No title

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

PP181-340731

QC: June 21, 2001

12:51

Style file version Nov. 19th, 1999

Archives of Sexual Behavior, Vol. 30, No. 5, 2001

A Typology Approach to Describing Parents as Communicators About Sexuality Doreen Rosenthal, PhD,1,3 Teresa Senserrick, PhD,1 and Shirley Feldman, PhD2

Teenagers in Grades 8 and 10 and their parents completed a questionnaire examining the frequency of parental communications about sexuality and the communicative style when discussing sexuality and in general. Respondents also assessed parents’ competence in communicating about sexual matters. For each set of respondents (teens reporting about mother, teens reporting about father, mothers’ self-reports, fathers’ self-reports), a cluster analysis yielded four clusters that were similar for each set. Relative to other parents, there was a group of parents that could be labelled as competent communicators and a group that could be labelled as problematic communicators about sexuality, with strong associations between cluster membership and score on the global measure of communicative competence. There were two intermediate categories that reflected more or less competence although the precise nature of these clusters differed as a function of informant group. Overall, fathers were rated as poorer communicators about sexuality than were mothers, at least by their teenage children. Consistent with other studies, mothers were more likely to be perceived as effective communicators by daughters and older teens. It appears that, independent of their level of competence, parents adjust their communication strategies according to the age and sex of their child, at least in the eyes of that child. Effective and problematic communicators among mothers were regarded as such by both sets of informants. This was not the case for fathers. We conclude that it is possible to classify parents usefully on the basis of perceived competence as communicators about sexuality. KEY WORDS: adolescents; sexuality; parents; communication.

1 La

Trobe University, Melbourne, Australia. University, Palo Alto, California. whom correspondence should be addressed at Faculty of Health Sciences, La Trobe University, Bundoora, Victoria, Australia 3086; e-mail: [email protected].

2 Stanford 3 To

463 C 2001 Plenum Publishing Corporation 0004-0002/01/1000-0463$19.50/0 °

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

464

PP181-340731

QC: June 21, 2001

12:51

Style file version Nov. 19th, 1999

Rosenthal, Senserrick, and Feldman

INTRODUCTION The importance of parents’ role as communicators about sexuality is increasingly recognized, largely as a result of concerns about sexually transmitted infections (STIs) among young people (see, for example, Kirby, 1997; Moore et al., 1996) and teenage pregnancy (Kirby, 1997). Given the pervasive role that parents play in their children’s decision making, it is reasonable to expect that responsible sexual decision making will be facilitated by parental involvement, not only in conveying information, values, and attitudes but also in the manner in which these are presented. In the present study, our primary goal was to describe different parental styles of communicating about sexuality. A secondary focus was to validate these styles by examining their relationship to teenagers’ evaluations of parents as communicators about sexual topics. In doing so, we go beyond the separate examination of various aspects of parent–teen sexual communication and attempt to derive a typology that enables us to identify those characteristics that define parents as competent communicators. We extend previous research in three ways. First, we take a multidimensional approach to communication about sexuality, second, we adopt a person-oriented rather than a variable-oriented approach, and third, we use multiple informants within the family. The Multidimensional Approach to Communication About Sexuality Early studies assumed that communication about sex was a unidimensional phenomenon, focussing primarily on amount rather than diversity and style of that communication. In this study, three domains are included as potentially influential aspects of teenagers’ judgments of their parents’ competence as sex communicators: frequency and quality of parents’ communication about sexuality and quality of parents’ general communication. Recent studies reveal that there is still limited sex-related discussion, that both parents and teens often find communication difficult, and that frequency of parental communication depends on the topic (Baumeister et al., 1995; Brock and Jennings, 1993; Nolin and Petersen, 1992; Rosenthal and Feldman, 1999). Parents’ manner of communicating about sexuality has had little attention despite the fact that adolescents report wishing that their parents would be open, supportive and empathic in their communications about sexuality (Brock and Jennings, 1993; Neer and Warren, 1988; Warren, 1995). In observational studies, compared to communications about other topics, maternal communications about sexuality are indirect, involve more dominance, less mutuality and turn-taking, more unilateral power assertion, and lower levels of comfort. Adolescent communications about sexuality involve more contempt and avoidance, and less honesty (Kahlbaugh et al., 1997; Lefkovitz et al., 1996; Levy et al., 1993; Yowell, 1997).

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

PP181-340731

QC: June 21, 2001

12:51

Parents as Communicators About Sexuality

Style file version Nov. 19th, 1999

465

Patterns of general communication have been associated with amount and quality of sex communication (Baldwin and Baranoski, 1990; Fisher, 1990; Kotva and Schneider, 1990). These findings raise the issue of whether sex-related communication is merely a subset of general communication (and therefore provides redundant information) or whether general communication provides additional information that enters into assessments of parents as effective communicators about sex. A Person-Oriented Approach This approach was prompted by a number of considerations. First are the highly equivocal conclusions to be drawn from the literature examining the impact of parental sex communication—considered one dimension at a time—on teen behaviors and beliefs. While many studies report that open, positive, and frequent parental communication about sex is associated with responsible sexual outcomes (see, for example, Jaccard and Dittus, 1991; Leland and Barth, 1993; Pick and Palos, 1995; Ward and Wyatt, 1994) other studies find either no association (see, for example, Casper, 1990; Cvetkovich and Grote, 1983; Fisher, 1993; Newcomer and Udry, 1985) or a negative association (Darling and Hicks, 1982; Fox and Inazu, 1980; Widmer, 1997). Miller (1998) concludes that the contradictory results may arise from the use of widely different measures of parent–teen communication. To this might be added the use of single, undifferentiated measures of that communication. Second, adolescents’ evaluations of their parents as communicators about sexuality depends on a number of separate factors, including frequency and style of communication about sex-related matters, and style of general communication, and these factors differ for mothers and fathers (Feldman and Rosenthal, 2000). We need to ascertain how these aspects of communication, both general and sexspecific, interact and how the meaning of one dimension is influenced by the presence or absence of another. In pursuing a typological approach, we follow the example of Baumrind (1966, 1978) and others who examined the influence of various aspects of parenting style in combination rather than in isolation. In similar vein, we hypothesise that parental characteristics associated with communication about sexuality would, in different combinations, have different outcomes. For example, frequent talk about sexual issues in the context of an empathic style might result in positive outcomes, whereas frequent discussions carried out in a controlling manner would not. A typological or person-oriented approach, in which possible effects from combinations or patterns of characteristics are explored (Magnusson and Bergman, 1990), has been largely ignored in the literature on adolescent sexuality, specifically (but see Buzwell and Rosenthal, 1996, for an account of adolescents’ “sexual

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

466

PP181-340731

QC: June 21, 2001

12:51

Style file version Nov. 19th, 1999

Rosenthal, Senserrick, and Feldman

selves”). However, in an interview study (Rosenthal et al., 1998) we were able to classify mothers according to a typology based on their perceptions of the frequency and content of their discussions about sexual issues. We included in this typology who initiated or maintained discussions, the comfort level of both mother and teenager, and the context in which communications took place. The effectiveness of their communication related not just to any one of these dimensions, but to their complex configuration. Using Different Informants The extent and effectiveness of communication about sexuality is dependent on the perspective of the informant. If we are to describe parental styles of communication and relate these to measures of effectiveness, we need to access parents’ as well as teenagers’ perspectives. Few studies have done this and even fewer still have focussed on fathers. Family members frequently give different accounts of the amount, nature, and quality of sex-related communications between parents and teens (Fisher, 1993; Jaccard et al., 1998; Kirby, 1995; Kotva and Schneider, 1990; Newcomer and Udry, 1984; Pick and Palos, 1995). Moreover, there is significant correspondence between mothers and their teenagers but not between fathers and teenagers (Feldman and Rosenthal, 2000). The amount and content of the communication is influenced by the sex of both parent and teen. Mothers are more frequent communicators than fathers; when fathers communicate, they tend to deal with the least intimate topics. Mothers are the preferred confidantes of both sons and daughters although they discuss sexuality more with daughters than with sons (Fisher, 1990; Greene and Grimsley, 1990; Kirby, 1995; Nolin and Petersen, 1992; Noller and Callan, 1990; Pick and Palos, 1995). Topics raised with daughters but less frequently with sons stress the negative, problematic aspects of sexuality (Darling and Hicks, 1983; Kirby, 1995). Because parents are likely to adapt their sex-related communications to the age and sex of their offspring, we have included 14- and 16-year-old boys and girls in our sample. Previous research has shown that relatively few 14-year-olds are sexually active and thus parents may consider sex-related communications as encompassing more “hypothetical” issues than those with 16-year-olds, of whom a substantial minority are already sexually active (Dunne et al., 1993; Rosenthal et al., 1999). In summary, this study addresses four questions. First, we ask whether or not we can classify parents, using frequency and quality of their sex-related communications, together with the quality of general parental communication. The goal of this classification is to produce a small number of groups such that parents within each group share communication patterns in common and these communication patterns differ across groups. Second, we investigate whether or not the groups

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

PP181-340731

QC: June 21, 2001

Parents as Communicators About Sexuality

12:51

Style file version Nov. 19th, 1999

467

that result from this classification differ in important ways and, if so, whether or not these groups relate to informants’ subjective evaluations of effective sex communicators. Third, we ask whether allocation to a group is a function of sex of parent and/or sex of teen. Finally, we examine whether or not the classifications derived from the teens’ scores match the classifications derived from parents’ scores. METHOD Participants The sample consisted of 215 children in 8th grade, 159 of their mothers and 80 of their fathers, and 209 in 10th grade, 156 of their mothers and 91 of their fathers. There were approximately equal numbers of boys and girls in each grade. We had complete data on 285 mother–teen dyads, 155 father–teen dyads, and 143 mother–father–teen triads. The families were primarily middle to upper middle class, as judged by parents’ education level, and almost all the parents were of English-speaking descent. Measures The frequency and quality of parents’ communications with their teenagers about sex was measured by two scales, and a global assessment of parents’ competence as communicators about sex was also included. The Frequency of Sex Communication Scale, an instrument designed for the present research (see Feldman and Rosenthal, 2000), consists of 20 items. Using the scale 1 = never, 2 = once, 3 = a few times, 4 = often, participants rate how often parents communicate about different sex-related topics. The scale consists of four factors: development and societal concerns (6 items); sexual safety (4 items); solitary sex (2 items); and the experience of sex (8 items). Examples of the sexual topics and alpha coefficients for parents’ and adolescents’ ratings are shown in Table I. Sex Communication Styles, an instrument also designed for the present research (see Feldman and Rosenthal, 2000), assessed quality of sex communications on several different dimensions. The scale consists of 30 items, describing both parental communications about sex and adolescents’ reactions to these communications, which are rated on 4-point scales (1 = strongly disagree, 4 = strongly agree). Teens filled out this instrument for their mothers and fathers separately. Parents filled out the scale describing their own behavior. The scale yielded five scores representing differing dimensions of sex communication: Comfortable (6 items); Controlling (9 items); Unwanted (6 items); and Avoidant (9 items). Sample items and alpha coefficients appear in Table I.

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340731

June 21, 2001

468

12:51

Style file version Nov. 19th, 1999

Rosenthal, Senserrick, and Feldman Table I. Alpha Coefficients and Sample Content of Measures Measures

Teens’ reports

Frequency of sex communications Development and .71–.84 societal concerns (6 items) Safe sex (4 items) .81–.90

Parents’ reports

Sample content of items

.79–.87

Physical development, menstruation, abortion, homosexuality

.87–.91

Contraception, safe sex, HIV/AIDS, STDs Dating and romance, sexual satisfaction, sexual desire, sexual pressure Wet dreams, masturbation

Experience of sex (8 items) Solitary sex (2 items) Sex communication Comfortable (6 items)

.79–.83

.75–.87

.48–.71

.71–.83

.71–.80

.69–.74

Controlling (9 items)

.70–.81

.66–.72

Unwanted (4 items)

.82–.90

.72–.82

Avoidant (9 items)

.72–.78

.76–.79

General communication Good (10 items)

.85–.91

.74–.83

Problematic (6 items)

.62–.71

.65–.75

Avoidant (4 items)

.55–.63

.51–.70

Parent communicates openly with teen, parent tries to make teen feel comfortable Parent decides what will be discussed, teen expects a lecture from parent Teen feels embarrassed, teen tries to change the topic Parent tends not to initiate discussions, parent only discusses if deemed necessary Teen is very satisfied with communications, parent tries to understand teen’s opinions Teen gives parent the silent treatment, parent tends to say things better left unsaid Teen avoids discussing some topics, teen is careful about what is said to parent

Note. Four α coefficients were calculated for 8th and 10th graders’ reports of mothers and fathers. Four α coefficients were calculated for mothers’ and fathers’ reports of 8th and 10th graders.

The evaluation of parents as communicators about sex consisted of a single question asked of teenagers about mothers and about fathers in separate parts of the survey: “Thinking about your mother’s (father’s) communication with you about sex and sex-related issues, how good a communicator is she (he)?” Teenagers rated each parent on a 6-point scale ranging from 1 (very poor), through 3 (average), to 6 (excellent). Parents rated themselves on a similar item. In addition, we asked about general communication in the family, using the Parent–Adolescent Communication Scale (Barnes and Olsen, 1982), a scale with good test-retest reliability (.78). The scale consists of 20 items rated on a 4-point scale (from 1 = strongly disagree to 4 = strongly agree). Items include both descriptions of parents’ behavior and teenagers’ feelings and behaviors in communicating with a parent. Guided by our earlier research (Feldman and Rosenthal, 2000), we created three composite scores: good (10 items), problematic

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

PP181-340731

QC: June 21, 2001

12:51

Style file version Nov. 19th, 1999

Parents as Communicators About Sexuality

469

(6 items), and avoidant (4 items). Sample items and alpha coefficients are presented in Table I. Procedure During class time, students completed separate questionnaires about their mothers and their fathers. Parents completed questionnaires only about their own behavior in interaction with their 8th or 10th grader. RESULTS Four sets of cluster analyses (teens’ ratings of mothers, teens’ ratings of fathers, mothers’ ratings, and fathers’ ratings) were conducted to identify groups of students and groups of parents with similar patterns of responses on the 11 scores (i.e., 4 frequency of sex communication factors, 4 sexual communication styles, 3 general communication styles). Analyses were performed in two steps. First, individuals were grouped according to a hierarchical method (Ward, 1963). Second, quick cluster analyses were performed to ensure appropriate assignment of cases to clusters. The final cluster centroids describe the profile for each cluster (see Aldenderfer and Blashfield, 1984; Hair et al., 1984). For each of the four analyses, a four-cluster solution was chosen at the first step based on dendrogram structure. At the second step, the four-cluster solutions were subjected to discriminant function analyses to check the validity of the solutions. In each case, analyses confirmed four significant discriminant functions at p < .001. For each of the four analyses, between 95.5 and 97.5% of cases were correctly classified. Next, cluster centroids were examined to establish cluster profiles. A series of MANOVAs was conducted including the 11 communication measures using each of the four sets of 4-cluster solutions as the grouping variable. As all four analyses were significant at p < .001, oneway ANOVAs using Scheffe comparisons were conducted for each clustering variable in order to identify differences across clusters. Results for each analysis are shown in Tables II, III, V, and VI. Mother–Teen Communication: Teens’ Perceptions Classification of Mothers Into Clusters Mean scores of clustering variables for teens’ ratings of mothers are presented in Table II. There were significant differences across clusters for each of the measures, and the four clusters differed significantly by gender [χ 2 (3) = 35.64, p < .001] and by grade [χ 2 (3) = 19.88, p < .001].

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340731

June 21, 2001

470

12:51

Style file version Nov. 19th, 1999

Rosenthal, Senserrick, and Feldman Table II. Mean Scores and F Values for Clusters of Teens’ View of Mother Variables

Frequency of sex communications Development and societal concerns Safe sex Solitary sex Experience of sex Sex communication style Comfortable Controlling Unwanted Avoidant General communication style Good Problematic Avoidant Effectiveness of parent as sex communicator (Teen report)

Effective (n = 40)

Reasonable (n = 123)

Disengaged (n = 133)

Problematic (n = 100)

F(3, 392)

3.13a

2.50b

1.51c

1.74d

141.39∗

3.16a 1.84a 2.50a

2.68b 1.30b 1.80b

1.41c 1.07c 1.27c

1.55c 1.07c 1.37c

149.59∗ 45.33∗ 134.95∗

3.45a 1.42a 1.45a 1.59a

2.99b 1.73b 2.12b 2.04b

2.66c 1.86c 2.29b 2.20c

2.24d 2.36d 2.92c 2.70d

94.90∗ 84.67∗ 73.30∗ 97.36∗

3.72a 1.81a 1.99a

3.10b 2.31b 2.57b

3.01b 2.02a 2.55c

2.39c 2.78c 3.05c

122.28∗ 67.27∗ 50.90∗

5.52a

4.62b

4.07c

3.02d

76.86∗

Note. Common superscripts indicate clusters do not significantly differ from each other. Scores for frequency and styles are derived from 4-point rating scales; scores for effectiveness are derived from 5-point rating scales. ∗p < .001.

Cluster 1: Effective Communicators (n = 40). Mothers in this group were considered to be frequent communicators about sexual issues, although this was topic dependent, and to have very positive sexual and general communication styles. Discussion focussed more on development, societal concerns, and safety, and less on solitary sex and the experience of sex. All scores for frequency, as well as positive communication styles, were significantly higher than for any other cluster and scores for negative communication styles were significantly lower. Students whose mothers were in this relatively small group were predominantly female (75%) and 10th graders (60%). Cluster 2: Reasonable Communicators (n = 123). Mothers in this cluster scored moderately well on positive communication styles, both sexual and general, as well as on frequency of discussion about sexual issues, although this was also topic dependent. These mothers, however, used a negative style of communicating with their teens. While this profile reflects reasonably good communication, scores on all measures were significantly less positive and more negative than for Cluster 1. There were slightly more females (55%) and older students (61%) in this group. Cluster 3: Disengaged Sexual Communicators (n = 133). These mothers’ communication about sexual issues was rated as very infrequent and significantly more negative in style than either of the first two clusters. However, general

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

PP181-340731

QC: June 21, 2001

12:51

Style file version Nov. 19th, 1999

Parents as Communicators About Sexuality

471

communication was relatively positive in style. Mothers in this group were more likely to have sons (72%) and 8th graders (65%). Cluster 4: Problematic Communicators (n = 100). A final group of mothers (N = 100) was rated as problematic communicators, with infrequent communication about sexual issues and a style of communication that reflected difficulties with both sexual and general communication. Positive communication styles were significantly lower and negative communication styles significantly higher than for any other cluster (particularly, avoidant general communication style). Students whose mothers were grouped in this cluster were approximately equally divided by gender and grade level. Evaluation of Mothers as Sex Communicators Our labelling of the clusters contains a clear evaluation of mothers’ communicative competence in the sexual domain. To examine if our evaluation concurs with that of the teenagers, we compared the scores of each group on the teens’ global ratings evaluating parents as sex communicators (see Table II). There was a significant difference in ratings of competence across the groups although, on average, mothers’ ratings were in the top half of the 6-point rating scale. Teens whose mothers were in the group labelled Effective rated their mothers highest (and as very good to excellent sexual communicators) while mothers in the group labelled Problematic Communicators were rated lowest. Mother–Teen Communication: Mothers’ Self-Perceptions Classification of Mothers Into Clusters Table III displays mean scores of clustering variables for mothers’ ratings. There were significant differences across clusters for each of the measures, and significant differences across the clusters by grade [χ 2 (3) = 42.51, p < .001], but not by gender. Cluster 1: Effective Communicators (n = 53). These mothers, the smallest group, rated themselves in similar ways to the first cluster of students, that is, as frequent communicators about sexual issues, with positive styles of sexual and general communication. Mothers in this cluster were more likely to be parents of 10th graders (62%) than of 8th graders. Cluster 2: Capable Communicators (n = 66). Mothers in this cluster also rated themselves positively overall in terms of their communication styles and equally so as mothers in the first group. However, communication about sexual issues was significantly less frequent for all topics. These mothers can be regarded as capable communicators when they do communicate about sex. There were more mothers of 8th graders (65%) in this cluster than of 10th graders.

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340731

June 21, 2001

472

12:51

Style file version Nov. 19th, 1999

Rosenthal, Senserrick, and Feldman Table III. Mean Scores and F Values for Clusters of Mothers’ View Effective (n = 53)

Variables

Frequency of sex communications Development and societal concerns Safe sex Solitary sex Experience of sex Sex Communication Style Comfortable Controlling Unwanted Avoidant General communication style Good Problematic Avoidant Effectiveness of parent as sex communicator (Parent report)

Capable (n = 66)

Inconsistent (n = 79)

Problematic (n = 90)

F (3, 284)

3.58a

2.87b

3.06b

2.18c

95.33∗

3.73a 2.75a 2.98a

2.92b 1.47b 1.89b

3.23c 1.75c 2.16c

1.98d 1.15d 1.36d

118.43∗ 91.53∗ 162.31∗

3.59a 1.36a 1.66a 1.53a

3.50a 1.48a 1.97b 1.74b

3.12b 1.81b 2.20c 2.05c

2.88c 1.79b 2.64d 2.34d

64.97∗ 36.99∗ 56.42∗ 103.96∗

3.41a 2.22a 1.90a

3.47a 2.01b 1.88a

3.02b 2.71c 2.36b

2.92b 2.41a 2.52b

46.43∗ 35.38∗ 40.35∗

4.65a

4.41a,b

4.01b

2.99c

49.57∗

Note. Common superscripts indicate clusters do not significantly differ from each other. Scores for frequency and styles are derived from 4-point rating scales; scores for effectiveness are derived from 5-point rating scales. ∗ p < .001.

Cluster 3: Inconsistent Communicators (n = 79). For the second largest group of mothers, discussions about sex were more frequent than for the mothers in Cluster 2, although this was strongly topic dependent. These mothers rated themselves as moderately comfortable in sexual discussions but had relatively high scores on negative characteristics when compared to mothers in Clusters 1 and 2. Compared to these (Clusters 1 and 2) mothers, they also had the least good and most negative ratings of their general communication style. Mothers of 10th graders (61%) were predominant. Cluster 4: Problematic Communicators (n = 90). This cluster had infrequent sexual discussions and a negative style of communicating about sexual issues. While these mothers described themselves as comfortable with sexual discussion and as having good general communication style, they also described unwanted and avoidant sexual styles and a problematic general style that were higher than for any other cluster. This reflects somewhat infrequent and problematic communication. This was the largest group of mothers, 60% of whom were mothers of 8th graders. Evaluation of Mothers as Sex Communicators Mothers’ global ratings of themselves as effective communicators about sexual issues significantly differed among the four groups (see Table III). Both

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

PP181-340731

QC: June 21, 2001

12:51

Style file version Nov. 19th, 1999

Parents as Communicators About Sexuality

473

Table IV. Cross-Tabulation Frequencies and Row Proportions of Clusters Representing Teens’ View of Mother by Mothers’ View Mothers view Teens’ view of mother

Effective (n = 47)

Capable (n = 53)

Inconsistent (n = 63)

Problematic (n = 74)

Effective (n = 23) Reasonable (n = 80) Disengaged (n = 80) Problematic (n = 54)

11 (.48) 21 (.26) 10 (.13) 5 (.09)

5 (.22) 22 (.28) 18 (.23) 8 (.15)

5 (.22) 24 (.30) 19 (.24) 15 (.28)

2 (.09) 13 (.16) 33 (.41) 26 (.48)

Note. N = 237 for this sample.

Effective Communicators and Capable Communicators evaluated themselves positively, with ratings, on average, in the range good to very good, while Problematic Communicators had the lowest self-rating. Comparison of Teens’ and Mothers’ Perceptions To establish if a match existed between teens’ and mothers’ perceptions, we cross-tabulated the two sets of cluster data. The analysis was limited to cases for which both mother and teen reports were obtained (N = 237). The analysis yielded a significant result [χ 2 (9) = 36.13, p < .001]. Standardised adjusted residuals were examined as representations of the extent of match and mismatch between clusters, with a score of 2 deemed significant (Cohen, 1960). Frequencies and row proportions for the solution appear in Table IV. Teens’ and mothers’ ratings of mothers as Effective Communicators and Problematic Communicators were consistent, and mothers rated as Effective Communicators by one set of informants were very unlikely to be classified as Disengaged, Inconsistent, or Problematic Communicators by the other set of respondents. Mothers classified as Reasonable Communicators by their teens were very unlikely to rate themselves as Problematic Communicators. Finally, Disengaged mothers (as rated by their teens) were very likely to be self-classified as Problematic Communicators by mothers. This pattern of results revealed a significant degree of concordance between teen-rated and mother-rated clusters, with this being most apparent in the most positive and most negative profiles. There were no instances of an association between clusters with substantially different meanings. Father–Teen Communication: Teens’ Perceptions Classification of Fathers Into Clusters Mean scores for clusters of teens’ perceptions of fathers are shown in Table V. There were significant differences across clusters for each of the measures, and

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

PP181-340731

QC: June 21, 2001

12:51

474

Style file version Nov. 19th, 1999

Rosenthal, Senserrick, and Feldman Table V. Mean Scores and F Values for Clusters of Teens’ View of Father Variables

Frequency of sex communications Development and societal concerns Safe sex Solitary sex Experience of sex Sex communication style Comfortable Controlling Unwanted Avoidant General communication style Good Problematic Avoidant Effectiveness of parent as sex communicator (Teen report)

Reasonable Inactive Ineffective Problematic (n = 60) (n = 48) (n = 158) (n = 53) F(3, 315) 2.34a

1.17b

1.34b

1.19b

126.26∗

2.73a 1.37a 1.95a

1.17b 1.03b 1.14b

1.45c 1.06b 1.21b

1.19b 1.04b 1.17b

131.85∗ 14.43∗ 78.52∗

2.82a 1.80a 2.08a 2.09a

2.41b 1.68a 1.95a 1.81b

2.44b 2.13b 2.52b 2.46c

1.81c 2.64c 3.39c 2.92d

45.19∗ 68.07∗ 65.12∗ 80.98∗

3.03a 2.05a 2.45a

2.83a 1.65b 2.06b

2.56b 2.38c 2.88c

1.85c 2.83d 3.21d

61.04∗ 63.98∗ 64.03∗

4.36a

3.98a

3.23b

1.96c

50.73∗

Note. Common superscripts indicate clusters do not significantly differ from each other. Scores for frequency and styles are derived from 4-point rating scales; scores for effectiveness are derived from 5-point rating scales. ∗ p < .001.

significant differences were found across clusters for both gender [χ 2 (3) = 8.90, p < .05] and grade [χ 2 (3) = 10.20, p < .05]. Cluster 1: Reasonable Communicators (n = 60). Students in this group characterized their fathers’ communication as moderately frequent concerning sexual issues, although the extent to which discussion occurred was strongly topic dependent. Styles of communication about sex were described in positive terms. However, although ratings for comfort with sexual communications were higher than for any other cluster, fathers in this group were also described as moderately avoidant and their communication as unwanted. General communication style was positive, although these fathers were perceived by their offspring to be moderately avoidant of communication. Male and female students were equally represented in this group, with fewer 8th graders (37%) than 10th graders. Cluster 2: Inactive Sexual Communicators (n = 48). Fathers in this group were perceived by students as very infrequent communicators about sex, although when they did communicate, they were moderately comfortable and rated low on negative styles. Relative to their peers, teens in this group rated fathers as, overall, moderately effective communicators. Students in this cluster were more likely to be sons (62%) than daughters and were predominantly in the 8th grade (63%).

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

PP181-340731

QC: June 21, 2001

12:51

Style file version Nov. 19th, 1999

Parents as Communicators About Sexuality

475

Cluster 3: Ineffective Communicators (n = 158). In this large cluster, discussion about sexual issues was very infrequent. On the whole, these fathers were regarded as having a mix of styles as communicators about sexual matters. A moderate degree of comfort in discussion was combined with somewhat controlling and avoidant behavior together with a moderate amount of unwanted communication. General communication style also showed a mix of positive and negative characteristics, with communication from these fathers rated as significantly less good, more problematic, and more avoidant than that of fathers in the first two groups. These fathers’ children were equally likely to be in either grade, and more likely to be sons (60%) than daughters. Cluster 4: Problematic Communicators (n = 53). Fathers in this cluster were rated by their teens as problematic communicators. As with fathers in Clusters 2 and 3, frequency of communication about sexual issues was particularly low. Importantly, on all measures of communication style—both sex-related and general— these fathers were rated more negatively than were fathers in the other three clusters. Comfort with communicating about sex was low, controlling and avoidant styles of communication were relatively frequent and young people rated these fathers’ communications as unwanted. In a similar vein, these fathers were rated as poor and avoidant communicators generally. When communication occurred, it was unlikely to be rated as good and more likely to be rated as problematic than communication from other fathers. Students whose fathers were in this cluster were more likely to be daughters (60%) and in the 10th grade (64%). Evaluation of Fathers as Sex Communicators Students’ global ratings of their fathers as effective communicators about sexual issues significantly differed among the four groups (see Table V). The highest evaluation of fathers by teens was a good rating for Reasonable and Inactive Communicators, with no significant difference between these two groups. Ineffective Communicators were judged as significantly less competent communicators than were fathers in the first two groups but significantly more able than the fathers in the fourth group, the Problematic Communicators.

Father–Teen Communication: Fathers’ Self-Perceptions Classification of Fathers Into Clusters Mean scores for clusters of fathers defined by fathers’ self-perceptions are shown in Table VI. There were significant differences across clusters for each of the measures, but no significant differences across the clusters by gender or grade.

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340731

June 21, 2001

476

12:51

Style file version Nov. 19th, 1999

Rosenthal, Senserrick, and Feldman Table VI. Mean Scores and F Values for Clusters of Fathers’ View Variables

Capable (N = 34)

Frequency of sex communications Development and societal concerns Safe sex Solitary sex Experience of sex Sex communication style Comfortable Controlling Unwanted Avoidant General communication style Good Problematic Avoidant Effectiveness of parent as sex communicator (Parent report)

Disengaged (N = 41)

Inconsistent (N = 43)

Problematic (N = 40)

F(3, 154)

2.69a

1.69b

2.57a

1.32c

76.05∗

2.86a 1.97a 2.01a

1.59b 1.06b 1.30b

3.08a 1.76a 1.95a

1.60b 1.06b 1.22b

65.51∗ 22.98∗ 36.46∗

3.46a 1.56a 1.91a 1.78a

2.73b 1.81b 2.30b 2.28b

2.83b 1.87b,c 2.49b,c 2.29b

2.79b 2.03c 2.69c 2.60c

34.54∗ 12.23∗ 20.18∗ 45.74∗

3.20a 2.28a 2.08a

3.07a 1.98b 2.11a

2.68b 2.61c 2.64b

2.69b 2.84c 2.92c

24.97∗ 30.81∗ 45.11∗

3.87a

3.28b

3.23b,c

2.72c

11.28∗

Note. Common superscripts indicate clusters do not significantly differ from each other. Scores for frequency and styles are derived from 4-point rating scales; scores for effectiveness are derived from 5-point rating scales. ∗ p < .001.

Cluster 1: Capable Communicators (n = 34). Fathers in this cluster reported moderate, topic-dependent frequencies of discussions about sexual issues. Their styles of communication were positive, with high levels of comfort and relatively low levels of controlling, unwanted, or avoidant behaviors. These fathers’ general communication style was, overall, more positive than that of fathers in other clusters. Cluster 2: Disengaged Communicators (n = 41). A second cluster of fathers reported low frequency of communication about sex-related issues and difficulties in their style of communicating. These fathers had moderate comfort levels but this was combined with evidence of controlling, unwanted, and avoidant behaviors. In contrast, these fathers rated themselves positively as general communicators, equally so as the first group of fathers in terms of being good communicators and less likely to engage in a problematic style. Cluster 3: Inconsistent Communicators (n = 43). Fathers in this cluster reported moderate frequency of communication about sexual issues and similar levels to fathers in Cluster 1. Frequency levels appeared to be more strongly topicdependent than for fathers in the first cluster. There was also considerable variation in their styles of communication, both about sexuality and in general. While these fathers were moderately comfortable with sexual discussions and relatively unlikely to be controlling in these discussions, they were also moderately avoidant

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340731

June 21, 2001

12:51

Style file version Nov. 19th, 1999

Parents as Communicators About Sexuality

477

and likely to perceive that these discussions were unwanted. Similarly, in general communication, there was evidence that they perceived their style to be good but also problematic and avoidant. Cluster 4: Problematic Communicators (n = 40). The final cluster of fathers were the least frequent communicators about sexual issues (but only slightly more so than the fathers in Cluster 2), had the least positive sexual communications styles and had mixed and relatively poor general communication styles. In particular, avoidant styles for both general and sexual communication were higher than for any other cluster. Evaluation of Fathers as Sex Communicators There was a significant difference across clusters in fathers’ overall evaluations of their sexual communication with their teens (see Table VI). Capable Communicators rated themselves significantly higher than those in the other clusters. Disengaged and Inconsistent Communicators rated themselves as average communicators, and fathers classified as Problematic Communicators reported the lowest self-ratings. Comparison of Teens’ and Fathers’ Perceptions The cross-tabulation analysis comparing clusters derived from teens’ and fathers’ perceptions (see Table VII) did not reach significance (χ 2 (9) = 0.49), suggesting that teens’ and fathers’ perceptions of father communication differed considerably. DISCUSSION Our attempt to develop a typology that reflected parents’ competence as communicators about sexual issues was successful. Four clearly defined groups Table VII. Cross-Tabulation Frequencies and Row Proportions of Clusters Representing Teens’ View of Father By Fathers’ View Fathers’ view Teens’ view of father

Capable (n = 25)

Disengaged (n = 31)

Reasonable (n = 27) Inactive (n = 14) Ineffective (n = 58) Problematic (N = 22)

8 (.30) 5 (.36) 9 (.16) 3 (.14)

4 (.15) 4 (.29) 18 (.31) 5 (.23)

Note. N = 121 in this sample.

Inconsistent (n = 36)

Problematic (n = 29)

9 (.33) 3 (.21) 18 (.31) 6 (.27)

6 (.22) 2 (.14) 13 (.22) 8 (.36)

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

478

PP181-340731

QC: June 21, 2001

12:51

Style file version Nov. 19th, 1999

Rosenthal, Senserrick, and Feldman

were derived for each of the four sets of informants (i.e., mother, father, and teens reporting about each parent). The resulting groups differed not only on the measures from which the classification was derived, but also on overall evaluation as communicators about sex. The classification of parents as communicators about sexual issues varied with the age and sex of the teenagers, especially when the classification was based on teenagers’ (rather than parents’) scores. Finally, the classifications of mothers, but not fathers, as communicators showed some validity in that there was moderate convergence between the classifications of teens and mothers. An important feature of our clusters is the relationship between the measures we used. For the most part, these are associated in expected ways. Thus, when sex communication is frequent, parents exhibit positive styles of communication, both about sex and in general. When sex communication is infrequent, parents are more likely to engage in negative styles of communication. Moreover, for most groups there is convergence between general communication style and communication style when discussing sexuality. We conclude from this that communication about sexuality may be regarded as a subset of general communication skills. Communicators with positive styles can transfer their skills to a specific domain, in this case, the sexual domain. Of interest are those few clusters where expected relationships do not occur. Among the clusters generated by teens in evaluating their mothers, there is a disjunction among the Disengaged communicators between their communication styles in discussing sexual matters (largely negative) and in general communication (largely positive). A similar disjunction in their self-perceptions occurs among the Disengaged fathers. For both these sets of respondents, it seems that, in spite of generally good communication, discussions about sex are problematic and off the agenda. Given that the majority of Disengaged mothers were mothers of sons and of Grade 8 students, allocation to this cluster might be a precursor to allocation to a cluster reflecting better communication about sexuality as their child grows older. In an earlier study (Rosenthal and Feldman, 1999), we found that some mothers wait to initiate discussions about these matters until they believe their child is “ready.” Disengaged fathers, on the other hand, appear to be those who find the sexual subset of communication particularly difficult and a task to be avoided. Given that these fathers are equally likely to have older and younger sons or daughters, there is little cause to believe that their communication patterns will change over time, unless external factors intervene. Finally, a third group, that of fathers designated as Inconsistent communicators, discussed sexual issues relatively frequently in comparison to their peers, but this occurred in the context of overall relatively ineffective communication styles. These were fathers who believed that they were receiving little encouragement from their teens to communicate with them. We speculate that, without this encouragement and without better communication skills, these fathers’ ability to engage their children in positive discussions around sexual issues will not improve.

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

PP181-340731

QC: June 21, 2001

Parents as Communicators About Sexuality

12:51

Style file version Nov. 19th, 1999

479

There is evidence of self-serving bias on the part of parents, a finding reported elsewhere in relation to parents’ beliefs about their teens’ sexual practices (Rosenthal and Collis, 1997; Feldman and Rosenthal, 2000). Teens consistently reported less positive and more negative perceptions of parents on every measure. It appears that parents give greater weight to their positive behaviors even though they engage in negative practices such as avoidance or control as well. This highlights the significance of parental reports of negative communicative practices. Our data suggest that very few parents are truly comfortable with discussing sex. Indeed, only a small minority of parents were rated by their teens as good communicators while a majority of mothers and almost all fathers were rated as problematic, inactive, or ineffective communicators. Consistent with other studies, mothers were more likely to be perceived as effective communicators by daughters and older teens. Mothers who were rated as the poorest communicators were equally as likely to have sons as daughters, and Grade 8 as Grade 10 children. Fathers do better with boys, according to their teens, although the clusters based on fathers’ reports revealed no differences as a function of age or sex of their teen. It appears that, independent of their level of competence, parents do adjust their communication strategies according to the age and sex of their child, at least in the eyes of that child. Given that much of the parent–child sex communication literature clearly points to better, more frequent communication between mothers and daughters and, in the few instances when fathers have been targeted, between fathers and sons, the finding that communication patterns do not differ by sex of their child is puzzling. We can invoke the self-serving nature of parents’ perceptions as one possible explanation. Parents, in overestimating their effectiveness as communicators, may be especially prone to do so with their opposite-sex children. Finally, the match between clusters derived from mothers’ perceptions and those of their teens signals a reasonable degree of synchrony in our classifications. Although the lack of correspondence between fathers’ and teens’ perceptions may be an artefact due to lack of statistical power because of the reduced number of cases, it is likely that this is a reflection of reality. Given fathers’ infrequent communications about sexuality, it is difficult to know the basis for informants’ judgments. Fathers may be actually exaggerating their contribution. Alternatively, students may be particularly critical of their fathers’ contributions or, indeed, be using standards of judgment that are different from those of their fathers. Whichever is the correct interpretation, it is not surprising that there is a mismatch between these two sets of clusters. In evaluating the usefulness of a typology approach, some caveats need to be borne in mind. First, our typology is dependent on the sample that we used and the questions we asked. Our volunteer sample may well have included parents who were, if not necessarily effective communicators about sexuality, conscious of the importance of their role as sex educators for their children. Moreover, a different set of measures may have produced other clusters. A useful next step would be

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

PP181-340731

QC: June 21, 2001

12:51

480

Style file version Nov. 19th, 1999

Rosenthal, Senserrick, and Feldman

to extend the measures used to include other factors that have the potential to impact on decisions about communicative competence. Second, the labels that we generated are value laden, and may be misleading. For example, fathers classified as Capable or Reasonable as sex communicators may be far removed from what experts advocate a good communicator should be. These “competent” fathers (designated as such, relative to other fathers) may be, and in fact are, quite different from the mothers classified as competent communicators. Finally, we were unable to examine if a relationship existed between the communication “types” we identified and teen outcomes, such as sexual activity. This remains a task for the future. In spite of these concerns, our data demonstrate that it is possible to classify parents usefully on the basis of perceived competence as communicators about sexuality, and that this competence is but part of general communicative competence. The perception of a competent communicator as one who is open, comfortable, willing to discuss sexual issues and to engage in dialogue is not simply a figment of researchers’ imagination. Rather, it is a culturally shared view, as evidenced by the links between parent and teen derived clusters and global competence measures. The task before us is to find ways to enable parents to become competent sex educators, with the communication skills that they and their teens recognize as important. ACKNOWLEDGMENTS This research was supported by a grant from the Australian Research Council and the Victorian Health Promotion Foundation. The authors are indebted to the schools, parents, and students who participated in the study. REFERENCES Aldenderfer, M. S., and Blashfield, R. K. (1984). Cluster Analysis, Sage, Beverly Hills, CA. Baldwin, S. E., and Baranoski, M. V. (1990). Family interactions and sex education in the home. Adolescence 25: 573–582. Barnes, H. L., and Olson, D. H. (1982). Parent-adolescent communication scale. In Olsen, D. H., and Markoff, R. (eds.), Family Inventories: Inventories Used in a National Survey of Families Across the Family Life Cycle, Family Social Science, University of Minnesota, St. Paul, pp. 33–48. Baumeister, L. M., Flores, E., and Marin, B. V. (1995). Sex information given to Latina adolescents by parents. Health Educ. Res. 10: 233–239. Baumrind, D. (1966). Effects of authoritative parental control on child behavior. Child Dev. 37: 887– 907. Baumrind, D. (1978). Parental disciplinary patterns and social competence in children. Youth Soc. 9: 239–276. Brock, L. J., and Jennings, G. H. (1993). Sexuality education: What daughters in their 30s wish their mothers had told them. Fam. Rel. 42: 61–65. Buzwell, S., and Rosenthal, D. A. (1996). Constructing a sexual self: Adolescents’ sexual selfperceptions and sexual risk-taking. J. Res. Adolesc. 6: 489–513. Casper, L. B. (1990). Does family interaction prevent adolescent pregnancy? Fam. Plann. Perspect. 22: 109–114.

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

PP181-340731

QC: June 21, 2001

Parents as Communicators About Sexuality

12:51

Style file version Nov. 19th, 1999

481

Cohen, (1960). A coefficient of agreement for nominal scales. Educ. Psychol. Meas. 20: 37–46. Cvetkovich, G., and Grote, B. (1983). Adolescent development and teenage fertility. In Byrne, D., and Fisher, W. A. (eds.), Adolescents, Sex and Contraception, Erlbaum, Hillsdale, NJ, pp. 109–123. Darling, C. A., and Hicks, M. W. (1983). Recycling parental sexual messages. J. Sex Marital Ther. 9: 233–243. Dunne, M. D., Lucke, J., Nilsson, R., and Raphael, B. (1993). 1992 HIV Risk and Sexual Behavior Survey in Australian Secondary Schools. Final report to the Department of Health, Housing and Community Services. Australian Government Publishing Service, Commonwealth of Australia, Canberra. Feldman, S. S., and Rosenthal, D. A. (2000). The effect of communication characteristics on family members’ perceptions of parents as sex communicators. J. Res. Adolesc. 10: 119–150. Fisher, T. D. (1990). Characteristics of mothers and fathers who talk to their adolescent children about sexuality. J. Psychol. Human Sex. 3: 53–70. Fisher, T. D. (1993). A comparison of various measures of family sexual communication: Psychometric properties, validity, and behavioural correlates. J. Sex Res. 30: 229–238. Fox, G. L., and Inazu, J. K. (1980). Patterns and outcomes of mother-daughter communication about sexuality. J. Soc. Issues 36: 7–29. Greene, A. L., and Grimsley, M. D. (1990). Age and gender differences in adolescents’ preferences for parental advice: Mum’s the word. J. Adolesc. Res. 5: 396–413. Hair, J. F., Anderson, R. E., and Tatham, R. L. (1984). Multivariate Data Analysis, 2nd edn., Macmillan, New York. Jaccard, J., and Dittus, P. (1991). Parent-Teen Communication: Toward the Prevention of Unintended Pregnancies, Spinger-Verlag, New York. Jaccard, J., Dittus, P., and Gordan V. (1998). Parent-adolescent congruency in reports of adolescent sexual behavior and in communications about sexual behavior. Child Dev. 69: 247–261. Kahlbaugh, P., Lefkovitz, E. S., Valdez, P., and Sigman, M. (1997). The affective nature of motheradolescent communication concerning sexuality and conflict. J. Res. Adolesc. 7: 221–239. Kirby, D. (1995). Parent-child communication about sexuality. Report to the Kaiser Foundation, ETR Associates, Santa Cruz, CA. (Unpublished manuscript). Kirby, D. (1997). No easy answers: Research findings on programs to reduce teen pregnancy, National Campaign to Prevent Teen Pregnancy, Washington. Kotva, H. J., and Schneider, H. G. (1990). Those “talks”: General and sexual communication between mothers and daughters. J. Soc. Behav. Pers. 5: 603–613. Lefkowitz, E. S., Kahlbaugh, P. E., and Sigman, M. D. (1996). Turn-taking in mother-adolescent conversations about sexuality and conflict. J. Youth. Adolesc. 25: 307–321. Leland, N. L., and Barth, R. P. (1993). Characteristics of adolescents who have attempted to avoid HIV and who have communicated with parents about sex. J. Adolesc. Res. 8: 58–76. Levy, S. R., Lampman, C., Handler, A., Flay, B. R., and Weeks, K. (1993). Young adolescent attitudes toward sex and substance use: Implications for AIDS prevention. AIDS Educ. Prev. 5: 340–351. Magnusson, D., and Bergman, L. (1990). A pattern approach to the study of pathways from childhood to adulthood. In Robins, L., and Rutter, M. (eds.), Straight and Devious Pathways From Childhood to Adulthood, Cambridge University Press, Cambridge, pp. 101–115. Miller, B. C. (1998). Pubertal development, parental communication, and sexual values in relation to adolescent sexual behavior. J. Early Adolesc. 18: 27–52. Moore, S. M., Rosenthal, D. A., and Mitchell, A. (1996). Youth, AIDS and Sexually Transmitted Diseases, Routledge, London. Neer, M. R., and Warren, C. (1988). The relationship of supportive communication to sex discussion in the home. Commun. Res. Rep. 5: 154–160. Newcomer, S. F., and Udry, J. R. (1984). Mothers’ influence on the sexual behavior of their teenage children. J. Marr. Fam. 46: 477–485. Newcomer, S. F., and Udry, J. R. (1985). Parent-child communication and adolescent sexual behavior. Fam. Plann. Perspect. 17: 169–174. Nolin, M. J., and Petersen, K. K. (1992). Gender differences in parent-child communication about sexuality: An exploratory study. J. Adolesc. Res. 7: 59–79. Noller, P., and Callan, V. J. (1990). Adolescents’ perceptions of the nature of their communication with parents. J. Youth Adolesc. 19: 349–362.

P1: GDB/GDX/GCZ/GFQ

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

482

PP181-340731

QC: June 21, 2001

12:51

Style file version Nov. 19th, 1999

Rosenthal, Senserrick, and Feldman

Pick, S., and Palos, P. A. (1995). Impact of the family on the sex lives of adolescents. Adolescence 30: 667–675. Rosenthal, D. A., and Collis, F. (1997). Australian parents’ beliefs about adolescent sexuality and HIV/AIDS. J. HIV Educ. Prev. Child Adolesc. 1: 57–72. Rosenthal, D. A., and Feldman, S. S. (1999). The importance of importance: The differentiated nature of parent-adolescent communication about sexuality. J. Adolesc. 22: 835–852. Rosenthal, D. A., Feldman, S. S., and Edwards, D. (1998). Mum’s the word: Mothers’ perspectives on communication about sexuality with adolescents. J. Adolesc. 21: 727–743. Rosenthal, D. A., Smith, A. M. A, and Lindsay, J. (1998). Change over time: High school students’ behaviours and beliefs, 1992 to 1997. Venereology 11(4): 6–13. Ward, J. (1963). Hierarchical grouping to optimize objective function. J. Am. Stat. Assoc. 58: 239–244. Ward, L. M., and Wyatt, G. E. (1994). The effects of childhood sexual messages on African-American and White women’s adolescent sexual behavior. Psychol. Women Quart. 18: 183–201. Warren, C. (1995). Parent-child communication about sex. In Socha, T. J., and Stamp, G. H. (eds.), Parents, Children and Communication: Frontiers of Theory and Research, Erlbaum, New Jersey. Widmer, E. D. (1997). Influence of older siblings on initiation of sexual intercourse. J. Marr. Fam. 59: 928–938. Yowell, C. (1997). Risks of communication: Early adolescent girls’ conversations with mothers and friends about sexuality. J. Early Adolesc. 17: 172–196.

P1: FYJ/LOV

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340732

June 22, 2001

12:46

Style file version Nov. 19th, 1999

Archives of Sexual Behavior, Vol. 30, No. 5, 2001

Sexual Child Abuse in a Defined Swedish Area 1993–97: A Population-Based Survey Anita Carlstedt, MSci, BA,1,2 Anders Forsman, MD,1 and Henrik Soderstrom, MD, BA1

Attempting to avoid some of the most common methodological problems involved in research on sexual child abuse, we collected data on crimes, perpetrators, and sanctions in all convicted cases of sexual child abuse in a defined population during a 5-year period. This approach provided amply documented and ascertained cases with precise definitions and descriptions of the crimes involved, no clinical referral bias, and minimal dependence on memory effects. The results are valid for the small proportion of cases that lead to conviction in the context of Swedish legislation. Structured data were collected from the court dossiers in all cases of sexual crimes against minors (less than 15 years of age) tried and sentenced at the courts in the V¨astra G¨otaland region of Sweden between 1993 and 1997. The total number of 496 sentences for sexual crimes during the study period included 203 cases of sexual child abuse (40.8%) with 283 victims and 196 perpetrators, all men. Girls were victims in 85% of the cases, boys in 12%, and boys as well as girls in 3%. Sexual penetration had occurred in 54.5% of cases and the total proportion of hands-on crimes was 83%. Most perpetrators, 72%, were well known to the child. The most severe offenses took place within the family. A wide range of acts were classified as sexual child abuse, but most common was sexual penetration of a female child by her biological father or a family friend. KEY WORDS: epidemiology; perpetrators; crimes; victims; sanctions.

INTRODUCTION Research on sexual child abuse has produced highly contradictory data depending on basic differences in study populations and designs. Victim surveys and 1 Department of Forensic Psychiatry, Institute of Clinical Neuroscience, G¨ oteborg University, Sweden. 2 To

whom correspondence should be addressed at Department of Forensic Psychiatry, Box 4024, 422 04 Hisings Backa, Sweden; e-mail: [email protected]. 483 C 2001 Plenum Publishing Corporation 0004-0002/01/1000-0483$19.50/0 °

P1: FYJ/LOV

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340732

June 22, 2001

12:46

484

Style file version Nov. 19th, 1999

Carlstedt, Forsman, and Soderstrom

clinically referred samples tend to yield the highest as well as the most disparate prevalences (Finkelhor et al., 1990; Finkelhor and Lewis, 1988; Jaffe et al., 1975; Painter, 1986) presumably because of variations in data collection techniques, to memory effects in either direction, and to scanty documentation in cases not tried in a court of law. The official crime statistics indicate a much lower incidence of sexual child abuse (Finkelhor and Hotaling, 1984; Russel, 1983), which suggests a large dark figure, especially for severe crimes in family settings. We have studied a total population of convicted perpetrators of sexual crimes against minors in a defined area to establish a cohort of ascertained and well-documented cases allowing detailed descriptions and comparisons to the background population.

SUBJECTS AND METHODS Background Population and Perpetrator Cohort The V¨astra G¨otaland region in Sweden has a population of about 1.5 million and includes G¨oteborg, the second largest city in the country. The total number of convictions for any type of sexual crime during the study period, 1993–97, was 496 according to statistics from the 14 district courts in the region. One of the authors (AC) visited all district courts to collect the documentation in all cases involving a minor (below the age of 15) as a victim. Of the 496 sentences, 203 (40.8%) concerned sexual child abuse and included 196 perpetrators, all men, 5 of whom were sentenced twice and 1 three times. Sociodemographic data from 1995, which was used as an average for the background population in V¨astra G¨otaland during the study period, was obtained from the Central Bureau of Statistics. Data on age, sex, and immigrant status was used for comparisons with the study group, whereas official statistics on civic status, education, and other variables was not specific enough for our purpose.

Data Collection The documentation collected in each case included the indictment, the complete police investigation (with video-taped or type-written interrogations of victims, perpetrators and witnesses, all technical evidence, and the personal case study), the sentence (including any revisions in higher courts), and the forensic psychiatric report in cases referred by the court to pretrial forensic psychiatric investigation. All data on sex, age, socioeconomic conditions, health status, previous criminality, index crime, relationship to victim, and legal consequences was registered by the same investigator (AC) in a standardized protocol for computerized analysis. Since some of the data sought, such as whether the perpetrator had been sexually abused in childhood, was unavailable in the majority of cases, this study

P1: FYJ/LOV

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340732

June 22, 2001

12:46

Sexual Child Abuse in a Defined Swedish Area 1993–97

Style file version Nov. 19th, 1999

485

is limited to basic data that could be clearly ascertained in all or almost all the cases. Statistical Analyses We used descriptive methods, chi-square tests, and tests for trend in contingency tables to test possible interrelationships between variables. Logistic regression was used to determine the relative contribution of several factors to the risk of a severe sanction; p-values are two-tailed. All statistics were computed by the SPSS 10.0 software (further references omitted). Specific Aims 1. To describe the full spectrum of crimes and perpetrators. 2. To register sanctions and medicolegal measures. 3. To analyze factors contributing to the risk of a severe sanction (defined as prison for more than 2 years or forensic psychiatric treatment with special court assessment for discharge). RESULTS Crimes, Victims, and Perpetrators The sexual acts involved in each of the 203 court cases are described in Table I. The victims were girls in 85% of all 203 cases, boys in 12%. Three percent involved both male and female victims. Further descriptive data for all cases (n = 203) are given in Table I and for all victims (n = 283, 41 boys and 242 girls) in Figs. 1(a) and (b) and 2(a) and (b). The most serious violations, that is, penetrating sexual intercourse, was significantly more often committed by biological relatives, household members, and family friends than by strangers (χ 2 = 22.521, df = 1, p < .0001). Boys were most often abused by strangers, but a close relative was the perpetrator in the few cases of aggravated abuse of boys (Table I). Median age of the perpetrators was 42 (range 18–86) years at the time of sentencing and 39 (range 12–79) years when the index crime (or the first violation included in the index sentence) was committed. Data describing the perpetrators are summarized in Table II. Perpetrators who confessed to their crimes (n = 111/203, 54.7%) were less often charged with violent abuse (χ 2 = 5.992, df = 1, p = .017) but were more often, than those who denied the crime, sentenced to prison for longer than 2 years (χ 2 = 8.161, df = 1, p = .005). The groups did not differ regarding substance abuse, psychiatric disorders, or gender of the victim.

1 1

Biological father (n = 42) Other close relative (n = 24) Stepfather (n = 28) Family friend (n = 52) Total stranger (n = 55) Total number of cases 26 15 18 26 6 91

Girls

2 3

1

Boys 7 2 6 6 2 23

Girls

0

Both sexes

3 3

Boys 5 4 2 11 9 31

Girls

0

Both sexes

1 1 1 3

Boys

1 1 1 3 22 28

Girls

3 3

Both sexes

Exhibitionism and other noncontact molestation (n = 34)

486

3

1

1 1

Both sexes

Other physical manipulation (n = 34)

June 22, 2001

4 9 15

Boys

Perpetrator

Genital manipulation (n = 26)

PP181-340732

Sexual penetration (n = 109)

P2: GMX/GFU/LOV

Relationship to victim

Archives of Sexual Behavior [asb]

Table I. Type of Violation, Sex of Victim, and Victim–Perpetrator Relationship in All 203 Sentenced Cases of Sexual Child Abuse

P1: FYJ/LOV QC: 12:46 Style file version Nov. 19th, 1999

Carlstedt, Forsman, and Soderstrom

P1: FYJ/LOV

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340732

June 22, 2001

12:46

Style file version Nov. 19th, 1999

(A)

(B)

Fig. 1. (a) Sexual activity and age of female victims and (b) sexual activity and age of male victims.

487

P1: FYJ/LOV

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340732

June 22, 2001

12:46

Style file version Nov. 19th, 1999

Fig. 2. (a) Relationship to the perpetrator and age of female victims and (b) relationship to the perpetrator and age of male victims.

488

P1: FYJ/LOV

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340732

June 22, 2001

12:46

Style file version Nov. 19th, 1999

Sexual Child Abuse in a Defined Swedish Area 1993–97

489

Table II. Demographic Data, Social Circumstances, and Previous Criminality in Perpetrators of Sexual Child Abusea

National origin Born and raised in Sweden Other Nordic countries Other European countries Countries outside Europe Maritial status Married Divorced Single Living alone but has a partner Not known Parenthood One or more biological children No biological children Not known Child custody Living with or sharing custody of children No regular contacts with children Not known Education University Upper secondary school Nine-year comprehensive school A few years of basic education Not known Occupation Fulltime employment or studies Part-time employment or studies Retirement or sickness pension Unemployed Not known Economic situation Meagre circumstances Income sufficient for necessities Well-to-do Not known Previous sexual criminality No previous conviction for sexual crimes Sexual crimes against adults Sexual child abuse Not known Substance abuse No substance abuse Alcohol abuse or dependence Not known an

= 196 (100%) for each group.

n

%

139 13 12 32

70.9 6.7 6.1 16.3

81 42 57 14 2

41.3 21.4 29.1 7.2 1.0

113 70 13

57.7 35.7 6.6

91 89 16

46.4 45.4 8.2

17 47 84 16 32

8.7 24.0 42.9 8.2 16.2

81 14 55 44 2

41.3 7.2 28.1 22.4 1.0

70 87 11 28

35.7 44.4 5.6 14.3

165 3 20 8

84.2 1.5 10.1 4.2

101 54 41

51.5 27.6 20.9

P1: FYJ/LOV

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340732

June 22, 2001

490

12:46

Style file version Nov. 19th, 1999

Carlstedt, Forsman, and Soderstrom Table III. Age-Wise Comparison Between the Background and Study Populations

Age group (years)

Number of men in the V¨astra G¨otaland region

Number of men in the study population

Number per 10,000 inhabitants

15–19 20–29 30–39 40–49 50–59 60–69 70–79 80–

49,987 110,349 100,574 104,365 69,991 62,150 38,911 9,726

19 20 46 63 23 14 8 3

3.8 1.8 4.6 6.0 3.3 2.3 2.1 3.1

The group of perpetrators with previous convictions for sexual crimes (n = 23/196, 11.7%) differed significantly from the nonrecidivists by having more substance abuse (χ 2 = 9.683, df = 1, p = .002) and psychiatric disorder before the crime (χ 2 = 3.430, df = 1, p = .011), and by a higher frequency of male victims (χ 2 = 14.385, df = 1, p < .001). They were more often a stranger to the victim (χ 2 = 16.384, df = 1, p < .001) and had more often committed crimes that did not involve sexual penetration (χ 2 = 9.708, df = 1, p = .002). Victims of the 203 sentenced crimes were 283 children (242 girls and 41 boys) from a background population of 145,503 girls and 154,051 boys in V¨astra G¨otaland. The yearly risk of becoming the victim of a subsequently tried and sentenced perpetrator of sexual child abuse was thus 33 per 100,000 girls and 5.2 per 100,000 boys. Assuming a constant risk in this region, 5 per 1,000 girls and almost 8 per 10,000 boys would be the victim in a sentenced case of sexual child abuse during his or her first 15 years. Compared to the total number of men above the age of 15 in the background population, the yearly risk for a man to be convicted of sexual child abuse was 7.44 in 100,000. Age-wise comparisons between the background and study populations are given in Table III. Data on immigration and age distribution in the general population were used to calculate an expected number of 36 immigrants in the study population of 196 perpetrators, which was significantly lower than the observed number of 54 immigrants ( p = .006, Poisson distribution). Medicolegal Assessments and Sanctions Forensic Psychiatric Investigation The court had ordered a “minor” pretrial forensic psychiatric investigation (i.e., a psychiatric interview) in 27% and a “major” forensic psychiatric investigation (i.e., a 4-week inpatient evaluation) in 8% of the cases. Referral to a minor or major forensic psychiatric investigation were significantly more frequent in

P1: FYJ/LOV

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340732

June 22, 2001

12:46

Style file version Nov. 19th, 1999

Sexual Child Abuse in a Defined Swedish Area 1993–97

491

cases where the crime involved violent features (χ 2 = 9.053, df = 1, p = .003), the victim was the same sex as the perpetrator (χ 2 = 8.377, df = 1, p = .007), the perpetrator had previous convictions for sexual offenses (χ 2 = 6.219, df = 1, p = .017), and, most significantly, the perpetrator had a history of psychiatric treatment (χ 2 = 14.403, df = 1, p < .001). No pretrial psychiatric evaluation was performed in 17 of 41 perpetrators with a diagnosed psychiatric disorder, or in 12 of 24 recidivists. The majority of perpetrators sentenced to imprisonment for more than 2 years (17 of 30) never saw a forensic psychiatrist before sentencing. Factors Influencing Severity of Sanctions Sanctions were imposed in 201 of the 203 court cases, since two perpetrators committed suicide before sentencing (Table IV). Severe sanctions (prison more than 2 years or forensic psychiatric treatment) were imposed in 42 cases, whereas 161 cases led to shorter prison terms, conditional sentences or fines. Severe sanction as an outcome was cross-tabulated with a number of possible predictors. Some of these variables were found to have no significant predictive value, including previous convictions (test for trend in contingency tables p = .2246), age of victim (test for trend in contingency tables p = .6888), being a stranger to the victim (χ 2 = 0.876, df = 1, p = .341), being under the influence of drugs at the time of the crime (χ 2 = 0.231, df = 1, p = .703), choosing a same-sex victim (χ 2 = 3.430, df = 1, p = .086), and being biologically related to the victim (χ 2 = 0.247, p = .712). There were, however, significant relationships to the perpetrator’s national origins, where immigrants had a significantly higher risk of a severe sanction (χ 2 = 22.841, df = 1, p < .001), violent features of the crime (χ 2 = 29.202, df = 1, p < .001), and to penetrating sexual activity (χ 2 = 5.020, df = 1, p = .036). The age of the perpetrator was not significant in a test for trend in contingency table but was introduced in the further analyses as a squared value to account for the possibility of a nonmonotonous relationship. Variables that were significantly associated with the risk of a severe sanction were entered into a logistic regression analysis (non-Swedish origins, penetrating sexual intercourse, and violence plus the squared and exact age of the perpetrator). In the presence of violence, penetrating intercourse lost its significant predictive Table IV. Sanctions in the 201 Court Cases Where a Sentence Could be Imposeda

Sexual penetration Genital manipulation Other physical manipulation Noncontact crimes a Two

Fines and/or probation

Prison <1 year

30 9 22 29

40 11 5 2

Prison Prison Special hospital 1–2 years >2 years treatment 10 2 1 0

convicted perpetrators committed suicide before being sentenced.

25 3 2 0

4 1 4 3

P1: FYJ/LOV

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340732

June 22, 2001

12:46

492

Style file version Nov. 19th, 1999

Carlstedt, Forsman, and Soderstrom

influence on the dependent variable and was therefore excluded from further analyses. Remaining as significant predictors in the final equation were age of the perpetrator, immigrant status, and violent features of the crime. The risk of having a severe sentence had a maximum at the age of 52. Beta-coefficients, p-values, and odds ratios (OR) of these factors were for violent features beta = 2.142, p < .0001, OR = 8.514, immigrant status beta = 1.615, p < .001, OR = 5.026, age of the perpetrator beta = 0.202, p = .016, age2 beta = −0.002, p = .037. Immigrant status alone thus increased the overall risk of having a severe sentence by a factor of 5. DISCUSSION We chose to study all sentenced cases of sexual child abuse in a defined population during a defined period to collect reliable information on the full spectrum of crimes and perpetrators in a nonclinical and well-documented cohort. Findings in such ascertained cases of sexual child abuse complement the picture drawn by studies of clinically referred or self-reported samples, but the presumably substantial proportion of nonsentenced and nonreported crimes must be considered in all interpretations of data. Our data support previous observations that children, particularly girls, are far more likely to be sexually abused in their familiar surroundings than by a stranger, and the figure is generally assumed to be considerably higher for familiar than for unfamiliar perpetrators of sexual child abuse. Severe sexual crimes were rarely committed by strangers in our cohort, where the most severe crimes were committed by biological fathers and other close relatives, and where even preschool children had been subjected to penetrating sexual intercourse. The high proportion of girls as victims agrees with the findings in the bulk of previous research. The proportion of boys corresponds to the general prevalence of homosexuality in society. In studies of victims, however, men report much higher prevalences of sexual child abuse experiences than reflected by studies on sentenced cases. Situations, gender of and relationships with perpetrators, and type of violations are likely to differ between boys and girls. The relative paucity of official register data on the life situations of the normal background population left few variables for statistical comparisons with the study population. Our cohort showed high prevalences of substance abuse, previous psychiatric treatment, unemployment, and dependence on public welfare. The recidivists had generally committed less severe crimes than the nonrecidivists, and the rate of recidivism was not as high as reported in previous studies (Proulx et al., 1997; Rice et al., 1991), which might reflect a lower incidence of pedophiles in our cohort. The prevalence of true pedophilia and other paraphilic impulse control disorders could not, however, be assessed in this study. The finding that perpetrators of non-Swedish origins were significantly overrepresented may be explained by increased attention to this category in the

P1: FYJ/LOV

P2: GMX/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340732

June 22, 2001

Sexual Child Abuse in a Defined Swedish Area 1993–97

12:46

Style file version Nov. 19th, 1999

493

community, and by a higher prevalence of confounding factors such as socioeconomic marginalization. According to the Swedish Penal Code, a sexual crime is not designated as rape unless the perpetrator has used violence to subdue the victim. Since physical violence is rarely required to restrain children, sexual child abuse is most often classified as “sexual intercourse with a minor.” The sanction for aggravated rape is a minimum of 4 and a maximum of 10 years in prison, whereas “sexual intercourse with a minor” is punishable by a maximum of 4 years. The most severe sentences in our material were primarily predicted by violent features of the crime (but not by penetrating sexual activity), by the age of the perpetrator (but not by the age of the victim), and by immigrant status. The latter, highly unexpected, finding certainly motivates further investigation on sentencing practices. REFERENCES Finkelhor, D., and Hotaling, G. T. (1984). Sexual abuse in the national incidence study of child abuse and neglect: An appraisal. Child Abuse Negl. 8: 23–33. Finkelhor, D., Hotaling, G., Lewis, I. A., and Smith, C. (1990). Sexual abuse in a national survey of adult men and women: Prevalence, characteristics, and risk factors. Child Abuse Negl. 14: 19–20. Finkelhor, D., and Lewis, I. A. (1988). An epidemiological approach to the study of child molestation. Ann. N.Y. Acad. Sci. 528: 64–78. Jaffe, A. C., Dynneson, R. N., and Bensel, R. W. (1975). Sexual abuse of children—An epidemiologic study. Am. J. Dis. Child. 129: 689–692. Painter, S. L. (1986). Research on the prevalence of child sexual abuse: New directions. Can. J. Behav. Sci. 18: 323–339. Proulx, J., Pellerin, B., Paradis, Y., McKibben, A., Aubut, J., and Ouimet, M. (1997). Static and dynamic predictors of recidivism in sexual aggressors. Sex. Abuse: J. Res. Treat. 9: 7–27. Rice, M. E., Quinsey, V. L., and Harris, G. T. (1991). Sexual recidivism among child molesters released from a maximum security psychiatric institution. J. Cons. Clin. Psychol. 59: 381–386. Russell, D. E. (1983). The incidence and prevalence of intrafamilial and extrafamilial sexual abuse of female children. Child Abuse Negl. 7: 133–146.

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

Archives of Sexual Behavior, Vol. 30, No. 5, 2001

An Evaluation of Self-Report Measures of Cognitive Distortions and Empathy Among Australian Sex Offenders1 David W. Tierney, MClinPsych,2 and Marita P. McCabe, PhD, FAPS2,3

The modification of deviant cognitions and the enhancement of victim empathy are central components in many treatment programs for sex offenders. There appear to be three broad problems with self-report measures of these factors: variations in the psychometric evaluation of measures; the transparency of items and thus the likely influence of social desirability; and the difficulty of determining which measures are specific to particular types of sex offenders. The aim of this study was to investigate these three issues among child molesters (CMs), and men convicted of sex offences against adults (ASOs). Data were collected from 36 CMs and 31 ASOs and from two comparison groups (33 men convicted of nonsexual offences and 40 nonoffenders from the community), to assess the reliability (internal and test-retest) and validity (discriminant, construct, and face) of measures, the influence of sexual social desirability on responding and the specificity of measures to both sex offender groups. Collectively, the results raise issues related to the assessment of sex offenders that require further investigation. They also have theoretical implications about the relationship between cognitive and emotive processes among sex offenders. KEY WORDS: cognitive distortions; empathy; sex offenders; assessment.

INTRODUCTION Sex offenders are characterized as having deviant cognitions and as lacking in empathy for their victims, and these variables are considered to be factors 1 Research supported by a scholarship paid to the first author by CORE, Department of Justice, Victoria,

Australia. of Psychology, Deakin University, Victoria, Australia. whom correspondence should be addressed at School of Psychology, Deakin University, 221 Burwood Highway, Burwood, Victoria, Australia 3125; e-mail: [email protected].

2 School 3 To

495 C 2001 Plenum Publishing Corporation 0004-0002/01/1000-0495$19.50/0 °

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

496

Tierney and McCabe

in the cause and maintenance of their sexual offending behavior, and are thus a major focus in many treatment programs and associated assessment processes (Marshall et al., 1995). Tierney and McCabe (in press) reviewed the literature on these measures as they related to the effectiveness of treatment programs for sex offenders. Three broad problems with the measures were identified: variations in psychometric evaluation; the transparency of items, and thus the possible influence of social desirability on responding; and a difficulty in determining which measures were specific to sex offenders generally, and which were specific to particular types of sex offenders. The authors concluded that the available literature made it very difficult for clinicians to determine the best measures of deviant cognitions and empathy in the assessment of sex offenders. The aim of this study was to empirically investigate the usefulness of measures of cognitive processes and empathy as determined by the criteria outlined earlier. This exploration was conducted with child molesters (CMs) and sex offenders against adults (ASOs). The following discussion will describe the evaluation criteria further, and outline the strategies used in this study to investigate these issues.

Evaluation of Psychometric Properties There is considerable variation in the extent to which the psychometric properties of measures of cognitive processes and empathy used to assess sex offenders have been evaluated (Tierney and McCabe, in press). Without establishing the reliability and validity of measures, it is unclear how reliable are the data that are collected using these measures, and it is also unclear what is being measured. In this study, the reliability of the measures was investigated by determining their internal reliability using Cronbach Alpha. The test-retest reliability of the measures was evaluated with a sample of CMs who were re-tested 2–3 weeks after initial testing. A range of practical problems made it difficult to assess the test-retest reliability of measures with the ASOs. Where possible, the discriminant validity of the measures was assessed by determining their ability to discriminate the sex offender groups from the comparison groups, and the CMs from the ASOs. Also, the construct validity of the measures was evaluated by assessing the relationship between measures purporting to assess the same or similar constructs.

The Influence of Social Desirability on Responding The transparency of items and the influence of social desirability on responding are considered major problems with the measures used with sex offenders (Gendreau et al., 1973; McGrath et al., 1998; Stermac et al., 1990). This, together with the fact that some sex offenders have been found to fake good responses

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

Assessment of Cognitive Processes and Empathy Among Sex Offenders

497

(Abel et al., 1975), raises questions as to whether the responses of sex offenders reflect their actual opinions and attitudes, or their desire to present themselves in a good light. Stermac et al. (1990) recommended providing confidentiality to encourage honest responding, and also recommended formally assessing the impact of social desirability on responding. Both of these strategies were utilized in this study. The Specificity of Measures The final problem identified with the measures of cognitive processes and empathy is related to the specificity of assessment tools. The measures range from those developed for the general population to those developed for particular sex offender types. Measures of general empathy, empathy for nonspecific victims of sexual abuse, and victim empathy have been used to assess sex offenders. Measures of cognitive processes include those that assess a tendency to deny or be defensive, deviant beliefs and attitudes, and offence specific variables (e.g. denial, minimization, and admission of sexual offences). At this point, it is difficult for researchers and clinicians to be clear about which measures are specific to the needs of particular types of sex offenders. The implications of this issue are best illustrated by research conducted by Marshall and colleagues (Marshall et al., 1995; Fernandez et al., 1999). These authors found that, compared to nonoffenders, CMs were able to accurately identify the emotions experienced by an accident victim, less able to identify the emotional state of nonspecific victims of child sexual abuse, and were markedly deficient in identifying the emotions of their own victims. However, the issue of the specificity of measures of empathy is confused by research that found that contrary to expectations, CMs had deficits in their recognition of the emotional states of adults as well as children (Hudson et al., 1993). Collectively, this research indicates that measures of both victim empathy and emotional recognition are of greater clinical utility among CMs than measures of general empathy. However, the issue of specificity has not been assessed with a sample of Australian CMs and ASOs, and this is one of the aims of this study. Two strategies were used to investigate the issue of specificity of measures to CMs and ASOs. First, where possible, the ability of measures to discriminate CMs and ASOs from each other and from non-sex offender comparison groups was investigated. Second, the relationship between particular measures that purport to assess the same or similar variable was assessed (similar to Fernandez et al., 1999; McGrath et al., 1998). Theoretically, one would assume that there would be predictable relationships between the various measures found to be reliable and valid with sex offenders. For example, given the research related to empathy, one would assume that general empathy would be unrelated to empathy for nonspecific victims of sexual abuse and unrelated to empathy for one’s own victim(s). However, it would be predicted that empathy for nonspecific victims of sexual abuse and empathy for one’s own victim(s) would be related.

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

498

Tierney and McCabe

METHOD Participants One hundred and forty men volunteered to participate: 36 had convictions for sexual offences against children (CMs); 31 had convictions for sexual offences against adults (ASOs); 33 were convicted of non-sexual offences (NSOs) (murder, attempted murder, false imprisonment, burglary, armed robbery, drug offences, fraud, assault, and culpable driving); and 40 nonoffenders were recruited from the community (COM). All offender groups were incarcerated. Both sex offender groups may have had additional non-sex-offence convictions, but their primary conviction was for a sexual offence. The sex offenders had not participated in treatment programs for their sexual offending behavior. Table I shows the demographic data for each group. Although the CM and COM groups reported similar mean ages and similar age ranges, the ASO and NSO groups were younger and had a narrower age range. However, Chi-Squared analysis found that, overall, the age difference between the groups was not significant [χ 2 (135) = 147, p > .21]. The COM group reported a significantly higher level of educational attainment than the other groups [χ 2 (15) = 49.50, p < .00]. Paid employment (full- plus part-time) was reported by

Table I. Demographic Data for Child Molester (CM), Adult Sex Offenders (ASOs), Non-Sexual Offenders (NSOs), and Community (COM) Groups

Mean age (range) Educational attainment (%)a Part high school High school Part tertiary school Tertiary school No response Employment (%)b Full-time paid Part-time paid Voluntary Unemployed Retired Relationship status (%) Single/never married Married/de facto Separated/divorced Widowed

CMs

ASOs

NSOs

COM

45.9 (23–78)

38.1 (22–57)

33.6 (18–59)

41.1 (19–71)

55.6 22.3 8.3 8.3 2.8

64.5 25.8 9.6 9.6 3.2

66.7 12.1 9.1 9.1 0

12.8 10.3 12.8 59 0

62.8 11.4 8.5 17.1 0

51.6 9.6 0 25.8 9.6

56.7 6.7 3.3 33.3 0

71.1 7.9 2.6 15.8 2.6

22.9 34.1 8.6 17.1

29 22.5 12.9 16.1

24.2 45.5 18.2 3

10.3 71.8 10.3 0

a Highest grade successfully completed. b For all offender groups this question related

to employment prior to being incarcerated.

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

Assessment of Cognitive Processes and Empathy Among Sex Offenders

499

more of the COM group than the other groups. The NSO and ASO groups reported a higher level of unemployment than the CM and COM groups, but this difference was not significant [χ 2 (18) = 16.39, p > .56]. The three offender groups were significantly more likely to be single or never married and less likely to be in a current relationship (married and de facto) than the COM group [χ 2 (21) = 76.6, p < .00]. Materials Table II lists the type of cognitive processes and empathy assessed and the measures used to assess these factors.

Assessment of a Tendency to Deny or Be Defensive The Sexual Social Desirability Scale was developed to assess the extent to which respondents attempt to present themselves in a socially desirable manner in relation to sexual issues (McGrath et al., 1998). The scale has 60 true/false items, 30 for each of two subscales. The Acquiescence subscale assesses a tendency to attribute positive but unlikely sexual behaviors, attitudes, and beliefs to oneself (e.g. “I buy magazines that have nude pictures mainly for the articles”). The Denial subscale assesses the tendency to deny negative but likely sexual behaviors, attitudes, and beliefs to oneself (e.g. “There have been times when I have had sexual thoughts or fantasies about someone else, even though I am involved in a relationship”). McGrath et al. (1998) found that among CMs, the Sexual Social Desirability Scale had good internal reliability (r = .88), good discriminant validity, and good concurrent validity with the Marlowe Crowne Social Desirability Scale (Crowne and Marlowe, 1960). McGrath et al. (1998) argued that the Denial subscale matches the defensive approach adopted by sexual offenders and to some extent this suggested that it has some face validity.

Assessment of Beliefs and Attitudes About Sex Between Adults and Children The Abel and Becker Cognition Scale (Abel et al., 1984) was developed to specifically measure the beliefs and attitudes of CMs about sexual contact between adults and children. This scale contains 29 items (e.g. “Having sex with a child is a good way for an adult to teach the child about sex”) that are considered to be consistent with clinical experience and to that extent the scale is considered to have good face validity (Murphy, 1990). Murphy (1990) cited the results of an unpublished doctoral thesis that identified subscales that had: acceptable reliability

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

500

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

Tierney and McCabe

(Cronbach alphas from .59 to .64 and test-retest reliabilities from .64 to .76); and that differentiated CMs from both normal controls and other sex offenders. Also, Stermac and Segal (1987) found that the Abel and Becker Cognition Scale differentiated CMs from both rapists of adult women and other non-sex-offender control groups. The Child Molester Scale was developed by McGrath et al. (1998) to measure cognitive distortions similar to those assessed by the Abel and Becker Cognition Scale (Abel et al., 1984). McGrath et al. (1998) found that among CMs, the Child Molester Scale had a moderate internal consistency coefficient (r = .65) and good discriminant validity. The items appear to be consistent with clinical practice, and to that extent the measure has some face validity. Further support for its validity is derived from the fact that McGrath et al. (1998) found that 17 CMs reported significantly fewer cognitive distortions after participating in a 24-week treatment program. Assessment of Offence Specific Cognitive Variables The Sexual Offence Attitudes Questionnaire (Home Office, 1996) assesses offence specific denial, minimization, and admission. The scale contains 50 statements related to the offender’s sexual offence (e.g. “The victim of my offences is also to blame for the offences happening”). Offenders are required to indicate how true or untrue they believe the statements are about their most recent sexual convictions. Although the psychometric properties of this scale have not been reported in the literature, the items are considered to reflect the general statements made by CMs and to that extent they are considered to have some face validity (Owen, 1997). It is also argued that the items on this scale reflect Salter’s multidimensional model (Salter, 1988) of denial: admission with justification, denial of responsibility, denial of seriousness, and complete denial (Owen, 1997). In a pilot study, ten CMs allocated the items to one of three categories: admission of offence, minimization of offence, and denial of offence (Tierney, 1999). These items comprise the subscales used in this study: the Admit, Minimize, and Deny subscales. A number of researchers have assessed the causal attributions for offending among CMs (Kennedy and Grubin, 1992; Larsen et al., 1995; McKay et al., 1996). McKay et al. (1996), for example, used the Causal Dimension Scale (Russell, 1982) as a basis for developing the Offence Questionnaire on which respondents are required to rate their attributions for their offences. The scale measures three causal dimensions described by Weiner’s attributional theory (Weiner, 1986; controllability of behavior, stability of behavior, and locus of causality), and contains nine items (three items per causal dimension). Each statement has a 5-point scale anchored by words that describe opposite ends of one of the three causal dimensions. The internal reliability of the scale across four types of offenders (CMs, rapists,

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

Assessment of Cognitive Processes and Empathy Among Sex Offenders

501

property offenders, and violent offenders) ranged from r = .91 to .98 (McKay et al., 1996). Mehrabian Emotional Empathy Scale Mehrabian Emotional Empathy Scale (Mehrabian and Epstein, 1972) is a measure of general empathy. It contains 33 statements that respondents are required to rate in the range +4 (very strong agreement) to −4 (very strong disagreement). In developing this scale, Mehrabian and Epstein (1972) only selected items that did not correlate with the Marlowe Crowne Social Desirability Scale (Crowne and Marlowe, 1960), to ensure that the scale was not confounded by social desirability. Further, each item was found to correlate significantly with the total score and the spilt-half reliability of the scale was .84. Marshall et al. (1995) cited two unpublished studies that used this scale with rapists, one found that they were less empathic than nonoffenders, but the second study was unable to confirm this finding. Langevin et al. (1988) found no differences between a range of sex offenders and a community control group on the Mehrabian Emotional Empathy Scale, and also found no differences between the different types of sex offenders. Langevin et al. (1988) found that the internal reliability of this scale was low (r = .59), but this property has not been assessed with a sample of Australian sex offenders. Empat(general) The Empat has two subscales – one assesses empathy for nonspecific victims of child sexual abuse [Empat(child)], which is discussed later, and the Empat(general) that measures general empathy (McGrath et al., 1998). McGrath et al. (1998) found that the Empat(general) had good internal reliability (r = .77), but that CMs were not deficient in general empathy compared to two control groups. The Empat(child) The Empat(child) was developed to investigate empathy for nonspecific victims of child sexual assault (McGrath et al., 1998), and is comprised of 34 items (e.g. “If someone was molested 18 years ago, they should be over it by now”). McGrath et al. (1998) found that the Empat(child) had good internal reliability (r = .92), and good discriminant validity. The items were generated from the observed deficits of CMs and from victim accounts, so one would expect them to have good face validity. Also, after completing a 24-week relapse prevention program, 17 CMs reported significantly more victim empathy as measured on the Empat(child), and this provides additional support for its validity (McGrath et al., 1998).

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

502

Tierney and McCabe

Victim Empathy Distortions Scale Beckett and Fisher (1994) developed the Victim Empathy Distortions Scale to assess (1) empathy for victims depicted in vignettes and (2) empathy for an offender’s own victim. This scale contains 30 questions, each with 5 possible responses. The Victim Empathy Distortions Scale was found to have high internal reliability (r = .89) in 140 untreated CMs, and high test-retest reliability (r = .90) in 40 untreated CMs (Beckett et al., 1994). Furthermore, the scale items appear to have good face validity, and the scale has been found to discriminate between CMs and a control group (presumably this was in relation to their responses to the vignettes, as the control group were not sex offenders).

Procedure The CMs, NSOs, and ASOs were recruited from six prisons in the State of Victoria, Australia. The CMs and NSOs responded to Public Notices asking for research volunteers. No ASOs responded to the Public Notices, so this sample was obtained by asking individuals if they were willing to participate. To obtain a sample size of 31 ASOs, 90 ASOs were invited to participate. The COM group responded to a Public Notice displayed in a Community Health Center. The CMs were assessed in groups of 2–12 within a Protection prison. The NSOs were also assessed in groups in Mainstream prisons. For security reasons, the ASOs, who were housed in Mainstream prisons, completed the questionnaires individually in an interview room where the purpose could be disguised (medical or educational rooms). The COM group completed the questionnaires and returned them by post to the principal researcher (40 out of 60 questionnaire packages were completed and returned). Research Design and Statistical Procedures All participants completed all measures except those related to sexual offences, which were completed by the CM and ASO groups only. This design enabled an assessment of the internal reliability of the measures for each group, and a comparison of the responses across the four groups provided an assessment of the discriminant validity of the measures. The CMs were reassessed 2–3 weeks after the initial test, and this provided an assessment of the test-retest reliability of measures. An investigation of the relationship between measures provided an assessment of their construct validity and their specificity to CMs and ASOs. Finally, the influence of social desirability on responding was investigated by assessing the relationship between the Acquiescence and Denial subscales of the Sexual Social Desirability Scale and the measures of cognitive processes and empathy.

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

Assessment of Cognitive Processes and Empathy Among Sex Offenders

503

RESULTS AND DISCUSSION The evaluation of the psychometric properties of the measures of cognitive and emotive variables will be discussed first, followed by an assessment of the influence of sexual social desirability on responding, and finally, an assessment of the issue of the specificity of measures. Evaluation of the Psychometric Properties of Measures of Cognitive Variables The descriptive data for the measures of cognitive processes are shown in Table II and these will be discussed first, followed by a discussion of the data related to the reliability (internal and test-retest) and validity (discriminant and construct) of measures. Descriptive Data On the Sexual Social Desirability Scale, the CMs obtained the highest mean score on the Denial subscale followed in order by the COM, NSO, and ASO Table II. Descriptive Statistics for the Measures of Cognitive Processes Among Child Molesters (CMs), Adult Sex Offenders (ASOs), Non-Sexual Offenders (NSOs), and Community (COM) Groups Measure Defensiveness Sexual Social Desirability Scale (Acquiescence) Sexual Social Desirability Scale (Denial) Beliefs and attitudes Child Molester Scale Abel and Becker Cognitions Scale Offence related cognitive processes Sexual Offence Attitudes Questionnaire (admit) Sexual Offence Attitudes Questionnaire (minimize) Sexual Offence Attitudes Questionnaire (deny) Causal Dimension Scale (locus) Causal Dimension Scale (control) Causal Dimension Scale (stability)

CMs, mean (SD)

ASOs, mean (SD)

NSOs, mean (SD)

COM, mean (SD)

17.5 (4.8)

15.5 (4.4)

18.2 (5.2)

14.8 (3.8)

14.2 (5)

10.6 (4.8)

11.3 (6.3)

12.1 (4.9)

65.5 (7.9) 118.1 (18.4)

62.0 (17.7) 127.5 (12.9)

66.5 (7.8) 121.6 (18.6)

70.3 (5) 132.9 (11.2)

27.8 (12)

38.6 (11.1)

39.1 (14.4)

41.4 (13)

10.2 (6.9)

13.0 (7.2)

11.1 (3)

10.4 (4.2)

10.5 (3.1)

10.1 (2.7)

7.7 (3.7)

6.4 (2.8)

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

504

Style file version Nov. 19th, 1999

Tierney and McCabe

groups. This indicates that the CMs had a greater tendency to deny negative but likely sexual behavior to themselves, and that the ASOs were the least likely to have this tendency. On the Acquiescence subscale, the NSOs obtained the highest mean score, followed in order by the CM, ASO, and COM groups. This suggests that the NSOs had a greater tendency than the other groups to attribute positive but unlikely sexual behaviors to themselves, and that the COM group were the least likely to do this. For the Child Molester Scale, the ASOs obtained the lowest group mean score, and the COM group the highest group mean score. The mean scores for the CM and NSO groups fell between the ASO and COM groups and were essentially the same. This pattern of results was unexpected, as it indicated that the ASOs had the most deviant beliefs and attitudes about sexual relations between adults and children, followed in order by the CM, NSO, and COM groups. With the Abel and Becker Cognitions Scale, however, the CMs obtained the lowest group mean score, followed by the NSO, the ASO, and the COM groups, who obtained the highest group mean. This indicated that the CMs had the most deviant beliefs and attitudes about sexual relations between adults and children, and the COM group had the least deviant. Thus, the descriptive data indicate that while the Abel and Becker Cognitions Scale performed in the expected manner, this did not occur for the Child Molester Scale. Of the two measures of sex offence related variables, the ASOs obtained higher mean scores than the CMs on the Admit and Deny subscales of the Sexual Offences Attitudes Questionnaire, whereas mean scores on the Minimize subscale were similar. The scores of the CMs and ASOs on the subscales of the Causal Dimensions Scale were similar. Internal Reliability Table III shows the internal reliability for all measures of cognitive processes. The alpha coefficients for the Sexual Social Desirability Scale ranged from low to moderate. An item analysis was conducted to investigate the low alpha on the Denial subscale for the CMs. The removal of Item 10 increased its internal reliability (from r = .53 to r = .78) for the CMs without altering the reliability for the NSO and COM groups, although the reliability for the ASOs was slightly reduced (from r =.66 to r = .64). Given this, and the fact that the content of that item (“I have just felt like going to sleep after having sex”) was not considered crucial, it would appear that this item should be removed. With the measures of beliefs and attitudes about sexual relations between adults and children, Coefficient Alpha for the Abel and Becker Cognitions Scale was consistently high across all groups, whereas those for the Child Molester Scale were consistently low. An item analysis of the Child Molester Scale indicated that its internal reliability would be improved by deleting a substantial number of items.

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

Assessment of Cognitive Processes and Empathy Among Sex Offenders

505

Table III. Cronbach Alphas for Measures of Cognitive Processes Among Child Molesters (CMs), Adult Sex Offenders (ASOs), Non-Sexual Offenders (NSOs), and Community (COM) Groups CMs

ASOs

NSOs

COM

.75 .53

.61 .66

.64 .88

.61 .8

.46 .94

.00 .96

.47 .84

.03 .74

.79 .83 .79 .35 .18 .66

.81 .79 .74 .76 .29 .01

NA NA NA NA NA NA

NA NA NA NA NA NA

Defensiveness Sexual Social Desirability Scale (Acquiescence) Sexual Social Desirability Scale (Denial) Beliefs and attitudes Child Molester Scale Abel and Becker Cognitions Scale Offense related cognitive processes Sexual Offence Attitudes Questionnaire (admit) Sexual Offence Attitudes Questionnaire (minimize) Sexual Offence Attitudes Questionnaire (deny) Causal Dimensions Scale (locus) Causal Dimensions Scale (control) Causal Dimensions Scale (stability)

However, each group required different items to be removed to improve the internal reliability for that group, and many of the items that were required to be removed were among those found to discriminate the CMs from the other groups. The Sexual Offences Attitudes Questionnaire subscales had moderate to high internal reliability for the CM and ASO groups. With the Causal Dimensions Scale, however, the internal reliability of two subscales were poor (locus and control), whereas the reliability of the third subscale (stability) was moderate with the CMs. The internal reliability for the subscales of this questionnaire for the ASOs ranged from very low (control and stability) to moderate (locus). Test-Retest Reliability The test-retest reliability for all measures for the CM group were moderate to high (ranging from r = .65 to r = .92), except for the Child Molester Scale and the Control subscale of the Causal Dimensions Scale, where the test-retest correlations were low (r = .58 and r = .52, respectively). Discriminant Validity With the measures of defensiveness (Sexual Social Desirability Scale), there were no significant differences between the groups on the Denial subscale, which indicated that there were no differences between the groups in their tendencies to deny negative but likely sexual behaviors to themselves. With the Acquiescence subscale, however, there was a significant mean difference between the NSO and COM groups (mean diff. = −3.47, p < .04), which indicated that the NSOs were more likely to attribute positive but unlikely sexual behaviors to themselves than the COM group. There were no other significant group differences.

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

506

Style file version Nov. 19th, 1999

Tierney and McCabe

For the Child Molester Scale, post hoc testing found no significant differences between the CMs and the comparison groups. The only significant difference was between the ASO and COM groups (mean diff. = −8.27, p < .018). For the Abel and Becker Cognitions Scale, post hoc testing found a significant difference between the CM and the COM groups (mean diff. = −14.32, p < .01), and a significant difference between the NSO and COM groups (mean diff. = −11.23, p < .043). No significant differences were found between the three offender groups. Given the unexpected pattern of results with the Child Molester Scale and the Abel and Becker Cognitions Scale, ANOVAS were conducted to determine whether particular items were able to discriminate the CMs from the comparison groups, and to determine the clinical significance or relevance of these differences. These analyses highlighted a number of issues. First, only 12 of 22 Child Molester Scale items and 15 of 29 Abel and Becker Cognitions Scale items differentiated the CMs from the comparison groups. Second, among the items where a statistically significant difference was found, the difference was due to the extent of agreement or the extent of disagreement, rather than the direction of agreement or disagreement. This finding applied for 5 of the 12 Child Molester Scale items and all of the Abel and Becker Cognitions Scale items. This finding raises questions about the clinical relevance of these items. Finally, five Child Molester Scale items showed a clear statistical difference that may have some clinical relevance: two of these items related to the likelihood of reoffending (31, 58), one related to the length of jail sentences for child sexual offences (20), one related to men being aroused by naked children (46), and one related to sex offenders attempting to get sympathy and avoid criticism by saying they feel badly after having sex with a child (50). Construct Validity Table IV shows the correlation between measures purporting to assess the same or similar construct, and the correlation between the subscales of the same measure. This assessment was not conducted with the Causal Dimensions Scale, as it had been found to be an unreliable measure. The construct validity of the Sexual Social Desirability Scale was largely supported by the pattern of correlations between its subscales and those of the Marlowe Crowne Social Desirability Scale. For all groups, the Acquiescence subscales of both measures were significantly related, as were the Denial subscales. Also, the lack of a relationship between the Acquiescence and Denial subscales of the Sexual Social Desirability Scale suggests that they are different constructs, and this is additional support for the construct validity of this scale. The relationship between the two measures of attitudes and beliefs about sexual contact between adults and children was not significant for the three offender

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

Assessment of Cognitive Processes and Empathy Among Sex Offenders

507

Table IV. Correlations Between the Measures of Cognitive Processes Among Child Molesters (CMs), Adult Sex Offenders (ASOs), Non-Sexual Offenders (NSOs), and Community (COM) Groups Measures

CMs

Denial/defensiveness Sexual Social Desirability Scale (Denial) vs. Marlowe Crowne Social Scale (Denial) Sexual Social Desirability Scale (Acquiescence) vs. Marlowe Crowne Social Scale (Acquiescence) Sexual Social Desirability Scale (Acquiescence) vs. Sexual Social Desirability Scale (Denial) Beliefs and attitudes Child Molester Scale vs. Abel and Becker Cognitions Scale Offence related variables Sexual Offence Attitudes Questionnaire (admit) vs. (deny) Sexual Offence Attitudes Questionnaire (admit) vs. (minimize) Sexual Offence Attitudes Questionnaire (admit) vs. (minimize) ∗p

ASOs

NSOs

COM

.51∗∗

.40∗

.73∗∗

.58∗∗

.67∗∗

.55∗∗

.71∗∗

.57∗∗

.08

.32

.13

.29

.07

−.21

−.01

.50∗∗

−.63∗∗

−.72∗∗

NA

NA

−.37∗

−.48∗

NA

NA

.45∗

NA

NA

.61∗∗

< .05; ∗∗ p < .01.

groups, but was significant for the COM group. This questions the construct validity of the Child Molester Scale and the Abel and Becker Cognitions Scale with offender groups. The pattern of correlations between the subscales of the Sexual Offence Attitudes Scale supports its construct validity. For both sex offender groups, the Admit subscale had a significant and negative relationship with the Minimize and Deny subscales, although Minimize and Deny subscales were significantly correlated. Conclusion The Sexual Social Desirability Scale was found to have low to moderate internal reliability with the CMs and ASOs, and good test-retest reliability with CMs. Its discriminative power was not strong. However, the descriptive data suggested that CMs were the most likely to deny negative but likely sexual behaviors and the second most likely of the four groups to attribute positive but unlikely sexual behaviors to themselves. This is consistent with the view that in assessment processes CMs attempt to present themselves in a good light (Abel et al., 1975). Interestingly, the ASOs were the third most likely to attribute positive but unlikely sexual behaviors to themselves and were the least likely to deny negative but likely sexual behaviors to themselves. With the measures of attitudes and beliefs about sexual relations between adults and children, the Abel and Becker Cognitions Scale displayed stronger psychometric properties than the Child Molester Scale. The internal reliability of

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

508

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

Tierney and McCabe

the Abel and Becker Cognitions Scale ranged from high to very high across all groups while the internal reliability of the Child Molester was consistently poor. Also, among the CMs, the Abel and Becker Cognitions Scale had higher test-retest reliability than the Child Molester Scale. The descriptive data show the responses to the Abel and Becker Cognitions Scale were in the expected direction—the CMs obtained scores that indicated that they had the most deviant attitudes and beliefs about sexual relations between adults and children. This was supported, in part, by post hoc testing, which found a significant difference between the CM and COM groups. One would have expected a significant difference between the CM and both the ASO and NSO groups, but this was not found, although the group means were in the expected direction. With the Child Molester Scale, however, unexpected trends were found: the group means indicated that the ASOs had the most deviant attitudes about sexual contact between adults and children, and the CM and NSO group means were essentially the same. This pattern of results raises doubts about the discriminant validity of the Child Molester Scale. An item analysis of the Child Molester Scale and the Abel and Becker Cognitions Scale highlighted a number of issues. First, only a subset of items from both scales had some discriminate validity (i.e. discriminated the CMs from the comparison groups). Also, for many of the items where a statistically significant difference was found, the difference related to the extent of agreement and the extent of disagreement rather than a clinical difference, and this raises questions about the clinical utility of those items. This issue requires further investigation. The inconsistent pattern of correlations between the Abel and Becker Cognition Scale and the Child Molester Scale across the groups raises some doubt about the construct validity of both scales. However, the results of this study would suggest that the Child Molester Scale has particularly poor psychometric properties and the poor correlation between the two measures may reflect this. It is interesting to speculate about why the measures of beliefs and attitudes did not statistically differentiate the CMs from all comparison groups. First, it is possible that in the 20 years since the development of the Abel and Becker Cognition Scale the amount of community discussion about the inappropriateness of sexual contact between adults and children has resulted in a shift in the knowledge of CMs. However, this does not explain why the ASOs and NSOs were not statistically different from the CMs in their responses. Also, this proposition does not explain the results with the Child Molester Scale, which is a more recently developed scale, although this scale was found to be unreliable. A second possibility is the incidental learning within the prison where the CMs were assessed. Although data were obtained only from CMs who had not participated in treatment, there was a considerable degree of contact between this group and CMs who were undergoing or who had completed treatment (i.e. common work and recreational areas and activities). It is possible the discussions

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

Assessment of Cognitive Processes and Empathy Among Sex Offenders

509

between these different groups may have influenced the attitudes of those who participated in this study. A third possibility is that beliefs and attitudes about sexual contact between adults and children are not sufficient to explain whether someone sexually abuses children or not. This was reflected in anecdotal comments written by two CMs on their responses to the Abel and Becker Cognition Scale; although their responses to particular items were not deviant, they wrote that their behavior would be different. Thus, although the Abel and Becker Cognition Scale and the Child Molester Scale are assessing beliefs and attitudes, the CMs responses to these measures may not be particularly different from other groups, although their behavior is different. Among the measures of offence related cognitive processes, the Sexual Offences Attitudes Questionnaire would seem a more robust measure than the Causal Dimensions Scale. The Sexual Offences Attitudes Questionnaire was found to have superior internal reliability and test-retest reliability, and its construct validity was supported by the pattern of correlations between its subscales.

Evaluation of the Psychometric Properties of Measures of Empathy Descriptive Statistics The descriptive data for the measures of empathy are shown in Table V. The mean scores for the Empat(general) indicated that there was little difference in general empathy between the groups. With the Mehrabian and Epstein Empathy Scale, however, the ASO group obtained a lower mean group score than the other groups. On the Empat(child), the CMs obtained a lower mean score than the other groups, indicating that they had less empathy for a nonspecific victim of child sexual abuse. Table V. Descriptive Statistics for the Measures of Empathy Among Child Molesters (CMs), Adult Sex Offenders (ASOs), Non-Sexual Offenders (NSOs), and Community (COM) Groups Measure General empathy Mehrabian and Epstein Empathy Scale Empat(general) Empathy for nonspecific victims of child sexual abuse Empat(child) Victim empathy Victim Empathy Distortions Scale

CMs, mean (SD)

ASOs, mean (SD)

NSOs, mean (SD)

33 (22.1)

20.7 (22.7)

28 (13.3)

34.5 (9)

59.7 (12.2)

62.3 (10.1)

63.5 (11)

64.7 (9.8)

123.7 (22.2)

131.3 (25.6)

129.6 (21.4)

63.2 (24.3)

NA

NA

100.4 (26) 82.7 (20.2)

COM, mean (SD)

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

510

Tierney and McCabe

Internal Reliability As shown in Table VI, the measures of empathy were found to have moderate to very high internal reliability across the four groups. Test-Retest Reliability The test-retest reliability of the empathy measures among the CMs ranged from moderate to very high for all measures: r = .58 for the Empat(child), r = .93 for the Empat(general), r = .82 for the Mehrabian and Epstein Empathy Scale, and r = .73 for the Victim Empathy Distortions Scale. Discriminant Validity There were no significant mean differences between the groups on the Empat(general), but on the Mehrabian and Epstein Empathy Scale the ASOs obtained significantly lower scores than both the CM (mean diff. = −14.29, p < .03) and the COM group (mean diff. = −15.99, p < 0.01). For the Empat(child), the CMs obtained significantly lower scores than each of the comparison groups [CMs vs. NSOs (mean diff. = −30.86, p < .05), CMs vs. COM (mean diff. = −29.23, p < .05), and CMs vs. ASOs (mean diff. = −22.8, p < .05)]. The difference between the NSO and COM groups and between the NSO and ASO groups on the Empat(child) were not significant (mean diff. = −1.21, p > .91, and mean diff. = 8.07, p > 0.68, respectively). Collectively, these results indicate that the CMs in comparison to the other groups were not deficient in general empathy, but reported significantly less empathy for a nonspecific victim of child sexual abuse than did the ASO, NSO, and COM groups. This is consistent with research with non-Australian samples that have found that CMs were not deficient in general empathy (Marshall et al., 1995), but were deficient in empathy for nonspecific victims (McGrath et al., 1998). The ASOs were found to have less general empathy Table VI. Cronbach Alphas for the Measures of Empathy Among Child Molesters (CMs), Adult Sex Offenders (ASOs), Non-Sexual Offenders (NSOs), and Community (COM) Groups

General empathy Mehrabian and Epstein Empathy Scale Empat(general) Empathy for nonspecific victims of child sexual abuse Empat(child) Victim empathy Victim Empathy Distortions Scale a Not

given to the NSO and COM groups.

CMs

ASOs

NSOs

COM

.76 .57

.69 .68

.65 .73

.58 .84

.93

.92

.69

.89

.91

.93

NAa

NAa

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

Assessment of Cognitive Processes and Empathy Among Sex Offenders

511

than the CM and COM groups on the Mehrabian and Epstein Empathy Scale, but not on the Empat(general). As discussed earlier, research using the Mehrabian and Epstein Empathy Scale with ASOs has been mixed, with some studies finding them to be deficient in general empathy and other studies failing to confirm this (Langevin et al., 1988; Marshall et al., 1995). Construct Validity Table VII shows the correlations between the measures of empathy. Both measures of general empathy correlated significantly with each other for all groups, and this supports the construct validity of the Empat(general), which is the less established measure. Neither of the two measures of general empathy, however, had significant correlations with the Empat(child) for any of the four groups or with the Victim Empathy Distortions Scale for the CMs and ASOs. This suggests that general empathy is a different construct from empathy for a nonspecific victim(s) of child sexual abuse, and also a different construct from the empathy CMs and ASOs have for their own sexual abuse victims. The pattern of correlations between the measures of general empathy and the Empat(child) and the Victim Empathy Distortions Scale questions the value of assessing general empathy among CMs. Among the CMs, the Empat(child) had a significant correlation with the Victim Empathy Distortions Scale, and this suggests that empathy for nonspecific victims of child sexual abuse was related to empathy for their own victims. This provides some support for the construct validity of these measures, and suggests the value in assessing empathy for nonspecific victims of child sexual abuse among CMs. With the ASOs, the Empat(child) did not correlate with the Victim Empathy Distortions Scale. This suggests that empathy for nonspecific victims of child sexual abuse was unrelated to empathy for their own adult victims among the Table VII. Correlations Between the Measures of Empathy Among Child Molesters (CMs), Adult Sex Offenders (ASOs), Non-Sexual Offenders (NSOs), and Community (COM) Groups

Measures of general empathy Empat(general) vs. Mehrabian and Epstein Empathy Scale General empathy vs. empathy for nonspecific victims Empat(general) vs. Empat(child) Mehrabian and Epstein Empathy Scale vs. Empat(child) General empathy vs. victim empathy Empat(general) vs. Victim Empathy Distortions Scale Mehrabian and Epstein Empathy Scale vs. Victim Empathy Distortions Scale Victim empathy vs. empathy for nonspecific victims Victim Empathy Distortions Scale vs. Empat(child) ∗p

< .05;

∗∗ p

< .01.

CMs

ASOs

NSOs

.43∗

.64∗∗

.45∗

.29 .08

−.02 .18

−.19 .26

−.09 .11

−.46∗

−.02

−.23 .01

COM .35∗ .01 .13

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

512

15:20

Style file version Nov. 19th, 1999

Tierney and McCabe

ASOs, and given that this would be expected, this also supports the validity of the Empat(child) and Victim Empathy Distortions Scale. Conclusions The measures of general empathy were found to have good internal reliability and good construct validity. However, the CMs were found not to be deficient in general empathy and this questions the value of assessing this variable with this group. The ASOs were deficient in general empathy on one measure but not the other, so this issue requires further investigation. Moreover, neither measure of general empathy was related to the measures of victim empathy or empathy for nonspecific victims of child sexual abuse, which suggests that they are different constructs, and again questions the value of assessing general empathy. The Empat(child) was found to have good internal reliability across the four groups, and its test-retest reliability with the CMs was also good. Its discriminative validity was supported by the fact that CMs were found to have significantly less empathy for nonspecific victims of child sexual abuse than the comparison groups. The construct validity of this scale was supported by its significant relationship with the Victim Empathy Distortions Scale among the CMs and by its non-significant relationship with the same measure among the ASOs. The Victim Empathy Distortions Scale was found to have good internal reliability with both sex offender groups and good test-retest reliability with the CMs. Its construct validity was supported by its pattern of correlations with the Empat(child).

Evaluation of the Influence of Sexual Social Desirability The psychometric evaluation of the Sexual Social Desirability Scale, described earlier, supports its use as a measure of the influence of sexual social desirability on responding. The relationship between the Acquiescence and Denial subscales of the Sexual Social Desirability Scale and the measures found to have the most sound psychometric properties are shown in Table VIII. The Abel and Becker Cognitions Scale did not correlate significantly with either the Acquiescence or Denial subscales for any of the four groups, suggesting that responding to this measure was not influenced by sexual social desirability. The subscales of the Sexual Offences Attitudes Questionnaire, however, had a number of significant correlations with the Acquiescence or Denial subscales. For both sex offender groups, the Admit subscale had a significant and negative correlation with the Acquiescence subscale, which suggests that admission of sexual offence(s) was negatively related to attributing positive but unlikely sexual behaviors to oneself. For the ASOs, the Minimize subscale had a significant correlation with the

< .01.

.02 −.47∗∗

−.15 .17

.08 .03

.16

−.12

.27 .25

.18

−.02 .03

−.27

−.05

−.26 −.28

−.18

Denial

COM

−.18

.26 .07

−.32

Acquiescence

−.15

.32 .00

−.08

Denial

15:20

< .05 ;

.44∗

−.33 −.28

−.37∗ −.26

NSO Acquiescence

June 21, 2001

∗∗ p

.10

.36∗

.46∗ .36

.16

.57∗∗

−.20

.33 .41∗

−.25

−.49∗∗

.15

−.40∗

−.06

.14

−.04

−.49

Denial

PP181-340733

∗p

Beliefs and attitudes Abel and Becker Cognitions Scale Offence related cognitions Sexual Offence Attitudes Questionnaire (admit) Sexual Offence Attitudes Questionnaire (minimize) Sexual Offence Attitudes Questionnaire (denial) General empathy Mehrabian and Epstein Empathy Scale Empat(general) Empathy for nonspecific child victims Empat(child) Measure of victim empathy Victim Empathy Distortions Scale

ASO Acquiescence

Denial

CM

Archives of Sexual Behavior [asb]

Acquiescence

Table VIII. Influence of Sexual Social Desirability on Responding: The Relationship Between the Acquiescence and Denial Subscales of the Sexual Social Desirability Scale Among Child Molesters (CMs), Adult Sex Offenders (ASOs), Non-Sexual Offenders (NSOs), and Community (COM) Groups

P1: GDB/GDX/LOV P2: FJY/GCR/GFU Style file version Nov. 19th, 1999

Assessment of Cognitive Processes and Empathy Among Sex Offenders 513

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

514

Tierney and McCabe

Acquiescence subscale, indicating that minimization of their offence(s) was related to a tendency to attribute positive but unlikely sexual behaviors to oneself among ASOs. For both the CMs and ASOs, the Denial subscale was significantly related to the Acquiescence subscale, which indicated that denial of sexual offence(s) was related to the tendency to attribute positive but unlikely sexual behaviors to oneself. Finally, among the CMs, their denial sexual offence(s) was also significantly related to the Denial subscale, or the tendency to deny negative but likely sexual behaviors. Among the measures of empathy, the Mehrabian and Epstein Empathy Scale had a significant negative correlation with the Acquiescence subscale for the CMs, whereas the Empat(general) had no significant relationships with either the Acquiescence or Denial subscales for any of the groups. The Empat(child) had no significant correlations with the Acquiescence or Denial subscale, and this indicated that empathy for nonspecific victims of child sexual abuse was not influenced by sexual social desirability across all four subject groups. Among the CMs, the Victim Empathy Distortions Scale had a significant and negative correlation with the Denial subscale, which suggested that empathy for their own victim(s) was negatively related to the tendency to deny negative but likely sexual behaviors. Finally, the ASOs’ responses to the Victim Empathy Scale were significantly related to the Acquiescence subscale, indicating that empathy for their own victim(s) was related to the tendency to attribute positive but unlikely sexual behaviors to themselves. Assessment of the Specificity of Measures Two strategies were used to investigate the issue of specificity. First, the discriminative power of measures was investigated and this property was discussed earlier. The second strategy was to assess the relationship between the measures found to have sound psychometric properties. The basis of this strategy was the assumption that if particular measures were specific to CMs and ASOs, then these measures should have clinically/theoretically meaningful relationships with each other. Some of the relationships between variables were discussed earlier, and thus will not be discussed again here. Rather, the focus of this discussion will be an assessment of the relationship between the measures of cognitive processes and the measures of empathy that are displayed in Table IX. Among the CMs, the Empat(child) and Victim Empathy Distortions Scale had significant correlations with the Abel and Becker Cognitions Scale, and this indicates that empathy for nonspecific victims of child sexual abuse and empathy for their own victim(s) were related to beliefs and attitudes about sexual relations between adults and children. The Empat(child) also had a significant negative relationship with the Denial subscale of the Sexual Offences Attitudes Questionnaire, and this indicates that the CMs’ empathy for nonspecific victims of child sexual

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

Assessment of Cognitive Processes and Empathy Among Sex Offenders

515

Table IX. Correlations Between the Measures of Cognitive Processes and Measures of Empathy Found to have Sound Psychometric Properties Among Child Molesters (CMs) Adult Sex Offenders (ASOs) Non-Sexual Offenders (NSOs) and Community (COM) Groups CMs Empat(child) vs. Abel and Becker Cognitions Scale Empat(child) vs. Sexual Offence Attitudes Questionnaire (admit) Empat(child) vs. Sexual Offence Attitudes Questionnaire (minimize) Empat(child) vs. Sexual Offence Attitudes Questionnaire (deny) Victim Empathy Distortions Scale vs. Abel and Becker Cognitions Scale Victim Empathy Distortions Scale vs. Sexual Offence Attitudes Questionnaire (admit) Victim Empathy Distortions Scale vs. Sexual Offence Attitudes Questionnaire (minimize) Victim Empathy Distortions Scale vs. Sexual Offence Attitudes Questionnaire (deny) ∗p

< .05;

∗∗ p

.59∗∗ .23 −.12 −.39∗ .46∗ .19

ASOs NA NA NA NA NA .48∗

−.47∗

−.40∗

−.19

.33

< .01.

abuse were negatively related to their denial of their sexual abuse offence(s). The Victim Empathy Distortions Scale had a significant positive relationship with the Abel and Becker Cognitions Scale, which indicated that the CMs’ empathy for their victim(s) were related to their beliefs and attitudes about sexual relations between adults and children. Finally, the Victim Empathy Distortions Scale also had a significant negative relationship with the Minimize subscale of the Sexual Offences Attitudes Questionnaire, and this suggests that the CMs’ empathy for their victim(s) were negatively related to the minimization of their sexual offences. Among the ASOs, the Victim Empathy Distortions Scale had a significant positive relationship with the Admit subscale and a significant negative relationship with the Minimize subscale of the Sexual Offences Attitudes Questionnaire. This suggests that the ASOs’ empathy for their victim(s) were related to their admission of their sexual offence(s) and negatively related to their minimization of their offence(s). Broadly, the pattern of correlations between the measures of cognitive processes and empathy are consistent with theoretical and clinical expectations, and therefore provide additional support for the specificity of the Empat(child) and the Abel and Becker Cognitions Scale to CMs, and the specificity of the Victim Empathy Distortions Scale and the Sexual Offences Attitudes Questionnaire to both the CM and ASO groups.

DISCUSSION The aim of this study was to investigate the psychometric properties, the influence of social desirability on responding, and the specificity of measures of cognitive processes, and empathy among a sample of Australian CMs and ASOs.

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

516

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

Tierney and McCabe

The results have a number of implications in relation to the assessment of cognitive processes and empathy among CMs and ASOs and these will be discussed later, as will the theoretical implications of this research. Implications for the Assessment Sex Offenders In a broad sense, the results highlight the need to evaluate the psychometric properties of measures used to assess CMs and ASOs. A number of measures that were found to have good psychometric properties in other studies (e.g. the Causal Dimensions Scale, McKay et al., 1996), were found to have poor psychometric properties with the Australian sample assessed in this study. In relation to the assessment of cognitive processes, the Sexual Social Desirability Scale was found to have good internal reliability across all groups and good test-retest reliability with CMs. Its discriminative power was not strong, however, the descriptive data suggested that CMs were the most likely to deny negative but likely sexual behaviors, and the second most likely of the four groups to attribute positive but unlikely sexual behaviors to themselves. To some degree this is consistent with the view that in assessment processes, CMs attempt to present themselves in a good light (Abel et al., 1975). Interestingly, the ASOs were the third most likely to attribute positive but unlikely sexual behaviors to themselves and were the least likely to deny negative but likely sexual behaviors to themselves. With the measures of attitudes and beliefs about sexual relations between adults and children, the Abel and Becker Cognitions Scale displayed stronger psychometric properties than the Child Molester Scale. However, an item analysis of these measures highlighted a number of issues worthy of further investigation, most particularly the failure of the Abel and Becker Scale to statistically differentiate the CMs from the ASOs and NSOs. The clinical utility of items where statistically significant differences between the groups were related to the extent of agreement and extent of disagreement rather than a clinical difference also needs to be explored further. The Sexual Offences Attitude Questionnaire was found to be a more robust measure of offence related cognitive variables than the Causal Dimension Scale for both the CM and ASO groups. It was found to have superior internal reliability and test-retest reliability, and its construct and face validity were supported by the pattern of correlations between its subscales. In relation to the assessment of empathy, the results clearly question the value of assessing general empathy among CMs who, compared to the other groups, were not deficient in this characteristic. With the ASOs, the results were mixed—on one measure of general empathy they were found to be deficient but on the other they were not deficient. This issue requires further investigation. Interestingly, however, across all groups, general empathy was unrelated to empathy for nonspecific victims of child sexual abuse and unrelated to the sex offender’s empathy for his

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

Assessment of Cognitive Processes and Empathy Among Sex Offenders

517

victim(s). This suggests that general empathy is a different construct to empathy for nonspecific victims of child sexual abuse, and also a different construct to the CMs’ and ASOs’ empathy for their own victim(s). Again this questions the value of assessing general empathy. The measure of empathy for nonspecific victims of child sexual abuse [Empat(child)] was found to have good internal reliability across all subject groups, good test-retest reliability with CMs, and its pattern of correlations with the Victim Empathy Scale for both sex offender groups provided support for its construct validity. The Victim Empathy Scale was found to have good internal reliability with both sex offender groups and good test-retest reliability with the CMs, and its relationship with the Empat(child) among the CMs and ASOs provided support for its construct validity. The psychometric properties of the Sexual Social Desirability Scale support its use for assessing the influence of social desirability. It was found to be reliable, and support for its construct validity was strong. Also, its items relate to sexual behavior, and thus would seem preferential to measures of general social desirability, such as the Marlowe Crowne Social Desirability Scale. A number of the measures that were found to have the most robust psychometric properties were found to be influenced by the offenders desire to present themselves as displaying sexually desirable behavior, and this highlights the value in assessing this factor (Stermac et al., 1990). Further investigation of the psychometric properties of the Sexual Social Desirability Scale, Sexual Offences Attitudes Questionnaire, Empat(child), and the Victim Empathy Distortions Scale are required. For example, it would be valuable to investigate the test-retest reliability of these measures among ASOs and to further assess their construct validity against therapists’ ratings of these variables. Also, it would be valuable to investigate how responses to these measures are impacted upon by treatment participation and how they relate to recidivism. Theoretical Implications of Results The results have a number of implications in relation to an understanding of empathy. Clearly, the results suggest that empathy is a complex construct. First, across all groups, general empathy was not significantly related to empathy for nonspecific victims of child sexual abuse. Second, among the sex offender groups, general empathy was not significantly related to their empathy for their sexual abuse victim(s). These results suggest that there are different types of empathy that are issue-specific, and that the relationship between these different types of empathy may not be strong. Second, a number of significant relationships were found between the more robust measures of cognitive processes and empathy. These relationships were broadly consistent with theoretical expectations about a relationship between

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

518

Style file version Nov. 19th, 1999

Tierney and McCabe

cognitive and emotive processes. This is an area worthy of further investigation, as it may enhance our understanding of the links between cognitive and emotive processes as they relate to sexual offending behavior. In conclusion, the Sexual Social Desirability Scale and the Sexual Offences Attitudes Questionnaire were found to have the most robust psychometric properties among the measures of cognitive processes. Support for the Abel and Becker Cognitions Scale was mixed and so further evaluation of this scale is required. Among the measures of empathy, The Empat(child) and Victim Empathy Distortions Scale were found have the most robust psychometric properties. The mixed results with the measures of general empathy among the ASOs suggest the need for further investigation. The psychometric properties of the Sexual Social Desirability Scale support its use as a measure of the influence of sexual social desirability on responding, and the results suggest that this factor was evident in the responses of both sex offender groups. Finally, assessments of discriminant validity and/or the relationship between measures, provide support for the specificity of the Victim Empathy Distortions Scale and the Sexual Offences Attitudes Questionnaire to both CMs and ASOs, and the Empat(child) and the Abel and Becker Cognitions Scale to the CM group. REFERENCES Abel, G., Becker, J., Blanchard, E., and Mavissakalian, M. (1975). Measurement of sexual arousal in male homosexuals: Effects of instructions and stimulus modality. Arch. Sex. Behav. 4: 623– 629. Abel, G. G., Becker, J. V., Cunningham-Rathner, J., Rouleau, J. L., Kaplan, M., and Reich, J. (1984). The Treatment of Child Molesters (available from Gene G. Abel, MD., Behavioral Medicine Institute, Paces Pavilion, Suite 202, 3195 Howell Mill Road, N.W., Atlanta, GA 30327). Beckett, R. C., and Fisher, D. (1994). Assessing victim empathy: A new measure. Paper presented at the 13th Conference of the Association for Treatment of Sexual Abusers (ATSA), San Francisco, USA. Beckett, R. C., Beech, A. R., Fisher, D., and Fordham, A. S. (1994). Community-based Treatment for Sex Offenders: An Evaluation of Seven Treatment Programs, Home Office Publication Unit, London. Crowne, D. P., and Marlowe, D. (1960). A new scale of social desirability independent of psychopathology. J. Consult. Psychol. 24: 349–354. Fernandez, Y. M., Marshall, W. L., Lightbody, S., and Sullivan, C. (1999). The Child Molester Empathy Measure: Description and examination of its reliability and validity. Sex. Abuse: J. Res. Treat. 11: 17–31. Gendreau, P., Irvine, M., and Knight, S. (1973). Evaluating response sets styles on the MMPI with prisoners: Faking good adjustment and maladjustment. Can. J. Behav. Sci. 5: 183–193. Home Office. (1996). Sexual Offense Attitudes Questionnaire, Unpublished paper. Hudson, S. M., Marshall, W. L., Wales, D. S., McDonald, E., Bakker, L. W., and McLean, A. (1993). Emotion recognition skills of sex offenders. Annals Sex Res. 6: 199–211. Kennedy, H. G., and Grubin, D. H. (1992). Patterns of denial in sex offenders. Psychol. Med. 22: 191–196. Langevin, R., Wright, P., and Handy, L. (1988). Empathy, assertiveness, aggressiveness, and defensiveness among sex offenders. Annals Sex Res. 1: 533–547. Larsen, J., Hudson, S. M., and Ward, T. (1995). Evaluation of attributional change in a relapse prevention program for child molesters. Behav. Change 12: 127–138.

P1: GDB/GDX/LOV

P2: FJY/GCR/GFU

Archives of Sexual Behavior [asb]

PP181-340733

June 21, 2001

15:20

Style file version Nov. 19th, 1999

Assessment of Cognitive Processes and Empathy Among Sex Offenders

519

Marshall, W. L., Hudson, S. M., Jones, R. L., and Fernandez, Y. M. (1995). Empathy in sex offenders. Clin. Psychol. Rev. 15: 99–113. McGrath, M., Cann, S., and Konopasky, R. (1998). New measures of defensiveness, empathy, and cognitive distortions for sexual offenders against children. Sex. Abuse: J. Res. Treat. 10: 25–36. McKay, M. M., Chapman, J. W., and Long, N. R. (1996). Causal attributions for criminal offending and sexual arousal: Comparison of child sex offenders with other offenders. Brit. J. Clin. Psychol. 35: 63–75. Mehrabian, A., and Epstein, N. (1972). A measure of emotional empathy. J. Person. 40: 525–543. Murphy, W. D. (1990). Assessment and modification of cognitive distortions in sex offenders. In Marshall, W. L., Laws, D. R., and Barbaree, H. E. (eds.), Handbook of Sexual Assault: Issues, Theories, and Treatment of the Offender, Plenum Press, New York, pp. 331–342. Owen, K. J. (1997). Convicted Incarcerated Sex Offenders: The Effect of Incarceration and Treatment on Denial, Unpublished Dissertation, Graduate Diploma of Psychology, Deakin University, Melbourne, Australia. Russell, D. (1982). The causal dimensions scale: A measure of how individuals perceive causes. J. Person. Soc. Psychol. 42: 1137–1145. Salter, A. C. (1988). Treating Child Sex Offenders and Victims: A Practical Guide, Sage, Newbury Park, CA. Stermac, L. E., and Segal, S. V. (1987). Cognitive assessment of child molesters. Paper presented at the 21st Annual Convention of the Association for the Advancement of Behavior Therapy, Boston. Stermac, L. E., Segal, S. V., and Gillis, R. (1990). Social and cultural factors in sexual assault. In Marshall, W. L., Law, D. R., and Barbaree, H. E. (eds.), Handbook of Sexual Assault: Issues, Theories, and Treatment of the Offender, Plenum Press, New York, pp. 143–160. Tierney, D. W. (1999). An investigation of motivation for change among men convicted of sexual offences against children. Unpublished paper, Deakin University, Melbourne, Australia. Tierney, D. W., and McCabe, M. P. (Unpublished). The assessment of cognitive distortions and victim empathy among pedophilic sex offenders: An evaluation of the utility of self-report measures. Weiner, B. (1986). An Attributional Theory of Motivation and Emotion, Springer-Verlag, New York.

P1: GDX/LZX/GGN/GDP/LOV/GGT Archives of Sexual Behavior [asb]

P2: GFU PP181-340734

June 28, 2001

9:6

Style file version Nov. 19th, 1999

Archives of Sexual Behavior, Vol. 30, No. 5, 2001

Victim-Choice Polymorphia Among Serious Sex Offenders Jean-Pierre Guay, MSc,1,2,3 Jean Proulx, PhD,1,2 Maurice Cusson, PhD,1 and Marc Ouimet, PhD1

The victim-choice polymorphia of 178 sexual aggressors divided into six subtypes, incest offenders, pseudoincest offenders, sexual aggressors of familiar children, sexual aggressors of unfamiliar children, sexual aggressors of familiar women, and sexual aggressors of unfamiliar women, was compared. Results showed that sex offenders remained stable in their choice of victim from one offence to another in terms of victim age, victim gender, and aggressor–victim relationship. Subjects characterised by high levels of polymorphia were pseudoincest offenders and sexual aggressors of familiar women, whereas sexual aggressors of both unfamiliar women and unfamiliar children were characterised by low levels of polymorphia. Recommendations regarding how to further refine sex offender typologies are discussed. KEY WORDS: sex offenders; specialization; offence patterns; polymorphia.

INTRODUCTION Researchers who study criminal conduct are essentially divided into two camps over the issue of offence-type stability. On the one hand, it is believed that the commission of crimes over a lifetime is unpatterned or “cafeteria-style” (Klein, 1984), that is, that delinquents tend to show versatility in their choice of offence. For the advocates of this hypothesis, the next crime of a person who has committed burglary is just as likely to be burglary as it is any other. It is argued that criminals ´ de Montr´eal, Ecole de criminologie, Centre International de Criminologie Compar´ee, Montr´eal, Qu´ebec, Canada. 2 Institut Philippe-Pinel de Montr´ eal, Centre de Recherche Philippe Pinel, Montr´eal, Quebec, Canada. 3 To whom correspondence should be addressed at Institut Philippe-Pinel de Montr´ eal, Centre de Recherche Philippe Pinel, 10 905 Boul. Henri-Bourassa est, Montr´eal, Qu´ebec, Canada H1C 1H1; e-mail: [email protected]. 1 Universit´ e

521 C 2001 Plenum Publishing Corporation 0004-0002/01/1000-0521$19.50/0 °

P1: GDX/LZX/GGN/GDP/LOV/GGT Archives of Sexual Behavior [asb]

522

P2: GFU PP181-340734

June 28, 2001

9:6

Style file version Nov. 19th, 1999

Guay, Proulx, Cusson, and Ouimet

offend without following any particular trend or pattern, merely seizing the criminal opportunities that arise at random. On the other hand, researchers studying criminal careers believe that there are several patterns of offending (Blumstein et al., 1988 a,b,c; Farrington, 1992; Greenberg, 1991, 1992). Gottfredson and Hirschi (1990) advanced that versatility in offending is the result of various shortcomings due to lack of self-control, which are particularly prominent within criminal populations. Individuals lacking self-control are much more likely to be involved in various types of crimes and deviant behaviors, including rape, theft, and even murder. Moreover, they have a strong tendency to drink, smoke, and get involved in all sorts of accidents, including household fires, automobile collisions, and unwanted pregnancies. According to Gottfredson and Hirschi, cases of “criminal specialization” may be attributed to “obvious opportunities,” as evidenced by a purse-snatcher living near a shopping centre (Gottfredson and Hirschi, 1990, p. 92). Although general theories of crime such as that by Gottfredson and Hirschi (1990) may exert a certain appeal initially, criminal conduct is far too heterogeneous a phenomenon to be accounted for through a single concept such as lack of self-control. Although the propensity for crime may be partly or largely due to a lack of self-control for a certain proportion of offenders, many do not seem to fit the portrait of the impulsive, self-centred, and opportunistic delinquent (Hanson et al. 1995). There is no denying that certain criminals tend to develop patterns of delinquency or preferences for certain types of crimes (Blumstein et al., 1988a). This is evident when offenders are examined from the standpoint of criminal career, which Blumstein et al. (1988a, p. 304) defined as the “longitudinal sequence of crimes committed by an individual offender, coupled with any intervention by the criminal justice system that may be associated with those crimes.” The criminal career approach focuses on onset and prevalence of delinquency, desistence from delinquency, and frequency and diversity of criminality (Blumstein et al., 1988a; Farrington, 1992). Although the versatility or criminal propensity approach and the criminal career approach may at first sight be at odds in explaining criminal conduct over the life span, each can be used for different purposes. According to Farrington (1992), the criminal propensity approach is most applicable in cross-sectional studies of self-reported delinquency involving teenagers, and the criminal career approach applies more to specific populations. “The criminal career approach is more appropriate to research that focuses more narrowly on serious offences, where the prevalence is lower and where the distinction between active offenders and true nonoffenders is more important” (Farrington, 1992, p. 522). One of the most cited examples of crimes supporting the stability hypothesis is certainly sexual offending (Farrington, 1992; Farrington et al., 1988; Hanson et al., 1995; Soothill et al., 2000). According to Hanson et al. (1995), child molesters are different from nonsexual criminals not only in terms of background variables (e.g.,

P1: GDX/LZX/GGN/GDP/LOV/GGT Archives of Sexual Behavior [asb]

P2: GFU PP181-340734

Victim-Choice Polymorphia

June 28, 2001

9:6

Style file version Nov. 19th, 1999

523

marital status and education), but also with respect to type of recidivism. Compared with nonsexual criminals, child molesters are more frequently rearrested for sexual crimes (35% vs. 1.5%) and less frequently for nonsexual violent crimes (1% vs. 33%) (Hanson et al., 1995). Whereas sexual aggressors of children may be paragons of stable offenders, sexual aggressors of women tell a different story. Hall and Proctor (1987) noted a greater variety of criminal acts in the latter than in the former. Furthermore, Barbaree et al. (1994) found that 37.3% of their sample of sexual aggressors of women were classified as opportunistic, which means that the offence was “an impulsive, typically unplanned, predatory act, controlled more by contextual and immediately antecedent factors than by any obvious protracted or stylised sexual fantasy” (Knight and Prentky, 1990, p. 44). However, even if sexual aggressors of women tend to be more involved than sexual aggressors of children in nonsexual crimes, a certain crime-type stability (or specialization) can still be observed in the former (Hall and Proctor, 1987). In an examination of the criminal careers of a sample of incarcerated sexual aggressors, Boutin (1999) found that sexual aggressors of children had committed a mean of 5.1 sex crimes, compared with 1.6 violent crimes (e.g., homicide and assault), 2.6 property crimes (e.g., burglary and auto theft), and 3.4 “other” crimes (e.g., parole violation, loitering, and drunk driving). These inmates were also more likely to be convicted for sexual crimes if their previous crime was also sexual. In contrast, sexual aggressors of adult women had committed a mean of 2.3 sex crimes, compared with 4.4 violent crimes, 5.1 property crimes, and 6.7 other crimes. These results showed that sexual aggressors of women had a tendency to be more versatile than sexual aggressors of children in their pattern of offending. Although sex offenders as a whole may appear stable in their choice of crime (i.e., sex offenders tend to commit sexual crimes nearly exclusively), the sexual offences they commit are of a wide variety (Barbaree et al., 1994; Knight et al., 1985; Knight and Prentky, 1990; Proulx et al., 1999a,b,c). In this regard, Gebhard et al. (1965) classified sex offenders on the basis of four main aspects: age of victim, sex of victim, level of coercion used to commit offence, and presence of an incestuous relationship between aggressor and victim. Their results showed that homosexual and incest offenders maintained the highest stability in their choice of victims, being “rather rigidly departmentalised in their sex-offence behaviour” (Gebhard et al., 1965, p. 708). Offenders found to be the least specialized were violent sexual aggressors of children and adolescents, with, respectively, 46 and 51% of their sexual offences targeting other types of victims. While certain sex offenders are considered stable in their choice of victims with respect to sex and age, others seem to have multiple paraphilia. Abel et al. (1987, 1988) found, for instance, that sexual aggressors of women had often had prepubescent victims and a history of exhibitionism and voyeurism. In addition,

P1: GDX/LZX/GGN/GDP/LOV/GGT Archives of Sexual Behavior [asb]

P2: GFU PP181-340734

June 28, 2001

524

9:6

Style file version Nov. 19th, 1999

Guay, Proulx, Cusson, and Ouimet

they also noted that a significant proportion of sex offenders shifted in their victim type. For example, among nonincestuous child molesters of a female victim, 35% had a history of nonincestuous child molesting against a male victim, 35% had incestuous relationships with a female victim, 12% had been involved in incest with a male victim, and 25% had committed rape. Compared with Abel et al., Marshall et al. (1991) reported a significantly lower proportion of subjects with multiple paraphilia: 14.0% versus 84.8% for heterosexual child molesters; 11.8% versus 71.7% for homosexual child molesters; and 7.9% versus 95.5% for incest offenders. Finally, Bradford et al. (1992) found, based on 443 subjects admitted in a Canadian hospital for forensic evaluation, that most of paraphiliacs tend to have multiple types of sexual aberrations. The aim of our study was twofold: (1) investigate victim polymorphia, that is, the degree to which sex offenders tend to switch from one victim type to another; and (2) identify the most common victim type transitions, as well as the least common or probable. METHOD Subjects Subjects in this study were 178 male sex offenders with more than one victim for which they were currently incarcerated at the Regional Reception Centre (RRC) in Ste-Anne-des-Plaines (Quebec), a maximum-security facility of Correctional Services of Canada. They were serving a mean sentence of 51.6 months (SD = 36.4). Treatment needs and correctional risk levels were evaluated during a 4- to 6-week procedure. At the time of evaluation, their mean age was 42.8 years (SD = 13.7; range of 20–74), 35.4% were married or living in common-law, and 64.6% were single or widowed. No subject was mentally retarded or severely mentally disordered (e.g., psychotic or mood disorder). Data sources were official records (police records, victims’ statements) and interviews with offenders. In the event of divergence, official information was given priority over self-reported information. Sex offenders were classified according to four criteria: age of official victims, aggressor–victim relationship, gender of the victims, and presence of an incestuous relationship between aggressor and victim. Sexual aggressors of children were those implicated with prepubertal victims aged under 13 years, hebephiles with pubertal victims aged 13–15 years, and rapists with adult victims aged 16 years or more. Those who were related to their victim as acquaintances, family members, friends, persons in a position of authority, husbands, ex-husbands, boyfriends, and ex-boyfriends were considered aggressors of familiar victims, whereas those who were related to their victims as neighbours and strangers were considered aggressors of unfamiliar victims. Finally, subjects who abused their biological

P1: GDX/LZX/GGN/GDP/LOV/GGT Archives of Sexual Behavior [asb]

P2: GFU PP181-340734

June 28, 2001

9:6

Style file version Nov. 19th, 1999

Victim-Choice Polymorphia

525

son or daughter or their spouse’s son or daughter were classified as incest and pseudoincest offenders, respectively. Transition Matrices To estimate the level of variety in the choice of victims for each of the eight proposed subtypes, matrices for the transition from one crime to another were used for the current sentence. Although for certain subjects up to seven victims were implicated in the current incarceration, only the first four were considered in the study. RESULTS First, we examined the polymorphia of offending in terms of age of victim. Table I shows the matrices for the transitions from Victim 1 to Victim 2, Victim 2 to Victim 3, and Victim 3 to Victim 4. Figures running diagonally from left to right represent the probability of targeting the same type of victim as the previous one. As we can see, these figures are much higher than average probabilities in the rest of Table I, which strongly suggests stability in the choice of victims in terms of age. Table I shows that subjects with prepubertal victims tended to stick with the same type of victim over a sequence of crimes, as evidenced by the percentages of 78.7% in the first transition, 73.9% in the second, and 83.3% in the third. Subjects aggressing adult victims also tended to remain stable, with 75.0, 64.7, and 83.3% choosing a victim of the same age category at each subsequent transition. However, subjects with pubertal victims were much less stable in their choice of victim, the probability of subsequently targeting a victim in the same age group being 47.1, 66.7, and 38.9% in the first, second, and third transitions, respectively. Table I. Age Categories Transition Matrix Transition probabilities First transition (n = 178) Prepubertal victim Pubertal victim Adult victim Second transition (n = 81) Prepubertal victim Pubertal victim Adult victim Third transition (n = 58) Prepubertal victim Pubertal victim Adult victim

Prepubertal victim

Pubertal victim

Adult victim

Total

78.7% (85) 29.4% (10) 8.3% (3)

19.4% (21) 47.1% (16) 16.7% (6)

1.9% (2) 23.5% (8) 75.0% (27)

100% (108) 100% (34) 100% (36)

73.9% (33) 33.3% (6) 0.0%

17.4% (9) 66.7% (12) 35.3% (6)

8.7% (4) 0.0% 64.7% (11)

100% (46) 100% (18) 100% (17)

83.3% (20) 55.6% (10) 8.3% (1)

8.3% (2) 38.9% (7) 8.3% (1)

8.3% (2) 5.6% (1) 83.3% (10)

100% (24) 100% (18) 100% (12)

P1: GDX/LZX/GGN/GDP/LOV/GGT Archives of Sexual Behavior [asb]

P2: GFU PP181-340734

June 28, 2001

526

9:6

Style file version Nov. 19th, 1999

Guay, Proulx, Cusson, and Ouimet Table II. Relationship Categories Transition Matrix Transition probabilities First transition (n = 175) Unfamiliar victim Familiar victim Second transition (n = 79) Unfamiliar victim Familiar victim Third transition (n = 54) Unfamiliar victim Familiar victim

Unfamiliar victim

Familiar victim

Total

86.0% (98) 8.2% (5)

14.0% (16) 91.8% (56)

100% (114) 100% (61)

90.2% (37) 15.8% (6)

9.8% (4) 84.2% (32)

100% (41) 100% (38)

72.4% (21) 8.0% (2)

27.6% (8) 92.0% (23)

100% (29) 100% (25)

Second, we examined the polymorphia of offending relative to the relationship between aggressor and victim. Figures running diagonally from left to right represent the probability of targeting the same type of victim as the previous one. Clearly, offenders tended to remain stable from one crime to the next. Nearly 92% of the subjects who targeted familiar victims remained consistent in their choice of victim at the second offence (see Table II). Percentages remained high for the second and third transitions as well (84.2 and 92.0%, respectively). A similar picture of stability was observed with the subjects who targeted unfamiliar victims: 86.0% at the first transition, 90.2% at the second, and 72.4% at the third. One possible point of interest was the relatively higher probability of transition from an unfamiliar to a familiar victim (range of 9.8–27.6%) than vice versa (range of 8.0–15.8%). Third, polymorphia as to the gender of the victims was examined. Table III shows that the proportion of subjects being bisexual aggressors tends to remain small and stable from one transition to another. In fact, approximately 10% of the subjects were switching from female to male, or from male to female victims. Thus, most of the subjects are stable as to the gender of their victims. To restrict the number of categories for subsequent analyses, subjects who targeted prepubertal and pubertal victims were merged into a single category on the basis of victim age: sexual aggressors of children. Table IV presents a victim-type Table III. Victim Gender Categories Transition Matrix Transition probabilities First transition (n = 177) Female victim Male victim Second transition (n = 81) Female victim Male victim Third transition (n = 54) Female victim Male victim

Female victim

Male victim

Total

91.3% (115) 15.7% (8)

8.7% (11) 84.3% (43)

100% (126) 100% (51)

90.0% (45) 12.9% (4)

10.0% (5) 87.1% (27)

100% (50) 100% (31)

92.9% (26) 11.5% (3)

7.1% (2) 88.5% (23)

100% (28) 100% (26)

6.1% (2) 62.5% (20) 2.6% (1) 2.7% (1) 30.0% (3) 0.0% 0.0% 54.5% (6) 4.0% (1) 6.3% (1) 0.0% 0.0% 11.1% (1) 80.0% (4) 0.0% 8.3% (1) 0.0% 0.0%

88.9% (8) 27.3% (3) 12.0% (3) 25.0% (4) 0.0% 0.0% 33.3% (3) 0.0% 0.0% 0.0% 0.0% 0.0%

Pseudoincest

72.7% (24) 12.5% (4) 0.0% 5.4% (2) 0.0% 0.0%

Incest

0.0% 0.0% 0.0% 8.3% (1) 100.0% (2) 0.0%

11.1% (1) 0.0% 5.9% (1) 0.0% 0.0% 100.0% (7)

0.0% 0.0% 0.0% 0.0% 0.0% 75.0% (9)

3.0% (1) 0.0% 10.5% (4) 0.0% 20.0% (2) 90.5% (19)

Unfamiliar women

100% (9) 100% (5) 100% (17) 100% (12) 100% (2) 100% (7)

100% (9) 100% (11) 100% (25) 100% (16) 100% (2) 100% (12)

100% (33) 100% (32) 100% (38) 100% (37) 100% (10) 100% (21)

Total

9:6

33.3% (3) 0.0% 11.8% (2) 41.7% (5) 0.0% 0.0%

0.0% 0.0% 0.0% 0.0% 50.0% (1) 0.0%

0.0% 3.1% (1) 2.6% (1) 5.4% (2) 30.0% (3) 0.0%

Familiar women

June 28, 2001

11.1% (1) 20.0% (1) 82.4% (14) 41.7% (5) 0.0% 0.0%

11.1% (1) 9.1% (1) 8.0% (2) 56.3% (9) 50.0% (1) 0.0%

9.1% (3) 9.4% (3) 7.9% (3) 73.0% (27) 10.0% (1) 0.0%

Familiar children

PP181-340734

0.0% 9.1% (1) 76.0% (19) 12.5% (2) 0.0% 25.0% (3)

9.1% (3) 12.5% (4) 76.3% (29) 13.5% (5) 10.0% (1) 9.5% (2)

Unfamiliar children

Archives of Sexual Behavior [asb]

First transition (n = 171) Incest Pseudoincest Unfamiliar children Familiar children Familiar women Unfamiliar women Second transition (n = 75) Incest Pseudoincest Unfamiliar children Familiar children Familiar women Unfamiliar women Third transition (n = 52) Incest Pseudoincest Unfamiliar children Familiar children Familiar women Unfamiliar women

Transition probabilities of K − 1/K th victim

Table IV. Victim Type Transition Matrix

P1: GDX/LZX/GGN/GDP/LOV/GGT P2: GFU Style file version Nov. 19th, 1999

Victim-Choice Polymorphia 527

P1: GDX/LZX/GGN/GDP/LOV/GGT Archives of Sexual Behavior [asb]

P2: GFU PP181-340734

June 28, 2001

9:6

528

Style file version Nov. 19th, 1999

Guay, Proulx, Cusson, and Ouimet

transition matrix for six subtypes of sexual aggressions. Those who abused a biological son or daughter were assigned to the incest category, while those who abused a spouse’s son or daughter were classified under the pseudoincest category. In the end, the six categories used to assess level of offender polymorphia over time in terms of victim type were the following: incest, pseudoincest, sexual aggression of a familiar child (nonincest), sexual aggression of an unfamiliar child, sexual aggression of a familiar woman and, finally, sexual aggression of an unfamiliar woman. The probability of targeting the same type of victim varied considerably from one transition to the next, with figures ranging from 30.0 to 90.5%. This attested to the varying degrees of polymorphia that exist within a sample of incarcerated sex offenders. Where aggressors of women were concerned, those targeting unfamiliar victims were clearly the more stable: 90.5% maintained their choice of an unfamiliar adult woman in the first transition, and 75% in the second. A certain percentage of aggressors of women were nevertheless unstable as they subsequently turned toward children. It was interesting to note that whenever a change in victim type occurred, sexual aggressors of unfamiliar women always switched to an unfamiliar child. Further individual analysis of these cases indicated that the victims were all female teenagers, of which three were aged 14 years and two were aged 15. None of these subjects targeted prepubertal or male victims. Whenever they changed victim type, they opted for a type similar to adult female, that is, an unfamiliar teenage girl. Aggressors of familiar women came in at the opposite end of the polymorphia spectrum with the lowest probability of subsequently targeting the same type of victim. Table IV shows that they had a 30.0% chance of assaulting the same type of victim in the first transition.4 Pseudoincest offenders, too, presented a relatively low probability of repeating the same victim type, with figures reaching 62.5 and 54.5%, respectively, in the first and second transitions. Incest offenders and aggressors of familiar children victims showed a declining stability pattern. The former went from 72.7 to 88.9 to 33.3% over the three transitions, and the latter from 73.0 to 56.3 to 41.7%. DISCUSSION This study evidenced considerable diversity in the level of victim-choice polymorphia among incarcerated sex offenders. In this regard, our results concur with those by Abel et al. (1987, 1988), Bradford et al. (1992), and Marshall et al. (1991). Our study also showed that serious sex offenders tend to maintain their choice of victim with respect to age: Aggressors of children tend to reoffend against children, and aggressors of adult women tend to reoffend against adult women. However, sex offenders who target pubertal victims tend to be less stable than 4 Owing

to the small number of subjects, the second and third transitions were not taken into account.

P1: GDX/LZX/GGN/GDP/LOV/GGT Archives of Sexual Behavior [asb]

P2: GFU PP181-340734

Victim-Choice Polymorphia

June 28, 2001

9:6

Style file version Nov. 19th, 1999

529

those who target women or children. This particular type of aggressor raises certain questions regarding victim choice. Two major hypotheses may explain their low probability (38.9–66.6%) of choosing the same type of victim repeatedly. First, hebephiliacs may not constitute a consistently distinct type of sexual offender. Second, the choice of a pubertal victim may be either a compromise when a “preferred” type of victim is not available or a decision determined by circumstance. Our results showed, however, that subjects with pubertal victims only were mainly homosexual, which to some extent supports the “subcultural hebephiles” subtype proposed by Gebhard et al. (1965). Sex offenders also tend to remain stable over time with regard to aggressor– victim relationship. However, among those who changed, most shifted from a familiar to an unfamiliar victim. A portion of these subjects reported that the first contacts with a familiar victim were “accidental,” which would in part explain the results. Having experienced pleasure or sexual gratification with their first familiar victim, some of them were eager to search for other similar victims or victims more compatible with their sexual preferences. Another explanation could be that certain subjects were forced to opt for what Proulx and Ouimet (1995) described as an alternative offending scenario. Constrained by the situational context, some may have decided to bypass their preferred victim-type and offending pattern and select another type of victim as a sex surrogate (Freund et al., 1972). Results yielded by our six-type classification based on victim age and aggressor–victim relationship also showed substantial heterogeneity in the probabilities of victim choice from one offence to the next. Certain offenders, such as aggressors of unfamiliar victims, tended to remain stable. Clearly, the most stable in their choice of victims were aggressors of unfamiliar adult women. Such consistency raises the hypothesis of predatory aggression, where the selection of victims is carefully made on the basis of personal characteristics. As for aggressors of unfamiliar children, a possible explanation could be that a high proportion of subjects classified as aggressors of unfamiliar victims were homosexual child molesters (20 out of 29 in the first transition and 12 out of 19 in the second). Their marital status may have favored stability in their victim’s choice. In fact, a considerable proportion of them were never married (51.7% for the first transition and 42.1% for the second). Consequently, they never had access to incest or pseudoincest victims. This same type of offender has elsewhere been labelled a fixated child molester (Cohen et al., 1969; Groth et al., 1982) and a high-fixation subject with a high number of contacts and with interpersonal meaning (Knight and Prentky, 1990). For aggressors of unfamiliar children and aggressors of unfamiliar women, deviant sexual preferences also could explain enduring and stable victim-type selection. Furthermore, our results indicated that pseudoincest offenders were, along with aggressors of familiar women, the subjects most characterized by high levels of polymorphia. Studies of differences between incest offenders and pseudoincest offenders may help explain these results (Poirier et al., 1997; Studer et al., 2000).

P1: GDX/LZX/GGN/GDP/LOV/GGT Archives of Sexual Behavior [asb]

P2: GFU PP181-340734

June 28, 2001

9:6

530

Style file version Nov. 19th, 1999

Guay, Proulx, Cusson, and Ouimet

Poirier et al. (1997) showed that most pseudoincest offenders were comparable to extrafamilial child molesters and significantly different from biological-incest offenders in terms of personality. These results raise an interesting hypothesis regarding the sexual abuse of children and new familial realities: broken homes may provide a certain type of child molester with the opportunity to “infiltrate” and abuse children within the family. Finally, our results not only revealed victim-choice polymorphia in certain subtypes of offenders but also specific patterns or profiles of transition. Certain offenders tend to switch from one type of victim to another, but not at random. Incest offenders remain stable in the first and second transitions, but tend to select familiar children in the third. As mentioned earlier, the desire for victims more compatible with their fantasies and the unavailability of new victims may push these offenders to target children close at hand. LIMITATIONS In addition to our preoccupations regarding the small number of subjects, especially in the second and third transitions, we had three major concerns regarding our data. Firstly, crimes were classified according to victim type based on official records, although many subjects confessed to additional victims during the evaluation at the RRC. It is possible that the use of official records introduces a bias in favor of stability, especially for aggressors of unfamiliar victims. When tracking down sexual aggressors, investigators have a tendency to connect similar crimes to a given perpetrator through modus operandi or “the offender’s signature.” Consequently, it stands to reason that the probability of being accused of similar crimes on similar types of victims is greater than the probability of being accused of a different type of crime on a different type of victim (which most of the time is committed under different circumstances). For example, while police may suspect that the rape of a 20-year-old woman was committed by an aggressor who just admitted attacking a similar victim, only in rare circumstances will police connect and interrogate this type of aggressor with the molestation of an 8-year old boy in a nearby park. Hence, what we might call a “police investigation bias” may have inflated the level of victim-choice stability among our subjects. Secondly, we had difficulty comparing sexual crimes that are intrinsically different. If we consider sexual aggressors to form a particularly heterogeneous group, we should recognise the same heterogeneity in sexual aggressions. A key factor that helps differentiate sexual aggressions is situational context, which in large part is determined by the relationship between aggressor and victim. One of the major difficulties encountered in comparing victim selection patterns stemmed from the very different situational contexts in which aggressions occurred. In the case of child abuse, situational context (especially familial) often determines victim availability. Choosing his girlfriend’s daughter as a victim allows an aggressor to

P1: GDX/LZX/GGN/GDP/LOV/GGT Archives of Sexual Behavior [asb]

P2: GFU PP181-340734

June 28, 2001

9:6

Victim-Choice Polymorphia

Style file version Nov. 19th, 1999

531

commit repeated aggressions, and his authority over the child allows him to do so over a long period of time, thereby reducing his need to search for another victim. Comparing multiple- and single-event aggressions resulted in a methodological complication inherent to our decision to classify subjects on the basis of their relationship with their victims. Although we considered this measure imperfect, it nevertheless served to bring out differences across the proposed subtypes. We could have hypothesized that subjects in situations allowing repeated contact with the same victim would have low switching-pattern probabilities compared with single-event aggressors. It is interesting to note, instead, that results showed substantial variability among both multiple- and single-event aggressors. To diminish discrepancies across situational contexts (especially familial vs. extrafamilial contexts), future investigations should partial out for number of events as well as for length of abuse period in the case of multiple-event aggressors. Finally, we had reservations about using only the current sentence. Doing so, it did not allow us to differentiate “serial” from “spree” sex offenders. Moreover, stability levels may have been inflated by not looking at the strategies and the victim-type polymorphia associated with previous sentences. It can be hypothesized that sexual aggressors change strategy and victim type especially after being apprehended. Victim-choice polymorphia among sex offenders has major implications for typological studies. Most modern typologies have brushed such inconveniences as sex offenders with more than one type of victim under the rug (e.g., rapists with prepubescent victims or extrafamilial child molesters with intrafamilial victims) (Knight and Prentky, 1990; Proulx et al., 1999a). Although this strategy has proven efficient at least for statistically distinguishing sex offender subtypes, considering fluctuations in specificity might help improve predictability and expand coverage of the sex offender population. Finally, future studies should also investigate criminal career indexes, such as the variety of criminal behavior at large, in order to compare subjects with high and low levels of victim-choice polymorphia. It could be hypothesized that offenders with low victim-type polymorphia are characterized neither by low self-control nor by shortcomings in managing daily life events. ACKNOWLEDGMENTS The authors thank Ren´e Carbonneau, Carlo Morcelli, and Paul Di Biase for their substantive and stylistic editing. Research was funded by the Correctional Services of Canada (245-M2-343), the FCAR (99-ER-3029), and the CRSH (410-97-1273). REFERENCES Abel, G. G., Becker, J. V., Cunningham-Rathner, J., Mittelman, M., and Rouleau, J. L. (1988). Multiple paraphilic diagnoses among sex offenders. Bull. Am. Acad. Psychiatry Law 16: 153–168.

P1: GDX/LZX/GGN/GDP/LOV/GGT Archives of Sexual Behavior [asb]

532

P2: GFU PP181-340734

June 28, 2001

9:6

Style file version Nov. 19th, 1999

Guay, Proulx, Cusson, and Ouimet

Abel, G. G., Becker, J. V., Mittelman, M., Cunningham-Rathner, J., Rouleau, J. L., and Murphy, W. D. (1987). Self-reported sex crimes of nonincarcerated paraphiliacs. J. Interpers. Viol. 2: 3– 25. Barbaree, H. E., Seto, M. E., Serin, R. C., and Amos, N. L. (1994). Comparison between sexual and nonsexual rapist subtypes: Sexual arousal to rape, offense precursors, and offense characteristics. Crim. Justice Behav. 21: 95–114. Blumstein, A., Cohen, J., Das, S., and Moitra, S. D. (1988a). Specialization and seriousness during adult criminal careers. J. Quant. Criminol. 4: 303–345. Blumstein, A., Cohen, J., and Farrington, D. P. (1988b). Longitudinal and criminal career research: Further clarifications. Criminology 26: 57–74. Blumstein, A., Cohen, J., and Farrington, D. P. (1988c). Criminal career research: Its value for criminology. Criminology 26: 1–35. Boutin, S. (1999). La carri`ere criminelle des agresseurs sexuels, Master dissertation, University of Montr´eal, Montr´eal, Qu´ebec, Canada. Bradford, J. M. W., Boulet, J., and Pawlak, A. (1992) The paraphilias: A multiplicity of deviant behaviours. Can. J. Psychiatry 37: 104–108. Cohen, M., Seghorn, T., and Calmas, W. (1969). Sociometric study of the sex offenders. J. Abnorm. Psychol. 74: 249–255. Farrington, D. P. (1992). Criminal career research in the United Kingdom. Brit. J. Criminol. 32: 521– 536. Farrington, D. P., Snyder, H. S., and Finnegan, T. A. (1988). Specialization in juvenile court careers. Criminology 26: 461–487. Freund, K., McKnight, K., Langevin, R., and Cibiri, S. (1972). The female child as surrogate object. Arch. Sex. Behav. 2:119–133. Gebhard, P. H., Gagnon, J. H., Pomeroy, W. B., and Christenson, V. V. (1965). Sex Offenders: An Analysis of Types, Harper & Row, New York. Gottfredson, M. R., and Hirschi, T. (1990). A General Theory of Crime, Stanford University Press, Stanford. Greenberg, D. F. (1991). Modeling criminal careers. Criminology 29: 17–46. Greenberg, D. F. (1992). Comparing criminal career models. Criminology 30:141–147. Groth, N. A., Hobson, W. F., and Gary, T. S. (1982). The child molester: Clinical observations. J. Soc. Work Hum. Sexuality 1: 129–144. Hall, G. C. N., and Proctor, W. C. (1987). Criminological predictors of recidivism in a sexual offender population. J. Consult. Clin. Psychol. 55: 111–112. Hanson, K. R., Scott, H., and Steffy, R. A. (1995). A comparison of child molesters and nonsexual criminals: Risk predictors and long-term recidivism. J. Res. Crime Delinq. 32: 325–337. Klein, M. W. (1984). Offence specialisation and versatility among juveniles. Brit. J. Criminol. 24: 185–194. Knight R. A., and Prentky, R. A. (1990). Classifying sexual offenders: The development and corroboration of taxonomic models. In Marshall, W. L., Laws, D. R., and Barbaree, H. E. (eds.), Handbook of Sexual Assault: Issues, Theories, and Treatment, Plenum, New York, pp. 23–52. Knight, R. A., Rosenberg, R., and Schneider, B. (1985). Classification of sexual offenders: Perspectives, methods and validation. In Burgess, A. W. (ed.), Rape and Sexual Assault: A Research Handbook, Garland, New York, pp. 222–293. Marshall, W. L., Barbaree, H. E., and Eccles, A. (1991). Early onset and deviant sexuality in child molesters. J. Interpers. Viol. 6: 325–335. ` Granger, L., McKibben, A., and Guay, J.-P. (1997). G´enitalisation et troubles de la Poirier, M.-E., personnalit´e chez les agresseurs sexuels d’enfants. Paper presented at the meeting of the Mental Health and Intervention Symposium, Montr´eal, Qu´ebec, Canada, Oct 1997. Proulx, J., and Ouimet, M. (1995). Criminologie de l’acte et p´edophilie. Rev. Int. Criminol. Police Teck. Sci. 3: 294–310. Proulx, J., Perreault, C., and Ouimet, M. (1999a). Pathways in the offending process of extrafamilial sexual child molesters. Sex. Abuse 11: 117–129. Proulx, J., St-Yves, M., Guay, J.-P., and Ouimet, M. (1999b). Les agresseurs sexuels de femmes: Sc´enarios d´elictuels et troubles de la personnalit´e. In Proulx, J., Cusson, M., and Ouimet, M. (eds.), Les Violences Criminelles, Presses de l’Universit´e Laval, Qu´ebec, pp. 157–185.

P1: GDX/LZX/GGN/GDP/LOV/GGT Archives of Sexual Behavior [asb]

P2: GFU PP181-340734

Victim-Choice Polymorphia

June 28, 2001

9:6

Style file version Nov. 19th, 1999

533

Proulx, J., Perreault, C., Ouimet, M., and Guay, J.-P. (1999c). Les agresseurs sexuels d’enfants: Sc´enarios d´elictuels et troubles de la personnalit´e. In Proulx, J., Cusson, M., and Ouimet, M. (eds.), Les violences criminelles, Presses de l’Universit´e Laval, Qu´ebec, pp. 187–216. Soothill, K., Francis, B., Sanderson, B., and Ackerley, E. (2000). Sex offenders: Specialists, generalists—or both? Brit. J. Criminol. 40: 56–67. Studer, L. H., Clelland, S. R., Aylwin, A. S., Reddon, J. R., and Monro, A. (2000). Rethinking risk assessment for incest offenders. Int. J. Law Psychiatry 23: 112–124.

P1: GVG/GCR/FJQ/GMF

P2: GDX/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340735

June 21, 2001

15:56

Style file version Nov. 19th, 1999

Archives of Sexual Behavior, Vol. 30, No. 5, 2001

Comparative Data of Childhood and Adolescence Molestation in Heterosexual and Homosexual Persons Marie E. Tomeo, PhD,1 Donald I. Templer, PhD,1,2 Susan Anderson, PhD,1 and Debra Kotler, PhD1

In research with 942 nonclinical adult participants, gay men and lesbian women reported a significantly higher rate of childhood molestation than did heterosexual men and women. Forty-six percent of the homosexual men in contrast to 7% of the heterosexual men reported homosexual molestation. Twenty-two percent of lesbian women in contrast to 1% of heterosexual women reported homosexual molestation. This research is apparently the first survey that has reported substantial homosexual molestation of girls. Suggestions for future research were offered. KEY WORDS: childhood molestation; adult sexual orientation; female pedophiles.

INTRODUCTION The purpose of this study was to determine and compare the incidence of childhood molestation of homosexual and heterosexual adults. The previous literature suggests that a history of childhood molestation may be more common in homosexual men and women than in their heterosexual counterparts. Bradford et al. (1994) reported that 25% of 1,925 lesbians surveyed had been sexually abused or raped as children. They said that 31% of the perpetrators were male relatives, 45% other known men, 45% male strangers, and 1% women (it is not known why this totaled 122%). In interviewing with 35 lesbian women in alcoholic recovery, 16 (46%) revealed sexual abuse, even though they were not asked about sexual abuse (Hall, 1996). The onset of the molestation ranged from <4 years to 12 years, with 10 of 1 California

School of Professional Psychology, Fresno, California.

2 To whom correspondence should be addressed at California School of Professional Psychology, 5130

East Clinton Way, Fresno, California 93727. 535 C 2001 Plenum Publishing Corporation 0004-0002/01/1000-0535$19.50/0 °

P1: GVG/GCR/FJQ/GMF

P2: GDX/GFU/LOV

Archives of Sexual Behavior [asb]

536

PP181-340735

QC: June 21, 2001

15:56

Style file version Nov. 19th, 1999

Tomeo, Templer, Anderson, and Kotler

the women being younger than 7 years at the time of molestation. Fourteen of the 16 women were molested only by men. One was molested by multiple intrafamilial males and one extrafamilial female. Another was molested by her father and by one extrafamilial female. Simari and Baskin (1982) determined the extent of incestuous experiences in 54 gay men and 29 gay women. Twenty-five (46%) of the gay men reported experiencing incest, all of the homosexual sort, between 9 and 16 years of age with a mean age of 13 years. Twenty-four (96%) of the 25 men who experienced incest viewed themselves as homosexual before the incest. Eleven (38%) of the 29 gay women reported experiencing incest, heterosexual in seven (64%) of the cases, with two (29%) of these women regarding themselves as homosexual before the incest. Four (36%) of the 11 women experienced homosexual incest, with 3 (25%) of them regarding themselves as homosexual before the incest. In research by Doll et al. (1992), 1,001 homosexual and bisexual men in sexually transmitted disease clinics were interviewed. Thirty-seven percent of these men reported having been forced or encouraged to have sex by an older and more powerful person, before the age of 19. The median age of these events was 10 years. The median age difference between the younger and older participants in the sex acts was 11 years. In 94% of the cases, the older person was male, in 5% female, and 1% involved both male and female perpetrators. Cameron and Cameron (1995) surveyed 5,182 adults and found a history of incest in 22.3% of homosexual men, 1.2% of heterosexual men, 3.4% of homosexual women, and 0.3% of heterosexual women. The authors did not provide any information about the age of molestation. Cameron and Cameron pointed out that their findings are consistent with the findings of Bell et al. (1981) who reported a history of sibling incest in 15.6% of homosexual men, 9.0% of heterosexual men, 10.4% of homosexual women, and 5.6% of homosexual women. Bell et al. did not specify the gender of the other partner in the incestuous acts. Gundlach (1977) surveyed 225 lesbian and 233 heterosexual women, and found that 30% of the heterosexual women and 21% of the lesbians had been raped. Of the 30 women who had been raped before the age of 14, 26 had an adult homosexual orientation while 9 had an adult heterosexual orientation. On the basis of the previous literature, it would appear that gay men and lesbian women have a greater history of molestation than do heterosexual persons. The perspective from the previous studies, however, is unclear. Not all of the cited studies employed a heterosexual control group, specified the criteria of molestation, specified the gender of the perpetrator, specified the present sexual preference, and that before the molestation, and employed both gay men and lesbian participants. Furthermore, some of the studies focused only upon rape or incest. This study intended to extend the perspective by the inclusion of gay men and lesbian women and heterosexual control men and women, by using specific

P1: GVG/GCR/FJQ/GMF

P2: GDX/GFU/LOV

Archives of Sexual Behavior [asb]

PP181-340735

QC: June 21, 2001

15:56

Comparative Data of Childhood and Adolescence Molestation

Style file version Nov. 19th, 1999

537

criteria of molestation, by requesting the gender and age of the perpetrator, and by the determination of both current sexual preference and that before the molestation.

METHOD Although questionnaires were obtained from 981 persons, the data was not used for 39 of them because they did not include gender or sexual orientation or both. The persons regarded as participants consisted of 942 nonclinical population persons, 675 of whom were graduate or undergraduate students currently enrolled and attending classes in colleges and universities in central California. Instructor permission and applicable campus regulations regarding human subjects requirements were sought prior to distributing questionnaires in the classroom. The vast majority of the students who received questionnaires completed and returned them, but the exact percentage is not known. The age range of the college students was 18–68 with a mean of 27.65 and a standard deviation of 10.44. The total years of education of the college students ranged from 12 to 15 with a mean of 14.77 and a standard deviation of 3.59. Two hundred sixty-seven additional participants with fully completed questionnaires were obtained at the homosexual pride events in the central California area. The principal investigator rented a booth at the event and asked individuals who approached the booth to participate in the research. All of the participants who accepted questionnaires completed and returned them. The age range of the homosexual pride attendees was 18–67 with a mean of 36.60 and a standard deviation of 10.43. The total years of education of the homosexual pride attendees was 8–26 with a mean of 15 and a standard deviation of 2.65. The total years of education with all participants combined was 8–26 with a mean of 14.84 and a standard deviation of 3.34. The 329 males had an age range of 18–68 with a mean of 29.85 and a standard deviation of 11.06. The total years of education for the males was 8–24 with a mean of 14.78 and a standard deviation of 11.33. The 613 females had an age range from 18 to 62 years with a mean of 30.36 and a standard deviation of 11.33. The total years of education for the females was 11–25 with a mean of 14.84 and a standard deviation of 3.33. The questionnaire contained in Table I was administered to all participants. The principal investigator contacted instructors at several California educational institutions to obtain permission to disperse questionnaires to students in their classes. There were no identifying marks on any of the questionnaires and the participants were informed of their rights to deny participation and be given confidentiality. Graduate students attending the California School of Professional Psychology, Fresno, were contacted through their individual student mailboxes and were given the same questionnaires and confidentiality.

P1: GVG/GCR/FJQ/GMF

P2: GDX/GFU/LOV

Archives of Sexual Behavior [asb]

PP181-340735

QC: June 21, 2001

538

15:56

Style file version Nov. 19th, 1999

Tomeo, Templer, Anderson, and Kotler Table I. Questionnaire Administered to All Participants Sexuality Questionnaire

Age

Male

Female

Ethnicity:

Caucasian African American Hispanic Asian American Pacific Islander Native American Other

Years of Education Completed

1. Do you regard yourself as predominantly a heterosexual person or predominantly a gay/lesbian person 2. If predominantly of heterosexual orientation, at what age did you begin to regard yourself as heterosexual? 3. If predominantly of homosexual orientation, at what age did you begin to regard yourself as homosexual? 4. Before you were 16 years old, did you ever have sexual contact with a woman or girl 5 or NO more years older than yourself and at least 16 years of age? YES 5. IF YES, at what age did this first occur? 6. What was your relationship to the person with whom this sexual contact occurred? 7. Before you were 16 years old, did you ever have sexual contact with a man or boy 5 or more years older than yourself and at least 16 years of age? YES NO 8. IF YES, at what age did this first occur? 9. What was your relationship to the person with whom this sexual contact occurred? 10. Have you ever had sex with a person of the same sex after you reached your 18th birthday? NO YES 11. Have you ever had sex with a person of the opposite sex after you reached your 18th NO birthday? YES

The criteria of molestation were identical to that used in the studies by Petrovich and Templer (1984) and Condy et al. (1987), on molestation of boys by women. These criteria require that the victim be under the age of 16 at the time of the molestation, the perpetrator being at least 16 years of age and at least 5 years older than the victim. Table II displays the number and percentage of molested homosexual participants who reported themselves as homosexual before the molestation.

Table II.

Number and Percentage of Homosexual Participants Self-Identified as Homosexual Before Molestation Before molestation

Before molestation by male

Before molestation by female

All homosexual participants

N

%

N

%

N

%

Homosexual men (N = 124) Homosexual women (N = 153)

84 95

68 62

81 110

65 72

102 57

82 37

P1: GVG/GCR/FJQ/GMF

P2: GDX/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340735

June 21, 2001

15:56

Style file version Nov. 19th, 1999

Comparative Data of Childhood and Adolescence Molestation

539

Table III. Number and Percentage of Molested Participants as a Function of Sexual Preference Molested Participants Male participants∗∗∗ All heterosexual men (N = 205) All homosexual men (N = 124) Fair homosexual men (N = 121) Female participants∗∗∗ All heterosexual women (N = 460) All homosexual women (N = 153) Fair homosexual women (N = 146)

N

Molested by men χ2

%

N

%

χ2

6.7

Molested by women N

%

50 24.4

12

61 49.2 20.17∗∗∗

56 45.5 61.19∗∗∗ 20 16.4

61 50.4

56 46.7

113 24.6

111 24.3

65 42.5 17.03∗∗∗

44 29.3

63 43.2

42 29.4

χ2

41 20.0 .44

20 16.8 5 1.23

1.1

33 21.6

75.47∗∗∗

33 22.6

Note. All chi squares calculated for percentage differences between all heterosexual participants and all homosexual participants of the same sex. ∗∗∗ p < .001.

RESULTS Table III displays the number and percentage of total male, total female, and categories of male and female participants who had been molested, molested by men, and molested by women. It is also to be noted that the percentages for gay pride men are similar to the percentages for all homosexual men because 97.6% of the homosexual men were gay pride participants. All six chi squares compare all homosexual subjects to all heterosexual subjects of the same sex. It is apparent that the homosexual men and women were more likely to have been molested and more likely to have been molested by a person of the same sex than did their heterosexual counterparts. DISCUSSION One of the more salient findings of this study is that 46% of homosexual men and 22% of homosexual women reported having been molested by a person of the same gender. This contrasts to only 7% of heterosexual men and 1% of heterosexual women reporting having been molested by a person of the same gender. Although the differences between the molestation rate of homosexual and heterosexual participants have to be viewed with great caution because of the differences in the sampling methods for each population, the absolute magnitude of same-sex molestation in homosexual subjects is worthy of attention. The rates of same-sex molestation in homosexual subjects were 45% for homosexual men

P1: GVG/GCR/FJQ/GMF

P2: GDX/GFU/LOV

Archives of Sexual Behavior [asb]

540

PP181-340735

QC: June 21, 2001

15:56

Style file version Nov. 19th, 1999

Tomeo, Templer, Anderson, and Kotler

and 22% for homosexual women. If the college students are excluded, the rates are 46 and 23%, respectively. Perhaps the most salient finding is that 22% of the homosexual women reported having been molested by a woman. It has been well recognized for decades that it is very common for men to molest girls and boys. Molestations by women, however, had been considered rare. Petrovich and Templer (1984) serendipitously found that 59% of male penitentiary inmates had been molested by a woman. Condy et al. (1987) extended the findings of Petrovich and Templer, using the same criterion of molestation that they used and that was used in this study, which requires the victims being under age 16 and the perpetrator being at least 16 years of age and at least 5 years older than the victim. Condy et al. found that 46% of male inmate rapists, 37% of child molesters, 47% of non-sex-offender inmates, and 16% of male college students had been molested by one or more women. We now know that women molesting girls is probably not rare. This study appears to be the first survey-based research that reported girls having been molested by other females. There have been, however, other studies that report case study or anecdotal female molestation of girls (Dunbar, 1995; Hislop, 1995; Hunter et al., 1993). In this study, we do not know the adult sexual preference of the women who molested. Most men who molest boys and most men who molest girls are heterosexual in their adult sexual preference (Groth and Birnbaum, 1978). In fact, we do not know to what extent these women are attracted to adults of either gender and to what extent their orientation is pedophilic. Nevertheless, since the female victims in the current study had a mean age of 13 at the time of abuse, and 68% of them were at least 12 years of age, we are dealing more with adolescent molestation than with child molestation. In one study of men who molest, the mean age of the children was 10.7 years (Jaffe et al., 1975). Petrovich and Templer (1984) found a mean age of 10.81 in boys who were molested by women. The present study reported a mean age of 11 for boys molested by men, and a mean age of 13 for girls molested by women. It would appear that men molest children at a younger age than women do. This suggests that molestation by women may have less of a pedophilic component than in men. We know that some male pedophiles have always had a strong pedophile preference and very weak sexual attraction to adults. Groth and Birnbaum (1978) called them “fixated” pedophiles in contrast to the “regressed” pedophiles whose fundamental preference, at least in the past, was predominantly for adults. Whether or not there are women whose primary or sole sexual preference is for children remains to be determined. More research is needed on both the gender and the age aspects of sexual preference of women who molest. Sixty-eight percent of the present homosexual male participants and 38% of the present homosexual female participants (68 and 36%, respectively, if including just the homosexual fair participants) did not identify as homosexual until after the molestation. This suggests that if molestation resulted in homosexuality,

P1: GVG/GCR/FJQ/GMF

P2: GDX/GFU/LOV

Archives of Sexual Behavior [asb]

PP181-340735

QC: June 21, 2001

15:56

Comparative Data of Childhood and Adolescence Molestation

Style file version Nov. 19th, 1999

541

this phenomenon occurs in a greater proportion of male homosexuals. It may not, however, be a casual factor in either gender. Perhaps children or adolescents with a higher potential for homosexual behavior are more likely to enter a situation that leads to same-sex molestation. It must also be borne in mind that the present homosexual participants may not be representative of homosexual persons. The overwhelming preponderance of homosexual participants was in the gay pride group. There were only three homosexual men and seven homosexual women in the college group. On the other hand, the gay pride participants certainly seem to be a nonclinical group. It is most unlikely that all the present findings apply only to homosexual persons who go to homosexual fairs and volunteer to participate in questionnaire research.

REFERENCES Bell, A. P., Weinberg, M. S., and Hammersmith, S. K. (1981). Sexual Preference: Its Development in Men and Women, Indiana University Press, Bloomington. Bradford, J., Ryan, C., and Rothblum, E. D. (1994). National Lesbian Health Care Survey: Implications for mental health care. J. Consult. Clin. Psychol. 62: 228–242. Cameron, P., and Cameron, K. (1995). Does incest cause homosexuality? Psychol. Rep. 76: 611–621. Condy, S. R., Templer, D. I., Brown, R., and Veaco, L. (1987). Parameters of sexual contact of boys with women. Arch. Sex. Behav. 16: 379–394. Doll, L. S., Joy, D., Bartholow, B. N., Harrison, J. S., Bolan, G., Douglas, J. M., Saltzman, L. E., Moss, P. M., and Delgado, W. (1992). Self-reported childhood and adolescent sexual abuse among adult male homosexual and bisexual men. Child Abuse Negl. 16: 855–864. Dunbar, T. M. (1995). Women who sexually molest female children. Dissert. Abstr. Int., Sec. B: Sci. Engng. 56(3-B): 1687. Groth, A. N., and Birnbaum, H. J. (1978). Adult sexual orientation and attraction to underage persons. Arch. Sex. Behav. 7: 175–181. Gundlach, R. H. (1977). Sexual molestation and rape reported by homosexual and heterosexual women. J. Homosex. 2: 367–384. Hall, J. M. (1996). Pervasive effects of childhood sexual abuse in lesbians’ recovery from alcohol abuse. Subst. Use Misuse 31: 225–239. Hislop, J. R. C. (1995). Female child molesters. Dissert. Abstr. Int. 1995-95013-106. UMI AAM9514340. Hunter, J. A., Lexier, L. J., Goodwin, D. W., and Browne, P. A. (1993). Psychosexual, attitudinal, and developmental characteristics of juvenile female sexual perpetrators in a residential treatment setting. J. Child Fam. Studies 2: 317–326. Jaffe, A. C., Dynneson, L., and ten Bensel, R. W. (1975). Sexual abuse of children: An epidemiological study. Am. J. Disease Child. 129: 689–692. Petrovich, M., and Templer, D. I. (1984). Heterosexual molestation of children who later became rapists. Psychol. Rep. 54: 810. Simari, C. G., and Baskin, D. (1982). Incestuous experiences within homosexual populations: A preliminary study. Arch. Sex. Behav. 11: 329–344.

P1: GVM/GCX

P2: GDB/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340736

June 21, 2001

16:3

Style file version Nov. 19th, 1999

Archives of Sexual Behavior, Vol. 30, No. 5, 2001

(Serious) Sadomasochism: A Protected Right of Privacy? Richard Green, MD, JD1,2,3

Several adult male members of a sadomasochistic sex club were arrested for violating a U.K. statute passed in 1861 – The Offences Against the Person Act. The sex acts spanned a 10-year period and were videotaped. They involved manipulation of the genitalia with hot wax, sand paper, fish hooks, and needles. The sexual interactions were consensual. This analysis tracks the case from trial, to the Court of Appeal, to the House of Lords, and on to the European Court of Human Rights. It examines whether or not these sexual behaviors should be protected under a right to privacy. It contrasts state concerns of bodily harm, albeit consented to, in sex with those incurred in sport. KEY WORDS: sadomasochism; homosexuality; sexual privacy; privacy.

INTRODUCTION In 1987 in the course of investigations into other matters, English police came into possession of video films of sadomasochist encounters made over a 10-year period involving 50 men. Sex consisted mainly of maltreatment of the genitalia, sometimes with hot wax, sandpaper, fish hooks, and needles plus ritualistic beatings, either with bare hands or implements including stinging nettles, spiked belts, and a cat-o’-nine tails. There was branding and infliction of injuries which bled. Activities were consensual. Administration of pain was subject to rules, including the provision of a code word to be used by any recipient to stop an act. Acts did not lead to infection or permanent injury. Video cameras recorded events and the tapes were copied and distributed among members of the group. 1 Institute

of Criminology, University of Cambridge. of Law, University of Cambridge. of Psychiatry, Imperial College School of Medicine, Charing Cross Hospital, Fulham Palace Road, London W6 8RF.

2 Faculty

3 Department

543 C 2001 Plenum Publishing Corporation 0004-0002/01/1000-0543$19.50/0 °

P1: GVM/GCX

P2: GDB/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340736

June 21, 2001

16:3

544

Style file version Nov. 19th, 1999

Green

In 1990 several participants were prosecuted and convicted at London’s Old Bailey of violating the Offences Against the Person Act 1861. That Act states, “whoever shall unlawfully and maliciously wound or inflict any grievous bodily harm on any other person, either with or without any weapon or instrument, shall be liable to imprisonment.” The case became popularly known as “The Spanner Case” reportedly because a police detective remarked after watching one video that he felt as though “someone had tightened a spanner (wrench) around his nuts.” This analysis focuses on whether or not these sexual behaviors should be protected, either under a general umbrella of a right to sexual privacy in the United Kingdom (as in the 1957 Wolfenden Report’s well-known pronouncement regarding homosexual behavior, because this is simply “not the law’s business”) or by extension of other rights protecting capacity to consent to actual bodily harm, as in professional boxing. PRIOR UK CASE LAW In 1882, prize fighting was held to be an unlawful assault because it led to a breach of peace from the gathering of a rowdy crowd (R v. Coney, 1882). In 1934, a man liaised in London with a 17-year-old female after previous telephone conversations had made it clear that he wanted to cane her for his sexual gratification. They then went to his garage for the caning. According to the physician who examined her 2 days later, she had marks indicative of a “fairly severe beating.” At trial, she denied having consented to the caning. Conviction followed. At appeal, Justice Swift wrote, “it is an unlawful act to beat another person with such a degree of violence that the infliction of actual bodily harm is a probable consequence and when such an act is proved, consent is immaterial.” However, his conviction was reversed because of errors by the trial judge (R v. Donovan, 1934). In 1980, two men quarrelled and decided to settle the disagreement with a fight. One suffered a bleeding nose and bruised face. Upholding conviction, the Court of Appeal held “it is not in the public interest that people should try to cause or should cause each other actual bodily harm for no good reason. It is immaterial whether the act occurs in private or in public. It is an assault if actual bodily harm is intended and/or caused” (Attorney-General’s Reference, 1980). In 1992, a military man used nipple clamps and nipple piercing during bondage with his female partner. At trial, his partner testified that she had consented. The judge directed the jury to acquit (Thompson, 1994, p. 242). In a more recent post-Spanner case, in 1996 a husband branded his initials with a knife on his wife’s buttocks with her consent. He was convicted but on appeal his conviction was overturned with the Court noting that the wife not only consented to the act, but she instigated it, and that there was no aggressive intent on his part. To

P1: GVM/GCX

P2: GDB/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340736

June 21, 2001

16:3

Style file version Nov. 19th, 1999

(Serious) Sadomasochism: A Protected Right of Privacy?

545

the Court, the branding was “not any more dangerous than tattooing.” Consensual activity between husband and wife in the privacy of a matrimonial home was declared not a proper matter for criminal investigation (R v. Wilson, 1996) THE “SPANNER” OR (PROPERLY) BROWN CASE At trial, the men pleaded guilty to assault charges after the judge ruled they could not rely on consent of the participants as an answer to the prosecution case. On 19 December 1990, they were convicted and sentenced to terms of imprisonment of between 2 and 4 years. The trial judge disclaimed a biasing influence from the sexual orientation of the members of this sadomasochistic (SM) club, declaring “the unlawful conduct now before the Court would be dealt with equally in the prosecution of heterosexuals or bisexuals if carried out by them.” On 19 February 1992, the Court of Appeal, Criminal Division, dismissed the appeal, but certified a further appeal to the House of Lords, in that a point of law of general public importance was involved. It asked “where A wounds or assaults B, causing him actual bodily harm in the course of sado-masochistic encounter, does the Prosecution have to prove lack of consent on the part of B before they can establish A’s guilt (under the 1861 Act)?” The Law Lords majority acknowledged that some other consensual activities that involved actual bodily harm are lawful, for example, ritual circumcision, tattooing, ear-piercing, and sports like boxing. However, they observed that duelling is unlawful and consent by the protagonists affords no defence against prosecution. They also observed that although suicide was no longer an offence, a person who assisted another to commit suicide was guilty of homicide. To the majority, the evidence disclosed that the practices of the appellants were unpredictably dangerous and degrading to body and mind, were developed with increasing barbarity, and were taught to persons whose consents were dubious or worthless; (this because) drink and drugs were employed to obtain consent and increase enthusiasm. The Lords pointed out that “it was not surprising that a victim (the Court’s designation of the participant) did not complain to the police when the complaint would involve giving details of acts in which he had participated.” With regard to the “absence” of medical treatment, they pointed to doctors being subject to a code of confidentiality (thus insulating medical treatment for injuries from public scrutiny). Lord Jauncey offered to Parliament the option (to date unaccepted) of legalising these SM acts with this graphic suggestion: if it was to be decided that such activities as the nailing by A of B’s foreskin or scrotum to a board or the insertion of hot wax into C’s urethra followed by the burning of his penis with a candle, or the incising of D’s scrotum with a scalpel to the effusion of blood were injurious to neither B, C and D, nor to the public interest, it was for Parliament with its accumulated wisdom . . . to declare them lawful.

P1: GVM/GCX

P2: GDB/GFU/LOV

Archives of Sexual Behavior [asb]

546

QC:

PP181-340736

June 21, 2001

16:3

Style file version Nov. 19th, 1999

Green

In dissent, Lord Slynn argued that there was no doubt there was consent here, “. . .indeed there was more than mere consent. Astonishing though it might seem,” he wrote, “the persons involved positively wanted and asked for the acts to be done to them. [Further] the accused were old enough to know what they were doing, [and] the acts were done in private.” In dissent, Lord Mustill held up boxing as an example of one person seriously injuring another with legal impunity. Each boxer aims to end the contest prematurely by infliction of brain injury serious enough to make the opponent unconscious or temporarily by impairing his central nervous system through a blow to the midriff, or cutting his skin to a degree which would ordinarily be well within the scope of the Act of 1861.

Lord Mustill argued against the appropriateness of prosecution under the 1861 Act, pointing to its title, “Offences Against The Person.” There was no animosity, aggression, or personal rancour from the assailant and no protest by the victim. On 11 March 1993, the appeal was dismissed 3–2. Following this narrow defeat in the House of Lords, the case was appealed to the European Commission of Human Rights. Although a majority there rejected the argument of the appellants 11–7, the case was referred by the Commission to the European Court of Human Rights on 11 December 1995. The Commission’s remit to the Court was to determine if the United Kingdom was in violation of Article 8 of the European Convention: “Everyone has the right to respect for his private and family life. . . . There shall be no interference . . . except . . . (as) is necessary in a democratic society in the interests of . . .protection of health or morals. . . . ” The men submitted that their case should be viewed as one involved in matters of sexual expression, rather than violence, and the line beyond which consent is no defence for physical injury should only be drawn at the level of intentional or reckless causing of serious disabling injury. The U.K. government conceded that the men’s prosecution was an interference by a public authority in their private life but argued that the interference pursued the legitimate aim of protection of health. The government argued that the State was entitled to punish acts of violence that could not be considered of a trifling or transient nature, irrespective of consent. In fact, in the present case, some of these acts could well be compared to “genital torture” and the Contracting-State could not be said to have an obligation to tolerate acts of torture because they are committed in the context of a consenting sexual relationship. The government further contended that the criminal law should seek to deter certain forms of behavior not only on public health grounds but also for broader moral reasons. In this respect, acts of torture may be banned on the ground that they undermine the respect human beings should confer upon each other.

P1: GVM/GCX

P2: GDB/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340736

June 21, 2001

(Serious) Sadomasochism: A Protected Right of Privacy?

16:3

Style file version Nov. 19th, 1999

547

The European Court, reviewing the judgment of the House of Lords, concluded that the opinions of the majority were based on the extreme nature of the practices and not the sexual orientation of the participants. The Court accepted that sexual orientation and activity concern an intimate aspect of private life. However, a considerable number of people were involved in the activities in question, which included, inter alia, recruitment of new “members,” the provision of several specially equipped “chambers,” and the shooting of many video tapes that were distributed among the members. It was thus an open question to the Court if these sexual activities fell entirely within the notion of private life. However, since the privacy point had not been disputed by the State or the appellants, the Court saw no reason to examine it. Assuming therefore that the prosecution and conviction of the applicants did amount to interference with their private life, the question was if the interference was “necessary in a democratic society.” The European Court did not find it necessary to determine if the interference with the applicants’ right to respect for private life would also be justified for protection of morals. But this finding, the Court pointed out, should not be understood as calling into question the prerogative of the State to deter acts of this kind on moral grounds. There was no evidence to support the men’s assertion, according to the Court, that the SM activities were essential to their happiness but the argument would be accepted if sadomasochism were only concerned with sex (and not violence) as the men contended. The Court considered that one of the roles the State is entitled to undertake is to regulate activities that involve the infliction of physical harm. This is so whether the activities occur in the course of sexual conduct or otherwise. Nor did the Court accept the argument that no prosecution should have been brought since injuries were not severe and no medical treatment had been required. In deciding whether or not to prosecute, State authorities were entitled to have regard not only for the actual seriousness of the harm but also for the potential for harm inherent in the acts in question. The European Court rejected the appeal 9–0. DISCUSSION John Stuart Mill set the standard for the contemporary libertarian argument for privacy in 1859. Society only has jurisdiction over a person’s conduct when it prejudicially affects the interests of others. No one has the right to say to another adult that “he shall not do with his life for his own benefit what he chooses to do with it. He is the person most interested in his own well-being” (Mill, 1859). Mill’s contemporary critic, Sir James Stephen, objected: “How can the State or the public be competent to determine any question whatever if it is not competent to decide that gross vice is a bad thing.” To enforce moral behavior, “society has at its disposal two great instruments . . . law and public opinion” (Stephen, 1873).

P1: GVM/GCX

P2: GDB/GFU/LOV

Archives of Sexual Behavior [asb]

548

QC:

PP181-340736

June 21, 2001

16:3

Style file version Nov. 19th, 1999

Green

A century later (1959) Mill’s philosophy was attacked by Lord Devlin. Society has the right to legislate against immorality, there being no area of private immorality that is not the law’s business. But law makers are not to ascertain the moral judgments of society by majority vote. It is the judgment of the “man in the street.” It is not nearly enough to say that a majority dislike a practice. The threshold is reached when the behavior invokes intolerance, indignation, and disgust (Devlin, 1965). Hart (1961) countered that the risk of democratic rule by the majority lay in telling the man in the street that if “he feels sick enough about what other persons do in private to demand its suppression by law, no theoretical criticism can be made of his demand.” Paradoxically it was Mill’s critic, Stephen, whose legal reasoning might have sustained the Brown (Laskey, Jaggard, and Brown v. the United Kingdom, 1995/1997) appeal. In his Digest of the Criminal Law (Stephen, 1883) and in a court decision (R v. Coney, 1882), he stated that consent was a defence to a charge of occasioning actual bodily harm and “ . . . everyone has a right to consent to the infliction upon himself of bodily harm not amounting to a maim.” This high threshold for legally tolerable acts was later disapproved in R v. Donovan (1934). Did these acts constitute “private life”? Was the SM club private? A definition of a private club is that it is open to members only, not the general public. These questions were rendered moot as the government conceded the intrusion on private life. Perhaps the Court would have ruled against this threshold issue. But the decision makes it irrelevant whether or not the acts constituted sexual privacy as the State was granted authority to intervene, nevertheless. Privacy and sexual orientation are intertwined under U.K. law. Until overturned on privacy grounds by the European Court of Human Rights in 2000 (ADT v. United Kingdom, 2000) sexual behavior involving more than two men at a time was illegal. This prohibition was included 33 years ago when male homosexual acts between consenting 21-year-olds were first permitted (Sexual Offences Act 1967). It was to prevent gay orgies. But sex between a male and more than one female was permitted, as was group sex by lesbians. Was the Spanner ruling heterosexist? Why was a husband branding his initials on his wife’s buttocks, with her consent, distinguished from the Brown case? Branding was one of the acts of the SM club. How many additional acts would this couple have had to engage in, anatomical constraints notwithstanding, before they fell under the Act of 1861? Although the Lords disclaimed prejudice based on the sexual orientation of the participants, and were unfaulted in this regard by the European Court, Lord Jauncey explained, “the House of Lords had to consider the possibility that homosexual sado-masochistic activities were practised by others who might not be so controlled and responsible as the appellants were claimed to be.” In another reference, Lord Jauncey “had no doubt that it would not be in the public interest

P1: GVM/GCX

P2: GDB/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340736

June 21, 2001

(Serious) Sadomasochism: A Protected Right of Privacy?

16:3

Style file version Nov. 19th, 1999

549

that deliberate infliction of actual bodily harm during the course of homosexual sado-masochistic activity should be held to be lawful.” Lord Lowry commented that “sado-masochistic homosexual activities could not be regarded as conducive to the enhancement or enjoyment of family life” (reference to a European Convention right for protection) and “A relaxation of the prohibitions . . . could only encourage the practice of homosexual sado-masochism’. (emphases added). Finally, Lord Jauncey, referring to an under-21-year-old participant in the SM activities, exuded relief in announcing, “It is some comfort at least to be told, as we were, that (he) is now it seems settled into a normal heterosexual relationship.” The Court held that the State can regulate infliction of physical harm “albeit consenting” and that the determination of a tolerable level of harm via consent is for the discretion of the State. However, this grants to the State an essentially unworkable authority. It provides no guidelines for regulating SM activities that in the seminal words of the U.S. Food and Drug Administration are “both safe and effective.” Other skin-penetrating—arguably masochistic or sadistic acts—are permitted. These include nipple rings, clitoral rings, and tongue-studs. Ritual circumcision of Jewish males is not consented to but permitted. Were the acts genuinely consented to? The House of Lords deemed the participants’ consent to the sexual acts dubious or worthless because drink and drugs were employed to obtain consent and increase enthusiasm. But this disqualification could criminalize much adult sexual experience. The Court was concerned with the smearing of faeces during some of the sex acts and the risk of infection. Yet for over three decades anal intercourse between two men had been legal (the prime source of HIV contagion) (Sexual Offences Act 1967). Some risk of infection does exist in these acts, with a potentially fatal disease such as hepatitis or HIV, or from a sadist overestimating tolerance by a masochist, and short-circuiting the safe switch-off word. Whereas governmental control of public risk is common, such as requiring motorcycle helmets, automobile seatbelts, and vaccinations, with SM activities, State regulation is less easily effected. These are not public acts. But licensing and inspection could be attempted. The Spanner Case triggered a Law Commission Consultation Paper in 1995. It recommended a change in the law under which these men were convicted. The Commission provisionally proposed that either intentional or reckless causing of injury, other than that which is seriously disabling, even if amounting to “grievous bodily harm” within the meaning of the 1861 Act, not be criminal, if the other person consented. In place of criminal prosecution, the Commission recommended licensing arrangements aimed at ensuring safe and hygienic practices. No action has been taken since these proposals were submitted. A slippery legal slope could result from this successful prosecution. There is a substantial UK prostitution economy based on dominance and discipline as anyone who has ever been in a London phone kiosk reading the card adverts knows. The

P1: GVM/GCX

P2: GDB/GFU/LOV

Archives of Sexual Behavior [asb]

QC:

PP181-340736

June 21, 2001

16:3

550

Style file version Nov. 19th, 1999

Green

Brown decision could be extended to prosecuting SM prostitution (prostitution is generally legal) and to prosecution of SM films such that State control bypasses the more complex legal definition of obscenity and invokes a statutory violation of the Act of 1861. What is a “good reason” for people to try cause each other actual bodily harm? In the legal precedent, Attorney-General’s Reference (1980), settling an argument by fist fight was deemed not a “good reason.” Why is experiencing sexual arousal not a “good reason”? Why is sport a “good reason”? In boxing most injuries are consequent to head trauma. Muhammad Ali can barely speak and he was usually the winner. Yet professional boxing does not require the wearing of head guards. There is no paternalistic response to sport, so why to sex? The legal precedent of barring fist fighting (Coney) primarily to prevent breach of peace by unruly spectators fails to support the legality of professional English football. There, hordes of police are required at each match in an effort (sometimes unsuccessful) to prevent deadly violence. In treating sex and sport on a legal par neither would be permitted in consequence of health risks. Or, both would be permitted with consent outweighing the State interest in harm avoidance. REFERENCES ADT v United Kingdom (Application No 35765/97). European Court of Human Rights. Attorney-General’s Reference (No. 6 of 1980) (1981) 2 All ER 1057 (1981). Queen’s Bench 715 (1981) 3WLR. Laskey, Jaggard, and Brown v the United Kingdom (109/1995/615/703-705). European Court of Human Rights, 19 February 1997. Devlin, P. (1965). Morals and the criminal law. In The Enforcement of Morals, Oxford University Press, Oxford. Hart, H. L. A. (1961). Immorality and treason. In Blom-Cooper, L. (ed.), The Law as Literature, The Bodley Head, London. Law Commission Consultation Paper No. 139 (1995). Mill, J. S. (1859/1978). On Liberty. Rapaport, E. (ed.), Indianapolis, Hackett. R v. Brown and other appeals, 2 All England Law Reports 75 (1993). R v. Coney, 51LJM C66, 8QBD 534, CCR (1882). R v. Donovan, 2 King’s Bench Reports, 498, All ER Rep. 207, 25. Cr App R1, CCA (1934). R v. Wilson, 3 Weekly Law Reports, 125 (1996). (Wolfenden) Report of the Committee on Homosexual Offences and Prostitution (1957) (Cmd 247). Sexual Offences Act (1967). Stephen, J. (1873). Liberty, Equality, Fraternity. pp. 125, 148. London, Smith Elder, Stephen, J. (1883). Digest of the Criminal Law. Thompson, B. (1994). Sadomasochism: Painful Perversion or Pleasurable Play? Cassell, London.

P1: VENDOR/GCO Archives of Sexual Behavior [asb]

PP181-340737-41

June 28, 2001

9:8

Style file version Nov. 19th, 1999

Archives of Sexual Behavior, Vol. 30, No. 5, 2001

BOOK REVIEWS Sex and Cognition. Doreen Kimura. The MIT Press, Cambridge, Massachusetts, 1999, 217 pp., $30.00. Reviewed by Melissa Hines, PhD1

Kimura is a well-established neuropsychologist with a prominent research record. Most recently, she has increasingly turned her attention to the study of sex differences in human cognition. This book offers a popular treatment of her perspective on the field. Her main point is that cognitive sex differences are large (the cover illustration suggests that men and women are as different in this respect as apples and oranges) and that they are caused primarily by gonadal hormones (androgens and estrogens). She also suggests that because of this biological basis, different sex ratios in certain professions, such as science and engineering, are unavoidable. Like Kimura, I am interested in hormonal influences on human cognition and I have been conducting research in this area for more than 20 years. What struck me most in reading this book was the extent to which two scientists with similar interests, looking at the same body of evidence, could reach strikingly different conclusions. From my perspective, sex differences in cognition exist, but are relatively small. For instance, the most important aspects of cognitive function, including general intelligence, do not differ between males and females. Nor do many other aspects. There are sex differences in certain specific abilities, such as some aspects of visuospatial ability, mathematical ability, and verbal ability, but, even here, other aspects of these broad categories do not differ for males and females. For instance, males are better than females at mental rotation tasks, but not at spatial memory tasks. Also, they excel at mathematical problem solving, but not at mathematical computations or mathematical reasoning. Similarly, women show better verbal fluency than men do, but do not excel at vocabulary or reading comprehension. Even in the limited areas where cognitive sex differences exist, they are not very big. The largest is about half the size of the sex difference in height. Finally, cognitive differences related to sex are smaller than cognitive differences related to other factors. One example is cultural and educational background. Children in many 1 Department

of Psychology, City University, Northampton Square, London EC1V 0HB, England. 551 C 2001 Plenum Publishing Corporation 0004-0002/01/1000-0551$19.50/0 °

P1: VENDOR/GCO Archives of Sexual Behavior [asb]

552

PP181-340737-41

June 28, 2001

9:8

Style file version Nov. 19th, 1999

Book Reviews

nations, such as Japan and Hungary, routinely outperform children in the United States on mathematics and science tests. These national differences are sometimes severalfold larger than sex differences in cognitive abilities. I also do not agree that there is convincing evidence that hormones play a major role, or perhaps even any role, in the limited cognitive differences between men and women. Research in other mammals has shown that hormones exert two types of influences on behavior. Prenatally and neonatally, they direct certain developmental processes in some parts of the brain and, consequently, permanently influence some behaviors. These early, permanent influences of hormones are called organizational effects. After puberty, the effects of hormones typically wax and wane as hormone levels rise and fall. These more transient, postpubertal, influences of hormones are called activational effects. Both types of effects have been studied most extensively in regard to the reproductive behavior of laboratory rats; however, they also have been demonstrated in other species, including nonhuman primates, and for other behaviors, including certain “cognitive” capabilities, such as maze learning and avoidance learning in rodents. It therefore is reasonable to investigate the hypothesis that hormones influence human behaviors that show sex differences, including cognitive functions that do so. In regard to organizational effects of hormones, most of the available information has come from situations where people have been exposed to unusual hormone environments because of medical disorders or because their mother was treated for medical reasons with hormones during pregnancy. Research to date suggests that the early hormone environment plays a role in children’s gender role behavior, such as their preferences for toys like dolls or trucks and for playmates of the same sex. Several studies using various measures and different control procedures have found that girls exposed prenatally to high levels of androgens (hormones typically produced by the male gonads prenatally), because of a genetic disorder called congenital adrenal hyperplasia (CAH), are more masculine-typical in these respects than are other girls; however, research on similar outcomes for cognitive development is far from conclusive. In three areas where sex differences are thought to exist (mental rotations ability, mathematical ability, and verbal fluency) evidence from CAH and other disorders is equivocal or negative (for a review, see Hines, 2000). In regard to activational effects of hormones, the most commonly studied model is the female menstrual cycle. In a recent review, Epting and Overman (1998) concluded that, among studies relating cognitive performance to hormone levels across the menstrual cycle, the “results have been equivocal.” Perhaps more tellingly, studies where cognitive abilities have been measured before and after hormonal manipulations have shown few or no changes in cognitive abilities that show sex differences. Of course, a complete absence of a relation between hormones and human cognition cannot be proved. It is possible that larger samples, or more sensitive measures, or treatment with different hormones, or at a different time might reveal

P1: VENDOR/GCO Archives of Sexual Behavior [asb]

Book Reviews

PP181-340737-41

June 28, 2001

9:8

Style file version Nov. 19th, 1999

553

something. The point is that these effects have not been demonstrated to date. To support her perspective, Kimura has described studies that are consistent with her thesis, omitting many that are not. Ironically, Kimura begins her book with a quote from Kenneth Hilborn: “When science ignores facts in [favor] of ideology . . . it ceases to be science and becomes propaganda for a dogma.” Apparently, Kimura is unaware that she could be promoting a dogma. Maccoby and Jacklin’s classic work, The Psychology of Sex Differences, remains useful for anyone hoping to understand research in this complex area (Maccoby and Jacklin, 1974). Although some of their specific conclusions no longer hold because of the passage of time and the accumulation of new information, their descriptions of pitfalls in research on sex differences remain accurate. Kimura’s conclusions could relate in part to one of these. That is, that a scientist’s preconceptions, or stereotypes, can influence his or her conclusions, both regarding individual data or the big picture that findings in a field suggest. Thus, those who already believe that sex differences are innate and irreversible are likely to see biological determinism where others do not and to remember studies that fit their preconceptions while forgetting those that do not. This general problem of preconceptions influencing conclusions has also been studied extensively from the perspectives of gender schema research and social constructionism. Kimura would not be the first respected scientist to be misled by ideology. As Gould (1981) has pointed out, in the nineteenth century, many scientists, including the distinguished neurologist Broca (who founded the Anthropological Society of Paris) and his colleague Le Bon (a founder of sociology), fell victim to the stereotypes of their era. They, for instance, viewed differences in brain size between men and women as indicating differences in general intellectual capability. According to Le Bon, “there are a large number of women whose brains are closer in size to those of gorillas than to the most developed male . . . psychologists who have studied the intelligence of women, as well as poets and novelists, recognize today that they represent the most inferior forms of human evolution . . .” (Gould, 1981, pp. 104–105). Other races also were viewed as inferior intellectually, based on then-current estimates of their brain size and brain shape. Meanwhile, eminent German scientists reported that German brains were larger than French brains, proving, according to the Germans, their innate superiority. What if future research should reveal hormonal influences on those aspects of cognition that differ for males and females? Could we ever say, as Kimura suggests, that hormones are more important than other factors, such as education or the social environment? Here too I differ with Kimura. The relative contributions of these factors are not fixed. A research study provides a snapshot of a situation at a particular point in time. It tells us how things are in our society at this moment, not how they might be in the future or could be if we wished them otherwise. Those studying genetic influences on human behavior, including human cognition, have

P1: VENDOR/GCO Archives of Sexual Behavior [asb]

PP181-340737-41

June 28, 2001

554

9:8

Style file version Nov. 19th, 1999

Book Reviews

noted that the question of whether genetic or environmental influences are more important misses a more fundamental point. That is, that by changing the contribution of one, we change the contribution of the other. For example, if we were all genetically identical, the variability in our behavior would be caused completely by nongenetic factors, whereas if we all experienced the same environment, the variability in our behavior would come from elsewhere, perhaps genetics. Hormones could be substituted for genetics and the concept would remain the same. We determine the importance of hormones by manipulating other factors. Thus, regardless of whether hormones eventually prove to play a role in human cognition, we decide the extent to which they control our destiny. Kimura’s book is well-written and nicely illustrated and could be viewed as providing a provocative, popular introduction to the idea that hormones can influence human behavior. Does it matter whether the specific points it makes are accurate? Of course, it does. One fundamental finding in psychological research is that expectations influence performance. Kimura’s book sends a message to parents, educators, and children themselves that boys are biologically programmed to do better than girls in areas related to scientific achievement. This could become a self-fulfilling prophecy. Based as it is on questionable information and interpretations, it disserves girls who might otherwise do very well in science. It also is unfair to society in general. In an era where scientific competence is increasingly important, we need all the well-trained scientists we can get. Finally, it impedes genuine scientific understanding of the factors, perhaps including hormones, that influence human cognition. If we assume we already have answers to the questions, we are unlikely to find the real answers or to ask more revealing questions. REFERENCES Epting, L. K., and Overman, W. H. (1998). Sex-sensitive tasks in men and women: A search for performance fluctuations across the menstrual cycle. Behav. Neurosci. 112: 1304–1317. Gould, S. J. (1981). The Mismeasure of Man, W. W. Norton & Co., New York. Hines, M. (2000). Gonadal hormones and sexual differentiation of human behavior: Effects on psychosexual and cognitive development. In Matsumoto, A. (ed.), Sexual Differentiation of the Brain, CRC Press, Boca Raton, FL. Maccoby, E. E., and Jacklin, C. N. (1974). The Psychology of Sex Differences, Stanford University Press, Stanford, CA.

Physical Attractiveness and the Theory of Sexual Selection. By Doug Jones. Museum of Anthropology, University of Michigan Press, Ann Arbor, Michigan, 1996, 174 pp., $24.00. Reviewed by John Marshall Townsend, PhD2

2 Department

of Anthropology, Syracuse University, Syracuse, New York 13244-2340.

P1: VENDOR/GCO Archives of Sexual Behavior [asb]

PP181-340737-41

June 28, 2001

554

9:8

Style file version Nov. 19th, 1999

Book Reviews

noted that the question of whether genetic or environmental influences are more important misses a more fundamental point. That is, that by changing the contribution of one, we change the contribution of the other. For example, if we were all genetically identical, the variability in our behavior would be caused completely by nongenetic factors, whereas if we all experienced the same environment, the variability in our behavior would come from elsewhere, perhaps genetics. Hormones could be substituted for genetics and the concept would remain the same. We determine the importance of hormones by manipulating other factors. Thus, regardless of whether hormones eventually prove to play a role in human cognition, we decide the extent to which they control our destiny. Kimura’s book is well-written and nicely illustrated and could be viewed as providing a provocative, popular introduction to the idea that hormones can influence human behavior. Does it matter whether the specific points it makes are accurate? Of course, it does. One fundamental finding in psychological research is that expectations influence performance. Kimura’s book sends a message to parents, educators, and children themselves that boys are biologically programmed to do better than girls in areas related to scientific achievement. This could become a self-fulfilling prophecy. Based as it is on questionable information and interpretations, it disserves girls who might otherwise do very well in science. It also is unfair to society in general. In an era where scientific competence is increasingly important, we need all the well-trained scientists we can get. Finally, it impedes genuine scientific understanding of the factors, perhaps including hormones, that influence human cognition. If we assume we already have answers to the questions, we are unlikely to find the real answers or to ask more revealing questions. REFERENCES Epting, L. K., and Overman, W. H. (1998). Sex-sensitive tasks in men and women: A search for performance fluctuations across the menstrual cycle. Behav. Neurosci. 112: 1304–1317. Gould, S. J. (1981). The Mismeasure of Man, W. W. Norton & Co., New York. Hines, M. (2000). Gonadal hormones and sexual differentiation of human behavior: Effects on psychosexual and cognitive development. In Matsumoto, A. (ed.), Sexual Differentiation of the Brain, CRC Press, Boca Raton, FL. Maccoby, E. E., and Jacklin, C. N. (1974). The Psychology of Sex Differences, Stanford University Press, Stanford, CA.

Physical Attractiveness and the Theory of Sexual Selection. By Doug Jones. Museum of Anthropology, University of Michigan Press, Ann Arbor, Michigan, 1996, 174 pp., $24.00. Reviewed by John Marshall Townsend, PhD2

2 Department

of Anthropology, Syracuse University, Syracuse, New York 13244-2340.

P1: VENDOR/GCO Archives of Sexual Behavior [asb]

Book Reviews

PP181-340737-41

June 28, 2001

9:8

Style file version Nov. 19th, 1999

555

This book is an entry in the University of Michigan, Museum of Anthropology, Anthropological Papers Monograph series. It brings together in a useful way the author’s dissertation research and data that have appeared in Human Nature and Current Anthropology. The first two chapters lay the theoretical foundation, describing the relationships among general evolutionary theory, sexual selection, psychology, and anthropology. Readers that are unfamiliar with modern evolutionary theory will have a tough time following the logic, and skeptics will not be convinced, but Jones’ coverage is appropriate for a monograph, and he cites pertinent introductory sources for interested readers. Chapter 3 describes methods and the five populations studied: University of Michigan students, university students in Moscow, Hiwi Indians of Venezuela, Ache Indians of Paraguay, and three sites in Bahia Brazil, which included lower and middle class subjects in and around the city of Salvador. Chapter 4 defines mate value and offers a good discussion of the interrelationships among age, attractiveness, and mate value. Because both short- and long-term mate values vary with age more for women than for men, men emphasize physical traits in their assessments of sexual attractiveness more than do women. This appears to be a human universal. In passing, Jones notes that no known society contradicts the general pattern of sex differences that he describes and that this finding is inconsistent with the view that sex differences in sexuality are purely social constructions. More specifically, ethnographic evidence does not support the feminist claim that the male emphasis on youth and beauty reflects a fear of older, more powerful women and a desire to control and dominate weaker, na¨ıve younger women. In fact, other things being equal, younger women have more bargaining power and are often more difficult for men to control because younger women are more sexually attractive. As Indian and non-Indian Brazilian men said, younger women are more fickle and “stingy with their vaginas” (p. 80). Data described in the first half of Chapter 5 indicated the following: (1) female faces that appeared younger than their actual ages were rated as more attractive by the five male samples; (2) a sample of U.S. professional female models had more neotenous facial proportions than did U.S. female undergraduates; (3) female faces artificially transformed to make them more or less neotenous were rated respectively as more or less attractive; and (4) associations between facial neoteny and male attractiveness were weak and/or inconsistent. Cross-culturally (and prior to European contact and colonialism) skin color at the lighter end of the population continuum was (and is) considered more attractive in women but not in men. Along with female age, a waist–hip ratio within the normal range for fecund females appears to be a universal index of fertility and therefore universal components of female sexual attractiveness. These indices are significantly less important in the assessment of male attractiveness. Another consistent sex difference superimposed on cross-cultural variation is that men find exaggerated secondary sex characteristics more attractive than women do. Within

P1: VENDOR/GCO Archives of Sexual Behavior [asb]

556

PP181-340737-41

June 28, 2001

9:8

Style file version Nov. 19th, 1999

Book Reviews

this general parameter, however, some preferences do seem to be largely culturally determined. Brazilian men, for example, prefer women with large buttocks and small breasts, whereas U.S. men prefer the opposite. In Chapter 6, Jones analyzes the effects of bodily symmetry, averageness, and health on attractiveness. Evolutionary theory suggests that, in general, proportions close to the population average should be more attractive—except for neotenous features and exaggerated secondary sex characteristics that tend to enhance female attractiveness. In Jones’ data, neither male nor female attractiveness correlated consistently with bodily symmetry (fluctuating asymmetry), but correlations might be stronger in environments with large loads of parasites and pathogens. After sketching a brief but interesting history of Brazilian society and race relations, Jones analyzes the interactions of race, social status, and attractiveness in Chapter 7. Attractiveness is partly determined by an individual’s deviation (somatic distance) from the population average for a trait (somatic norm) and partly by the social status of the group who predominantly exhibits a particular trait. Hence, Jones concludes that in Brazil the effects of social status and the population average have combined to produce somatic prejudice against pronounced African features; these effects, however, interact to ensure that pronounced European features are not favored over intermediate features. In this chapter, and in Jones’ study generally, human nature emerges not as a hard-wired hierarchy of values, “but as a set of mechanisms for generating standards as a function of individual experience and surrounding cultural values” (p. 144). The Conclusion (Chapter 8) offers a useful summary of the major findings and conclusions. Overall, this is an interesting monograph based on a sophisticated analysis of pertinent literature and unique data sets. As Symons observes in his Foreword, Jones’ research program is the only one that applied systematic methods across samples that included preliterate peoples with little exposure to industrialized societies and their cultural products. There is, however, an important lacuna in his argument. Jones notes that the following traits are more important in determining female attractiveness than male attractiveness: facial neoteny, skin color, exaggerated secondary sex characteristics, blondism, and age. But if women do not emphasize these traits in their judgments of male attractiveness, what do they emphasize? Jones mentions this sex difference in sexual attractiveness and its evolutionary explanation. But there is no systematic exploration or explanation of this sex difference. Probably the author should not be faulted for this because his emphasis is on physical attractiveness, as his title indicates. This emphasis, however, obscures important differences in sexual psychology—differences for which evolutionary psychology provides a cogent explanation. Evolutionary psychologists explain sex differences in sexual attraction in terms of differential parental investment (Buss, 1989; Townsend, 1998). What might be a simple act of copulation for a male can result for the female in the

P1: VENDOR/GCO Archives of Sexual Behavior [asb]

Book Reviews

PP181-340737-41

June 28, 2001

9:8

Style file version Nov. 19th, 1999

557

medical risks of pregnancy and delivery, and the lifelong investment of motherhood. Symons (1979) argued that this discrepancy in minimum possible investment caused, through natural selection, females’ mate evaluation to be more selective than males’, with more emphasis on partners’ potential for parental investment— social dominance, prowess, nurturance—and less emphasis on physical attributes that serve as cues to fertility. In comparison, natural selection favored males who were attracted to a great variety of partners, and who assessed these partners’ acceptability for intercourse largely on the basis of physical attributes that serve as cues to fertility: signs of age, facial proportions, waist-to-hip ratio. Most studies of attractiveness have failed to include (or to manipulate systematically) the traits that influence women’s judgements of attractiveness more than men’s. When information that is more important to women is limited or absent from research designs, women base their judgements on facial features or other physical traits by default, thus creating a misleading picture of sex differences and similarities. Morse et al. (1976) pointed this out more than 20 years ago but this crucial consideration apparently got lost in the mania for “sophisticated” methods and number crunching (Townsend, 1998). A corollary problem is that researchers (both evolutionary and social– psychological) have reified the concept of attractiveness. Investigators typically require subjects to rate or rank photographs of persons (or statements about persons) in terms of “attractiveness,” “sexual attractiveness,” or “physical attractiveness”. These procedures assume that when men and women rate these abstractions, they are evaluating the same traits. This assumption ignores evidence that certain traits have significantly different or even opposite effects on men’s and women’s ratings. In independent studies, women’s judgements, but not men’s, of physical, dating, or sexual attractiveness, were affected by peer opinion, target person’s costume, dominance, and apparent socioeconomic status. Male law students, but not females, were affected by models’ physical attractiveness when models were portrayed as having high status. Flesh and physique display increased women’s attractiveness to men, but reduced men’s attractiveness to women. Appearing to be older than 25 but under 50 had neutral or positive effects on men’s attractiveness to women, but negative effects on women’s attractiveness to men (for review, see Townsend, 1998). Jones also mentions sensitive period learning in his analysis. Although all his examples are male, he fails to discuss the apparent sex difference in the importance of this type of learning. Literature on the development of the paraphilias, sexual orientation, and gender transposition suggests that the mechanisms for mate evaluation and sexual arousal differ profoundly in men and women (Weinrich, 1988). In a process similar to imprinting, prepubertal boys appear to internalize specific physical attributes of a class of objects. Because of this process, virtually all males, heterosexual or gay, nonfetishists as well as fetishists, share the following

P1: VENDOR/GCO Archives of Sexual Behavior [asb]

PP181-340737-41

June 28, 2001

558

9:8

Style file version Nov. 19th, 1999

Book Reviews

characteristics: (1) once internalized, the physical attributes tend to be relatively inflexible; (2) because they can be sexually stimulated by viewing any superior example of the favored class, virtually all males experience the desire to possess a great variety of individuals (or objects) within that class. Compared to boys, girls do not internalize ritualized acts, or the physical attributes of a class of objects, the mere sight (or thought) of which can then stimulate sexual arousal. Consequently, true sexual fetishism is virtually unknown among women. Thus, compared to men’s attractiveness criteria, women’s criteria involve more nonphysical factors, focus more on partners’ potential for investment, and are more flexible and more contingent on a woman’s own age, socioeconomic status, and lifestyle. Evidently, most of the traits Jones discusses (age, skin color, secondary sex characteristics, skin tone, facial neoteny, sensitive period learning) are paramount in men’s evaluations of women’s attractiveness, but relatively unimportant in women’s evaluation of men’s attractiveness. But Jones is not alone in his omissions, and given his topic and emphasis, they are excusable. This is an interesting monograph, based on unique, careful research. It deserves serious consideration by anyone who is interested in evolutionary theory and sexual attraction. REFERENCES Buss, D. M. (1989). Sex differences in human mate preferences: Evolutionary hypotheses tested in 37 cultures. Behav. Brain Sci. 12: 1–49. Morse, S., Gruzen, J., and Reis, H. (1976). The “eye of the beholder”: A neglected variable in the study of physical attractiveness. J. Pers. 44: 209–225. Symons, D. (1979). The Evolution of Human Sexuality, Oxford University Press, New York. Townsend, J. M. (1998). What Women Want—What Men Want: Why the Sexes Still See Love and Commitment So Differently, Oxford University Press, New York. Weinrich, J. D. (1988). The periodic table model of the gender transpositions: Part II. Limerant and lusty sexual attractions and the nature of bisexuality. J. Sex Res. 24: 113–129.

Maternal Personality, Evolution, and the Sex Ratio: Do Mothers Control the Sex of the Infant? By Valerie J. Grant. Routledge, London, 1998, 222 pp., $24.99. Reviewed by Anthony F. Bogaert, PhD3

This book presents an interesting hypothesis about the origins of the human (and animal) sex ratio. It also touches on related issues (e.g., ethics of sex selection). Knowledge about the factors that affect the sex ratio is, of course, interesting from a number of perspectives. Animal breeders are interested because they often want to 3 Departments

of Community Health Sciences and Psychology, Brock University, St. Catharines, Ontario L2S 3A1, Canada.

P1: VENDOR/GCO Archives of Sexual Behavior [asb]

PP181-340737-41

June 28, 2001

558

9:8

Style file version Nov. 19th, 1999

Book Reviews

characteristics: (1) once internalized, the physical attributes tend to be relatively inflexible; (2) because they can be sexually stimulated by viewing any superior example of the favored class, virtually all males experience the desire to possess a great variety of individuals (or objects) within that class. Compared to boys, girls do not internalize ritualized acts, or the physical attributes of a class of objects, the mere sight (or thought) of which can then stimulate sexual arousal. Consequently, true sexual fetishism is virtually unknown among women. Thus, compared to men’s attractiveness criteria, women’s criteria involve more nonphysical factors, focus more on partners’ potential for investment, and are more flexible and more contingent on a woman’s own age, socioeconomic status, and lifestyle. Evidently, most of the traits Jones discusses (age, skin color, secondary sex characteristics, skin tone, facial neoteny, sensitive period learning) are paramount in men’s evaluations of women’s attractiveness, but relatively unimportant in women’s evaluation of men’s attractiveness. But Jones is not alone in his omissions, and given his topic and emphasis, they are excusable. This is an interesting monograph, based on unique, careful research. It deserves serious consideration by anyone who is interested in evolutionary theory and sexual attraction. REFERENCES Buss, D. M. (1989). Sex differences in human mate preferences: Evolutionary hypotheses tested in 37 cultures. Behav. Brain Sci. 12: 1–49. Morse, S., Gruzen, J., and Reis, H. (1976). The “eye of the beholder”: A neglected variable in the study of physical attractiveness. J. Pers. 44: 209–225. Symons, D. (1979). The Evolution of Human Sexuality, Oxford University Press, New York. Townsend, J. M. (1998). What Women Want—What Men Want: Why the Sexes Still See Love and Commitment So Differently, Oxford University Press, New York. Weinrich, J. D. (1988). The periodic table model of the gender transpositions: Part II. Limerant and lusty sexual attractions and the nature of bisexuality. J. Sex Res. 24: 113–129.

Maternal Personality, Evolution, and the Sex Ratio: Do Mothers Control the Sex of the Infant? By Valerie J. Grant. Routledge, London, 1998, 222 pp., $24.99. Reviewed by Anthony F. Bogaert, PhD3

This book presents an interesting hypothesis about the origins of the human (and animal) sex ratio. It also touches on related issues (e.g., ethics of sex selection). Knowledge about the factors that affect the sex ratio is, of course, interesting from a number of perspectives. Animal breeders are interested because they often want to 3 Departments

of Community Health Sciences and Psychology, Brock University, St. Catharines, Ontario L2S 3A1, Canada.

P1: VENDOR/GCO Archives of Sexual Behavior [asb]

Book Reviews

PP181-340737-41

June 28, 2001

9:8

Style file version Nov. 19th, 1999

559

control the sex of livestock (e.g., cattle) to increase productivity. Many (potential) parents are interested in sex/gender selection so that they can have a child of a particular sex that balances their family’s sex ratio or merely because they have a preference for a child of a particular sex. The human sex ratio is also of interest to a range of academics for a variety of reasons, not the least of which is that there are puzzling attributes associated with it. For example, all animals have a 100/100 secondary sex ratio, that is, 100 male live births to 100 female live births, whereas humans consistently have a higher (secondary) sex ratio, that is, 105/100. In addition, the human sex ratio seems to vary as a function of certain variables, such as ethnicity/race, socioeconomic status, and sexual orientation in men (e.g., Blanchard, 1997; James, 1987). The crux of Grant’s hypothesis is that biological factors related to both maternal personality and stress levels can affect a mother’s egg cells to be more or less receptive to X- or Y-bearing sperm. The personality trait in question is dominance, where women with a more dominant personality are more likely to have male children. The biological basis for a dominant personality and the mechanism underlying receptivity to X- or Y-bearing sperm is testosterone levels, where higher levels of testosterone increase a female’s dominance level and her likelihood of conceiving a male fetus. The book contains 10 chapters, the first two of which give an overview of Grant’s thesis and attempt to define and measure dominance. For example, Grant uses a definition of dominance offered by Fiske (1971): “The core of dominance can be identified as acting overtly so as to change the views or actions of another” (p. 98). Grant also makes a distinction between “comfortable or natural dominance” and “environmental or unnatural dominance,” the latter of which is not true dominance and might be best termed domineering or a need for control/dominance. In Chapter 2, Grant also outlines existing human studies, some of which are her own, that support the maternal dominance/sex-of-infant hypothesis. Although individual studies were weak, a meta-analysis demonstrated a strong dominance effect (Grant, 1994). It is of note that some of the studies measured dominance in women prior to pregnancy, strengthening the argument that it is maternal personality prior to pregnancy, and not merely carrying a male fetus, that raises a mother’s testosterone and hence her scores on dominance personality measures. Most of the remaining chapters outline and attempt to support in detail various aspects of Grant’s arguments. For example, Chapter 3 explores the biological basis of dominance, reviewing studies showing that testosterone and dominance are related in men, women, and children. In this chapter and in Chapters 5 and 8, Grant also introduces evidence that testosterone rises in women during times of stress. Grant argues that this is an adaptive response because women may need to become tougher (and more dominant) during such times, which increased testosterone (via the adrenal glands) provides. This trend in women, Grant argues, is the cause of a well-cited increase in males born during periods of stress such as famine or wars.

P1: VENDOR/GCO Archives of Sexual Behavior [asb]

560

PP181-340737-41

June 28, 2001

9:8

Style file version Nov. 19th, 1999

Book Reviews

Grant also argues that these findings could explain why males generally may be at risk for various forms of affliction: “If more males are conceived at a time when mothers are at the upper limits of conceiving at all, because of stressful conditions, then such males may be at a disadvantage from the beginning” (p. 107). In Chapter 4, Grant reviews some factors that seem to affect the sex ratio and how these factors can be interpreted as consistent with the maternal dominance/sexof-infant hypothesis. For example, Grant cites research showing that women who have more masculine body types, that is, lower waist-to-hip ratios (and hence likely to have higher testosterone levels and be more dominant), are more likely to have male children. Somewhat disappointingly, however, Grant doesn’t mention findings on sibling sex ratio and sexual orientation, where gay men (particularly feminine gay men) have been found to have higher sex ratios (Blanchard, 1997). One reason for this absence is that a fair proportion of the data on sex ratio and sexual orientation was likely published after this volume was written. However, if Grant’s argument about stress increasing the chance of having a male child is correct, then it may suggest that elevated sex ratios in some groups of gay men result from their mothers having had stressful lives and thus having produced a higher number of boys relative to girls. Such stress would also, presumably, raise the vulnerability of some of these males to atypical development, including a homosexual orientation. Interestingly, there has been some support for a relationship between maternal stress and homosexuality, although there is contradictory evidence (Bailey et al., 1991). In Chapter 9, Grant suggests there is an evolutionary advantage for dominant mothers to raise boys and nondominant mothers to raise girls. Dominant (high testosterone) mothers should be better suited to raise boys because these mothers are more likely to engage in active and boisterous interactions with male infants that in turn leads to more aggressive and competitive boys and men; in contrast, nondominant mothers may be better suited to raise girls to be less aggressive and assertive and more nurturing. To support her case, Grant refers to literature in developmental psychology showing marked differences in how parents respond to infants of different sexes. However, in forcefully arguing this point, I wonder if she partially undermines the biological arguments that she uses to generate her original hypothesis. In particular, if a biological factor such as testosterone is critical to gender and individual differences in dominance (and other behaviors), why is there such a need for nurturance on the part of mothers to produce gender differences? Additional questions remain to be answered as well. For example, Grant argues that race/ethnicity differences in the sex ratio are probably the result of sampling problems. In particular, she argues that Blacks have a lower sex ratio than Whites have because of the fact that the black populations studied were living within regions with large white populations who hold a more dominant and superior position in society. Grant implies that this weaker societal position relative to Whites lowers Black people’s testosterone level and makes them more likely to have female offspring. Yet, to my knowledge, research indicates that Blacks (at

P1: VENDOR/GCO Archives of Sexual Behavior [asb]

PP181-340737-41

June 28, 2001

9:8

Style file version Nov. 19th, 1999

Book Reviews

561

least men) have a higher level of testosterone than do Whites (Ellis and Nyborg, 1992). In summary, this book provides a detailed account of an interesting hypothesis regarding the human and animal sex ratio, although some aspects of the account are less satisfying than are others. It also provides extensive information and citations on related topics, and, as such, academics who are interested in issues of sex ratio, sex/gender selection, and/or dominance as a personality characteristic might be informed and stimulated by reading it. REFERENCES Bailey, J. M., Willerman, L., and Parks, C. (1991). A test of the maternal stress theory of human male homosexuality. Arch. Sex. Behav. 20: 277–294. Blanchard, R. (1997). Birth order and sibling sex ratio in homosexual versus heterosexual males and females. Ann. Rev. Sex. Res. 8: 27–67. Ellis, L., and Nyborg, H. (1992). Racial/ethnic variations in male testosterone levels: A probable contributor to group differences in health. Steroids 57: 72–75. Fiske, D. W. (1971). Measuring the Concepts of Human Personality, Aldine, Chicago. Grant, V. J. (1994). Maternal dominance and the conception of sons. Br. J. Med. Psychol. 67: 343–351. James, W. H. (1987). The human sex ratio. Part I: A review of the literature. Hum. Biol. 59: 721–752.

Athletes and Acquaintance Rape. By Jeffrey R. Benedict. Sage, Thousand Oaks, CA, 1998, 105 pp., $23.50. Reviewed by Robert H. Pollack, PhD4

This book illustrates the use of selected case histories to make a point about the presumed causes of predatory sexual behavior by professional athletes, particularly African American athletes. The three cases are well documented and presented in a dispassionate manner. In addition, the life histories of the perpetrators are described in some detail to provide the rationale for the emergence of assaultive behavior. The behavior of the employers of these athletes in the protection of their human investments is detailed as well. Benedict’s conclusion concerning the reasons for the public acceptance of antisocial behavior by athletes and sports entrepreneurs appears to be well supported by the anecdotal material presented in the narrative. Athletes, particularly those from disadvantaged, non-middle-class background, exploited by coaches and alumni boosters in high school and college, and later by professional teams or managers, are rewarded for athletic ability to the exclusion of education or respect for common societal norms. As long as their performance satisfies the emotional needs of alumni and fans, and later the financial goals of their employers, they 4 Department

of Psychology, University of Georgia, Athens, Georgia 30602.

P1: VENDOR/GCO Archives of Sexual Behavior [asb]

PP181-340737-41

June 28, 2001

9:8

Style file version Nov. 19th, 1999

Book Reviews

561

least men) have a higher level of testosterone than do Whites (Ellis and Nyborg, 1992). In summary, this book provides a detailed account of an interesting hypothesis regarding the human and animal sex ratio, although some aspects of the account are less satisfying than are others. It also provides extensive information and citations on related topics, and, as such, academics who are interested in issues of sex ratio, sex/gender selection, and/or dominance as a personality characteristic might be informed and stimulated by reading it. REFERENCES Bailey, J. M., Willerman, L., and Parks, C. (1991). A test of the maternal stress theory of human male homosexuality. Arch. Sex. Behav. 20: 277–294. Blanchard, R. (1997). Birth order and sibling sex ratio in homosexual versus heterosexual males and females. Ann. Rev. Sex. Res. 8: 27–67. Ellis, L., and Nyborg, H. (1992). Racial/ethnic variations in male testosterone levels: A probable contributor to group differences in health. Steroids 57: 72–75. Fiske, D. W. (1971). Measuring the Concepts of Human Personality, Aldine, Chicago. Grant, V. J. (1994). Maternal dominance and the conception of sons. Br. J. Med. Psychol. 67: 343–351. James, W. H. (1987). The human sex ratio. Part I: A review of the literature. Hum. Biol. 59: 721–752.

Athletes and Acquaintance Rape. By Jeffrey R. Benedict. Sage, Thousand Oaks, CA, 1998, 105 pp., $23.50. Reviewed by Robert H. Pollack, PhD4

This book illustrates the use of selected case histories to make a point about the presumed causes of predatory sexual behavior by professional athletes, particularly African American athletes. The three cases are well documented and presented in a dispassionate manner. In addition, the life histories of the perpetrators are described in some detail to provide the rationale for the emergence of assaultive behavior. The behavior of the employers of these athletes in the protection of their human investments is detailed as well. Benedict’s conclusion concerning the reasons for the public acceptance of antisocial behavior by athletes and sports entrepreneurs appears to be well supported by the anecdotal material presented in the narrative. Athletes, particularly those from disadvantaged, non-middle-class background, exploited by coaches and alumni boosters in high school and college, and later by professional teams or managers, are rewarded for athletic ability to the exclusion of education or respect for common societal norms. As long as their performance satisfies the emotional needs of alumni and fans, and later the financial goals of their employers, they 4 Department

of Psychology, University of Georgia, Athens, Georgia 30602.

P1: VENDOR/GCO Archives of Sexual Behavior [asb]

PP181-340737-41

June 28, 2001

562

9:8

Style file version Nov. 19th, 1999

Book Reviews

will be shielded from the consequences of illegal assaultive behavior. Only when the misbehavior degrades performance or somehow offends enough fans is the offender punished. In Roman times, the tarnished Gladiator would be killed in the arena. Today, we take away the livelihood of drug using and gambling athletes. The use of the particular cases makes for interesting reading, and they do serve as vivid examples in support of Benedict’s argument, but the case could have been made more strongly by the introduction of statistics dealing with arrest rates for sexual and other assaultive crimes committed by athletes, both in college and professional ranks compared with arrest rates for nonathletes matched for age, ethnicity, education, etc. There was mention made recently of very high arrest rates of athletes in press and televison stories. There are also questions to be raised about the sports chosen: football, basketball, and boxing, fields dominated by Black athletes. What about baseball with its large Latino representation and the violent sport of ice hockey with its virtually all White personnel? In short, this slim volume is the most valuable as a call to solid scientific research dependent on detailed study of large samples of athletes from a variety of high-paying and high-prestige sports populated by varied ethnic and socioeconomic groups. In the meantime, since this book was published, the public’s almost universal forgiveness for athletes’ antisocial behavior seems to be diminishing. Arrests of high profile college football players for assault and stealing, and the indictments of two premier professionals from Baltimore and Charlotte seem to have spooked the selectors in this year’s National Football League’s College Draft. They have suddenly discovered “character” as an indication of a player’s reliability along with his injury record. Benedict should be encouraged to continue his research within the canons of social science research. His work could well spur a call for a change in the way schools and society at large socialize budding athletes and perhaps lead to a reappraisal of the role of competitive spectator sports.

Barbie Culture. By Mary F. Rogers. Sage, Thousand Oaks, CA, 1999, 171 pp., $21.95 (paperback), $27.95 (hardcover). Reviewed by Adrian Furnham, D.Phil., D.Sc.5

It is precisely this sort of book that gets sociology a bad name among behavioral, medical, and physical scientists. It is badly researched, biased, trivial, repetitive, and tedious. It is not the topic per se that is the problem, but the research method, the crypto-agenda, and the lack of theory. 5 Department

of Psychology, University College London, 26 Bedford Way, London WC1, England.

P1: VENDOR/GCO Archives of Sexual Behavior [asb]

PP181-340737-41

June 28, 2001

562

9:8

Style file version Nov. 19th, 1999

Book Reviews

will be shielded from the consequences of illegal assaultive behavior. Only when the misbehavior degrades performance or somehow offends enough fans is the offender punished. In Roman times, the tarnished Gladiator would be killed in the arena. Today, we take away the livelihood of drug using and gambling athletes. The use of the particular cases makes for interesting reading, and they do serve as vivid examples in support of Benedict’s argument, but the case could have been made more strongly by the introduction of statistics dealing with arrest rates for sexual and other assaultive crimes committed by athletes, both in college and professional ranks compared with arrest rates for nonathletes matched for age, ethnicity, education, etc. There was mention made recently of very high arrest rates of athletes in press and televison stories. There are also questions to be raised about the sports chosen: football, basketball, and boxing, fields dominated by Black athletes. What about baseball with its large Latino representation and the violent sport of ice hockey with its virtually all White personnel? In short, this slim volume is the most valuable as a call to solid scientific research dependent on detailed study of large samples of athletes from a variety of high-paying and high-prestige sports populated by varied ethnic and socioeconomic groups. In the meantime, since this book was published, the public’s almost universal forgiveness for athletes’ antisocial behavior seems to be diminishing. Arrests of high profile college football players for assault and stealing, and the indictments of two premier professionals from Baltimore and Charlotte seem to have spooked the selectors in this year’s National Football League’s College Draft. They have suddenly discovered “character” as an indication of a player’s reliability along with his injury record. Benedict should be encouraged to continue his research within the canons of social science research. His work could well spur a call for a change in the way schools and society at large socialize budding athletes and perhaps lead to a reappraisal of the role of competitive spectator sports.

Barbie Culture. By Mary F. Rogers. Sage, Thousand Oaks, CA, 1999, 171 pp., $21.95 (paperback), $27.95 (hardcover). Reviewed by Adrian Furnham, D.Phil., D.Sc.5

It is precisely this sort of book that gets sociology a bad name among behavioral, medical, and physical scientists. It is badly researched, biased, trivial, repetitive, and tedious. It is not the topic per se that is the problem, but the research method, the crypto-agenda, and the lack of theory. 5 Department

of Psychology, University College London, 26 Bedford Way, London WC1, England.

P1: VENDOR/GCO Archives of Sexual Behavior [asb]

Book Reviews

PP181-340737-41

June 28, 2001

9:8

Style file version Nov. 19th, 1999

563

We are told in the Appendix (pp. 156–157) that the “crucial data” for this book comes from written statements—mainly high school and student undergraduate essays. Rogers also notes with disarming honesty that “Additional written statements came from five of the eight family members and friends whom I wrote in hope that they, too, would share their experiences and perceptions of Barbie” (p. 157). Written statements, interviews with collectors, and heavy reliance on a publication called Barbie Bazaar seem to constitute the research effort upon which this book was based. Barbie is a doll, first created in 1959 and phenomenally popular. The book is about what the doll represents as a core cultural icon. So, we have in the first two chapters a lot of sociobabble: “ . . . Barbie is an icon of the femininity associated with the middle reaches of contemporary Western societies” (p. 11) and “The doll is ubiquitous, pervasive and a role model of intensely gendered, emphatic femininity” (p. XX). Barbie has unobtainable perfection and ageless puberty. But, Barbie is not an icon of self-sacrificing motherhood—she succeeds in a male, professional world. And there are many different Barbies—Black and White, brides, fire fighters, etc. I would love to have seen a comprehensive list of the different Barbies, when they were introduced (and withdrawn), and why. We are told some most interesting “facts” about the doll. Boys enjoyed torturing Barbies. The doll cannot stand up unsupported and the figures are joined so it cannot hold anything—all symbolic you see. Barbies do not have a big bust—she has a small waist. But, presumably, this leads to a good waist-to-hip ratio that makes her fecund. Barbie, we are told, is not only an icon of femininity but also an icon of consumerism and materialism. Chapter 4 is the most interesting but where all the weaknesses of the book are exposed. It is a thinly disguised attack on Mattel (and all toy manufacturers) and advertising. Mattel is clearly a very clever manufacturer doing continuous market research and updating the project. It is no accident that it is a high success. But Rogers, who is clearly an anticapitalist, spends a number of pages attacking Mattel as an employer (pp. 101–108). Is this relevant to the topic one wonders? I would have liked to have seen some serious research on Mattel’s market research and their philosophy. I wanted to know about sales figures in different countries; a psychographic breakdown of the purchaser; the major competition; and Far Eastern copies. But we get no real research from the perspective of the advertiser, marketeer, or manufacturer. We know nothing of how children play with the doll; how many are bought by children, their parents, relatives, etc. My copy soon became dog-eared with scribbles and highlights. I highlighted variations in style, postmodernist impenetrable jargon, curious facts, and poor writing. I ended up profoundly impressed by Mattel’s research and marketing to keep a product alive for over 40 years.