Content area
Tomboys are girls who behave like boys and, as such, challenge some theories of sex-typing. Bailey et al recruited tomboys ages 4-9 through the media and compared them with their sisters and brothers on measures of playmate preference, sex-typed activities and interests and gender identity. On nearly all measures, tomboys were substantially and significantly more masculine than their sisters, but they were generally less masculine than their brothers.
Full text
Tomboys are girls who behave like boys and, as such, challenge some theories of sex-typing. We recruited tomboys (N = 60) ages 4-9 through the media and compared them with their sisters (N = 15) and brothers (N = 20) on measures of playmate preference, sex-typed activities and interests, and gender identity. On nearly all measures, tomboys were substantially and significantly more masculine than their sisters, but they were generally less masculine than their brothers. We outline some scientific benefits of studying tomboys and describe some goals and initial findings of the Tomboy Project.
KEY WORDS: tomboys; gender nonconformity; sex-atypicality; gender roles: gender development.
INTRODUCTION
The Tomboy Project is a longitudinal study of girls identified by their parents as tomboys. In this article, we outline the intentions of the Tomboy Project, present descriptive data from the first cohort of tomboys and their families, and demonstrate how data from tomboys can address important questions about gender development.
What Is a Tomboy?
The word "tomboy" conjures up similar images in most people's minds: a young girl, possibly with short hair and a baseball cap, who likes sports, plays with boys, and has little use for dolls and dresses. The Merriam-Webster dictionary (Merriam-Webster, 1998) defines "tomboy" as "a girl who behaves in a manner usually considered boyish." Although few would dispute this definition, it leaves unanswered several important questions.
For example, do tomboys behave like boys in all respects or only in some? Tomboys may be male-typical in their interest in sports and their aversion to dresses, but are they also more physically aggressive than other girls? Increasingly, psychologists have conceptualized sex-typed behaviors as a multidimensional space (Ruble & Martin, 1998), meaning that individuals can be male-typical in some respects and female-typical in others. In which behaviors do tomboys differ from other girls?
The answer to this question actually depends on two separate issues: our collective notion of "tomboy" and the actual covariation of sex-typed traits. It is possible, for example, that people understand "tomboy" merely to mean a girl who likes sports and other boy-typical activities, and that the concept has nothing to do with aggression. But it would still be possible that tomboys are more aggressive than other girls if aggressiveness and male-typical interests were positively associated among females.
The initial aim of the Tomboy Project was to explore traits that distinguish tomboys from other girls. Based on our intuitions and informal surveys of friends and colleagues, we hypothesized that girls identified as tomboys would differ greatly from other girls in their activities and interests, and that these differences would be particularly evident in their play behavior. Specifically, tomboys should be highly interested in sports and boys' toys, and they should lack interest in stereotypically female play activities. A second general area of hypothesized difference concerns peer preference. Tomboys should be more likely than other girls to prefer associating with boys. A third area of likely difference is gender identity, that is, degree of comfort with assigned sex and desire to be a member of the other sex. We are concerned with affective gender identity rather than cognitive gender identity, that is, believing that one is a girl versus aboy. The distribution of the affective component is continuous, ranging from normal comfort with and unquestioned acceptance of assigned sex to extreme discomfort and rejection. At the most extreme during adulthood, this is transsexualism. We hypothesized that, on average, tomboys would be elevated on measures of atypical gender identity.
The meaning of "tomboy" may be culturally influenced, and indeed, may be changing. For example, Morgan (1998) found that older women were less likely than younger women to recall tomboyish behavior during childhood. Thorne (1993) has suggested that, in contrast to adults, many children do not use the "tomboy" label. Thus, we emphasize the need to consider our study in its cultural context.
Why Study Tomboys?
Since beginning the Tomboy Project, we have encountered intense curiosity in the topic. We believe that this interest is well justified because tomboys can shed light on matters of fundamental importance, especially the causes and development of sex differences. Furthermore, as a conspicuous and heretofore neglected minority, tomboys deserve scientific attention for their own sake. For example, do tomboys have difficulty adjusting, and if so, does their behavior become less masculine as they age? This question not only has potential implications for the causes and maintenance of sex differences, but also for the mental health of tomboys. We summarize several related general issues that can be informed by data from tomboys.
Theories of the Development of Sex Differences
Because they behave like boys in some respects, tomboys provide an opportunity to explore why the sexes differ in those ways. Although the causes of differences within the sexes may differ from the causes of differences between the sexes, it is a reasonable hypothesis that tomboys and typical boys have been subject to similar influences with respect to some traits. But, because tomboys behave like typical girls in many respects, they also provide an opportunity to delineate factors that operate universally within-sex, making members of one sex similar to each other and different from members of the other sex.
The most persistent controversy in the realm of sex differences is the degree to which they are caused by biological factors, gender cognitions, or social experiences (Ruble & Martin, 1998). The most influential biological hypothesis is that prenatal androgens cause sex differences in the brain that subsequently cause behavioral sex differences. Studies of congenital adrenal hyperplasia (CAH), in which genetic females are exposed to high levels of prenatal androgens, suggest that a core component of tomboyism-sex-typed toy play-is influenced by early androgen exposure (Berenbaum & Hines, 1992). Thus, it is reasonable to hypothesize that tomboys were also subject to atypically high early androgen action, either because of high androgen levels or sensitive receptors. However, in contrast to girls with CAH, tomboys do not appear to show morphological changes that would suggest that they were exposed to excess androgens early in development, such as masculinized genitalia. A classic study of female rhesus monkeys showed that genital and behavioral masculinization could occur independently, depending on the timing of hormonal androgenization (Goy, Bercovitch, & McBrair, 1988). Consistent with this possibility, Udry (2000) has reported that women exposed to higher levels of sex hormone binding globulin (a molecule necessary for testosterone action) during the second trimester of fetal life were more psychologically masculine than other women; masculinity was unrelated to hormone exposure during the first or third trimesters. Unfortunately, it is currently impossible to determine retrospectively which females have been exposed to the highest level of prenatal androgen. However, if biological markers of early androgen exposure are discovered, then tomboys are an obvious group to study.
Social explanations of sex differences have stressed factors such as differential treatment (e.g., by parents, peers, teachers) and observational learning (Bussey & Bandura, 1999; Ruble & Martin, 1998). By these theories, tomboys would have experienced atypical socialization leading to sex-atypical behaviors. For example, perhaps their parents have atypical attitudes about sex role socialization, or perhaps their mothers are themselves somewhat masculine. The extreme variation in sex-typical behaviors shown in tomboys makes it far easier to study socialization contributions to these behaviors than is possible in typical samples of boys and girls. Because the Tomboy Project has focused almost exclusively, thus far, on tomboys and their siblings, we cannot currently identify differences between tomboys' parents and other parents. But, we might in principle identify differences in experiences between tomboys and their nontomboy sisters. Of course, it is important to be cautious in inferring that different experiences between tomboys and the sisters are the causes, rather than the effects, of their behavioral differences. One way that we have attempted to study the importance of socialization is by assessing parental attitudes and responses toward their tomboy daughters' masculine behavior. We can examine longitudinally whether tomboys whose parents are most disapproving become less masculine than other tomboys do.
In the future, we will directly study observational learning in tomboys, although we will examine this as a correlate, not as a cause of tomboyism. It will be interesting to examine whether tomboys are more likely to model opposite-sex rather than same-sex others, and whether this applies in all situations (including sex-neutral ones) or only in situations where the model is engaging in behaviors characteristic of tomboys (such as play with boys' toys).
Tomboys also provide an opportunity to examine cognitive theories of gender development, including relations between gender cognitions and the development of sex-typed behavior (Martin & Halverson, 1981; Ruble & Martin, 1998). Because gender schemas are thought to facilitate children's acquisition of sex-typed behavior (Martin & Halverson, 1981), tomboys would be expected to use gender cognitions in a different way than typical girls. For example, because of their sex-atypical behavior, tomboys might be expected to have particularly flexible gender schemas, making them less likely than typical girls to use sex to guide their own behavior or to make inferences about the behavior of others. Again, such studies provide information only about associations, not causes, but even this information is important in evaluating the validity of cognitive theories.
Causes of Sex Differences: Intervening Variables
Tomboys provide an opportunity to examine some specific hypotheses about the origins of particular sex differences. For example, consider visuospatial ability, which typically shows a moderate to large sex difference in favor of males (Linn & Peterson, 1985). Some researchers have suggested that this difference results from boys' and girls' different play experiences (Serbin, Zelkowitz, Doyle, Gold, & Wheaton, 1990). If this is true, then tomboys' visuospatial ability should be higher than that of other girls. But, if tomboys do not differ from other girls in visuospatial ability, the hypothesis would be falsified. Further, within the group of tomboys, girls with more boy-typical play experiences should have higher spatial ability than girls with fewer of those experiences. This is an issue being addressed in the Tomboy Project (Berenbaum, Bryk, Levine, Huttenlocher, & Bailey, 1999).
Adolescent and Adult Correlates of Tomboyism
The Tomboy Project will follow tomboys into adolescence and adulthood and assess a variety of sex-typed traits as potential adult correlates of tomboyism. It seems likely that on average, tomboys remain sex-atypical. Personality traits (Caspi & Bern, 1990; Conley, 1984, 1985) and interests (Swanson, 1999) have typically shown at least moderate stability over the lifespan, and there is no reason to suspect that tomboyism would differ in this respect. However, it is likely that the behavioral indicators of tomboyism may change in type, because the repertoire of available behaviors changes over the lifespan. For example, if tomboys have male-typical interests, they may be more likely to enter male-typical occupations, such as science and engineering. Indeed among adults, interest in male-typical versus female-typical occupations and recreational activities yields a large sex difference that corresponds closely to lay notions of masculinity-femininity (Lippa, 1995; Lippa & Connelly, 1990). In support of this supposition, a masculine pattern of interests in girls with CAH persists into adolescence (Berenbaum, 1999). Thus, occupational and recreational interests are good candidates for adolescent and adult correlates of tomboyism.
A second candidate for an adult correlate of tomboyism is sexual orientation. An association between childhood sex atypicality and homosexual orientation has been well established for males in both retrospective (Bailey & Zucker, 1995) and prospective (Green, 1987; Zuger, 1988) studies. Lesbians recall a higher rate of tomboyism than heterosexual women do (Bailey & Zucker, 1995), but this association has not been investigated prospectively. People are less likely to expect tomboys to become homosexual than they are to expect feminine boys to do so (Martin, 1990), and tomboyism is probably less predictive of a homosexual outcome (Bailey & Zucker, 1995). This is because more girls than boys exhibit sex-atypical behavior (Zucker & Bradley, 1995), but fewer women than men are homosexual (Laumann, Gagnon, Michael, & Michaels, 1994). Based on these findings from retrospective studies, we predict that tomboys will have an elevated rate of homosexuality, but that most will become heterosexual women (Bailey & Zucker, 1995).
Previous Research on Tomboys
Research on tomboys has been limited. A few studies have focused on beliefs about, or attitudes toward tomboys (Martin, 1990, 1995). We know of only two systematic studies focusing on parent-identified tomboy participants. The first (Green, Williams, & Goodman, 1982; Williams, Goodman, & Green, 1985; Williams, Green, & Goodman, 1979) was similar to this report in the comparison of parent-referred tomboys and nontomboy girls on a number of characteristics thought to be related to tomboyism, including interests and activities, playmate preferences, and gender identity. Although large differences between tomboys and nontomboy girls were obtained, measures were restricted to parent report. The second study (Plumb & Cowan, 1984) focused on sex-typed interests in self-defined tomboys and other children, and showed that the tomboys were more interested than other girls in traditional male activities. The sample of the second study had two potentially important limitations. First, tomboys were self-selected, and so were probably not very extreme. For example, self-reports usually reveal that more than half of girls and women tend to consider themselves tomboys (Hyde, Rosenberg, & Behr-man, 1977). Further, participants were rather old (some were in tenth grade) to be called tomboys, a term typically used for children.
Overview of This Study
In this paper, we present the first wave of data from our first cohort. Our primary intention was to characterize our sample of tomboys. Unlike prior studies, we used a variety of methods, including parent report, child self-- report, and direct observation. We also chose as comparison groups nontomboy sisters and brothers in order to control for general genetic and environmental background. When possible, we also provide data relevant to specific hypotheses.
METHOD
Participants
Participants were recruited through media reports to participate in a "study of tomboys." All girls aged 4-9 identified as tomboys by their parents were eligible. Sisters and brothers were included for comparison. There were 60 tomboys, 15 sisters, and 20 brothers ranging in age from 4 to 10 years (one girl was 9 when her parents called but 10 by the time we tested her). The mean ages of the three groups were similar and did not differ significantly: Tomboys, M = 7.2 yrs, SD = 1.7; Sisters, M = 7.0 yrs, SD = 1.9; Brothers, M = 6.8 yrs, SD = 2.4. Two tomboys and one brother were African American, two tomboys and one sister were Asian American, and one tomboy was Hispanic; all other participants were Caucasian. The majority of parents had college degrees.
Measures
Playmate Preference
Preference for boy versus girl playmates was assessed with two tasks. In the first task (Playmate Preference), participants were shown pictures of pairs of children and asked, "Who would you rather play with?" Thirteen items showed a boy and a girl, 3 showed two boys, and 3 showed two girls. The pictured children were matched on attractiveness, attire, and activities. The score was the number of boys chosen in the 13 opposite-sex pairings. This measure shows a large sex difference with little overlap between boys and girls (Berenbaum & Snyder, 1995). Scores were available for 60 tomboys, 15 sisters, and 20 brothers.
In the second task (Playmate and Activity Preference), participants were shown two playground scenes, each with 7 children (4 boys and 3 girls) engaged in sex-- typical activities, and 7 children (4 boys and 3 girls) engaged in sex-atypical activities. For each of 20 pairings of opposite-sex children, participants were asked with whom they would rather play. The score was the number of boys chosen.
A Playmate Preference composite was computed as follows. First, both aforementioned measures were standardized. Second, if one of the measures was missing, its score was imputed via regression with the available measure as the predictor. Third, the two scores were summed. Coefficient alpha for the composite (computed before imputation) was .81.
Sex-Typed Activities and Interests
Activities and interests were assessed by observation, child self-report, and parent report. Each participant was observed individually for 10 min playing with a standard set of toys, some typically preferred by girls and others by boys (Huston, 1983; Liss, 1981). The girls' toys were dolls and accessories, kitchen supplies, play jewelry, and cosmetics. The boys' toys were vehicles, Lincoln Logs, transformer, and gun and holster. Play was videotaped for later scoring for total time spent in play with boys' toys and girls' toys. Each tape was scored by one of three raters, who did not know whether female subjects were tomboys or sisters, and who rated similar tapes with high reliability (interrater correlation of .98). We refer to this measure as "Time Playing With Boys' Toys." Each participant was also asked to choose in private a toy to keep from a selection of four toys. The choice was scored as boy-typical (toy vehicle or football) or girl-typical (stuffed animal or purse), and the respective measure is called "Toy to Keep."
Each child reported interest in each of 64 sex-typed games and activities from the modified Child Game Participation Questionnaire (CGPQ; Meyer-Bahlburg, Feldman, & Ehrhardt, 1985). Parents indicated on a written form of this measure whether the child regularly participated in the activities. We counted the number of responses to stereotypically masculine games and activities (Masculine Interests, Child- and Parent-Report), that is, those which showed moderate to large sex differences in normal boys and girls (Sandberg & Meyer-Bahlburg, 1994). Scores on this subscale showed very large sex differences (> 1.5 SD) in another sample and differences between girls with CAH and control girls (unpublished data).
An Activities and Interests composite was computed similarly to the Playmate Preference composite. Coefficient alpha for the composite (computed before imputation) was .60.
Gender Identify
Only girl participants were assessed with respect to gender identity, because our measures can only be sensibly compared within-sex. Gender identity was assessed by child interview and parent report. All girls received a 14-item interview (Gender Identity Interview; Zucker et al., 1993). Eight questions from the Affective Gender Confusion Scale were used to measure gender identity; total scores could range from 0 (complete female gender identity) to 16 (complete male gender identity). Items on the Cognitive Gender Confusion scale were used to establish that girls had acquired gender constancy. One tomboy and one sister did not have gender constancy, so they were excluded from analyses.
Parents completed the Child Behavior Attitude Questionnaire (CBAQ, Parent Report), a 68-item questionnaire assessing sex-typed relations and behavior problems (Meyer-Bahlburg et al., 1985). Parents rated on a 5-point scale how often the child engaged in each behavior (1 = Never to 5 = Always). Eight items were considered to assess gender identity by consensus of 6 advanced graduate students' independent categorization of CBAQ items. The items concern the child's stated wish to be a boy or man, imitation of and identification with men versus women, and role-playing. Scores on these items were added, with total scores ranging from 8 (complete female gender identity) to 40 (complete male gender identity).
A composite was constructed from these two measures, as for the other traits. Coefficient alpha was .61.
Appearance
Participants were photographed standing, and the photographs were subsequently rated independently for masculine appearance, on a scale from I (very feminine) to 5 (very masculine), and for physical attractiveness, on a scale from I (very unattractive) to 5 (very attractive). The raters did not know whether female participants were tomboys or sisters.
Tomboy Composite
A Tomboy Composite was computed for female participants by summing the z-scores for the measures mentioned so far, except for physical attractiveness. All nine measures included seem relevant to the core tomboy concept. Missing values were imputed using data from scales in the same broad category (e.g., if only one score from the Activities and Interests domain was missing, then scores from the other three tests were used to estimate it). Coefficient alpha for the composite was .84.
Aggressiveness
Using a modification of an inventory developed by Reinisch (1981; Reinisch & Sanders, 1986), parents of subjects were asked how the child would respond if presented with six situations involving interpersonal conflict. (Parents were questioned because it is difficult to use selfreport in young children. Our experience indicates that direct questioning by an examiner produces socially desirable responses.) There are four possible responses, each indicating either verbal aggression, physical aggression, withdrawal, or nonaggressive coping. For each situation, there are six pairs of written forced-choice alternatives representing all possible combinations of the four behavioral responses (e.g., physical aggression vs. verbal aggression, verbal aggression vs. withdrawal). A score of 1 is obtained each time a specific behavioral response (physical aggression, verbal aggression, withdrawal, nonaggressive coping) is selected, so scores for each type of response range from 0 (that alternative is never selected) to 18 (that alternative is selected all three times in each of the six situations). We used the physical aggression subscale.
Because aggressiveness is not a central feature of stereotypes about tomboys, we were uncertain whether tomboys would be more aggressive than their sisters. Aggressiveness and physical attractiveness were the only measures included here for which we had no strong a priori hypotheses, although indirect evidence would be consistent with higher aggressiveness among tomboys. Boys and men are more likely than girls and women to report that they would use physical aggression (differences are large, 0.7-1.1 SDs) (Reinisch, 1981; Reinisch & Sanders, 1986). Furthermore, girls who had prenatal exposure to masculinizing hormones are more likely than their unexposed sisters to report that they would use physical aggression (Berenbaum & Resnick, 1997; Reinisch, 1981).
Procedure
Children were tested individually at the university. Parents completed questionnaires while the children were tested or later at home. Testing took approximately 2 hr. Some children were missing data because of examiner error or incomplete parent questionnaires, but subjects with complete data are representative of the full sample with respect to scores on all measures (Bechtold, 1996).
RESULTS
Relations Among Core Measures
Table I contains the correlations among measures that assess three traits that we have hypothesized are fundamental to tomboyism: playmate preference, sex-typed activities and interests, and gender identity. Each of these traits was assessed by at least two measures, and thus we could examine whether the convergent validities were larger than were the correlations between measures of different traits. In general, there was no evidence for convergent validity. The convergent validities averaged .34, compared with an average of .41 for the other correlations. All correlations were positive, and most were significant, suggesting that a general factor may be present. The correlations did not change much when boys were excluded.
A pattern of uniform positive correlations might occur either because all measures were indicators of a continuous latent dimension along which participants varied or because they were indicators of different types of people who were included in the sample. For example, a combined sample of men and women would yield a negative correlation between hair length and height, but this correlation is entirely due to the fact that women both wear their hair longer and are shorter than men. Including data from girls and boys, or from tomboys and sisters, in generating correlations risks analogous artifacts. Thus, we examined trait covariation among the tomboys alone, focusing on the three composites, and these results are presented in Table II. Correlations were still moderate and statistically significant. This supports the likelihood that tomboys represent the extreme of a latent dimension along which they too, vary.
We examined whether there were age effects on tomboyism, separately for tomboys and their sisters, by correlating age with the Tomboy composite. For tomboys, r(50) = .04 (ns), and for sisters r(10) = -.46 (ns).
Differences Between Groups
Differences among sisters, tomboys, and brothers on the primary measures of this study are shown in Fig. 1. (Because we had no a priori hypothesis about physical attractiveness, the results are not included in Fig. 1. The groups did not differ significantly [p > .40].) In order to make the differences comparable across different measures, the differences are presented as effect sizes, with sisters anchored at zero, and tomboys' and brothers' scores in (within-group) standard deviation units. Positive effect sizes represent more masculine scores.
All measures given to both girls and boys showed large differences between the nontomboy sisters and brothers, with Cohen's d ranging from 0.8 to 3.4 (Cohen, 1988). Tomboys were also substantially elevated on the measures compared with their sisters, with effect sizes ranging from 0.4 to 2.8, and all but two effect sizes exceeding 1.0. On the Tomboy composite, 48 of 50 tomboys with nonmissing scores exceeded the highest scoring sister.
For statistical testing of the group differences, each variable was analyzed via a one-way ANOVA with subsequent Bonferroni tests of differences between each pair of means. The overall F test was significant with an alpha level of .01 for all measures except one (Physical Aggressiveness), which was significant with an alpha level of .05. On all measures given to both sexes, tomboys' brothers were significantly more masculine than their sisters (p < .01). On all measures except one (Physical Aggressiveness), tomboys were significantly more masculine than their sisters (p < .01). Tomboys differed from their brothers significantly on only two measures (Playmate Preference and Masculine Appearance), and in both cases, tomboys were less masculine (p < .05).
Variation Among Tomboys
Scores on the Gender Identity Interview illustrate the range of behavior we observed. Fifty-eight percent of the tomboys exceeded the highest-scoring sisters (who received scores of 4), and one tomboy was near the maximum, with a score of 14. On the other hand, 42% of tomboys received scores of 4 or below, and 3 tomboys received scores of zero. Visual inspection of the distribution of the Gender Identity Interview scores and of the Tomboy Composite (not shown) suggested that the tomboys came from the same (variable) distribution.
DISCUSSION
Results of our study confirm some common notions about tomboys. Specifically, the tomboys in this sample were more likely than their sisters to play with boys and to prefer boys' activities and toys. Indeed, on most measures of playmate and activities preferences, they were more similar to their brothers than to their sisters. They were also somewhat more likely than their sisters to express feelings of wanting to be boys, and of not wanting to be girls.
There was, however, considerable variability among the tomboys. Undoubtedly, this partly reflected our recruitment strategy, which was to rely on parents' classifications of their daughters as tomboys. Parents probably differed in the stringency of their criteria. This would explain why two girls in the tomboy subsample had Tomboy Composites lower than some nontomboy sisters. There was no evidence for bimodality of the Tomboy Composite, suggesting that the tomboys differed from each other only in degree, and not in kind. One implication of this variability is that tomboys' developmental outcomes may depend on how extreme their tomboyism is. We can examine this prospectively.
In order to place the tomboys' behavior in perspective, it is useful to compare tomboys and girls with CAH who are exposed to sex-atypical prenatal hormones and who have been assessed previously using several measures included in the current study: Playmate Preference, Time Playing With Boys' Toys, Toy to Keep, CGPQ Masculine Interests (Berenbaum & Snyder, 1995), and Physical Aggressiveness (Berenbaum & Resnick, 1997). In each case, on average, tomboys' scores were higher than those of girls with CAR Furthermore, another sample of tomboys was found to be significantly elevated on a scale of atypical gender identity compared with girls with CAH (Berenbaum & Bailey, 1998). On the one hand, these findings are not surprising because tomboys have been selected precisely because they are behaviorally masculine. On the other hand, the higher scores of tomboys generally, and especially those of the most atypical tomboys, underscore the complex origins of sex-typed behavior. Although we did not measure environment, our impression was that tomboys did not have very unusual rearing environments. Furthermore, their siblings were sex-typical. Nor is there evidence that tomboys were exposed to sex-atypical hormone levels early in development, although it is possible that they were exposed for a short time during a later critical period, that their brains have extremely sensitive androgen receptors, or that hormones causing tomboyism do not affect the external genitalia.
Tomboys were slightly, but not significantly more aggressive than their sisters. Of course, a failure to demonstrate a significant difference does not mean that no difference exists. However, assuming for the moment that tomboyism is uncorrelated with physical aggressiveness, what would this mean? This would suggest that the causes of the tomboys' sex atypicality in other domains (e.g., sex-typed interests and activities) do not overlap with the causes of the sex difference in aggressiveness. This would suggest, for example, that boys are not more aggressive because of exposure to different toys than girls are typically exposed to.
Tomboys were much more likely than their sisters to prefer male playmates. This finding is particularly interesting in view of Maccoby's work on sex segregation (Maccoby, 1998). Beginning in mid-childhood, children associate almost exclusively with same-sex peers. Maccoby (1998) has hypothesized that same-sex peer groups comprise an important vehicle for sex socialization. It is possible that tomboys' increased desire to associate with boys causally precedes their sex atypicality. More likely, tomboys' masculine behavior and interests may make associating with boys more attractive. In either case, increased association with boys (and decreased association with girls) may delay or prevent tomboys from learning some aspects of sex-typical behavior from their peers. We might expect, for example, that tomboys would begin dating later than other girls.
A related question, neglected in the present study, is whether tomboys' atypical interpersonal behavior affects their mental health. To our knowledge none of the girls in our study had been seen by mental health professionals for sex atypicality, and our impression was that few of the parents were concerned about their daughters' tomboy status. This contrasts with girls with gender identity disorder (GID), who are often treated for their extreme masculinity (Zucker and Bradley, 1995). Although we cannot provide direct comparisons, we suspect that the tomboys were less extreme in their sex atypicality compared with girls referred for GID. Another difference between tomboys and girls with GID concerned attractiveness; tomboys were not less attractive than their sisters, but one study found girls with GID to be less attractive than other girls (Fridell, Zucker, Bradley, & Maing, 1996).
Clearly, the bulk of the Tomboy Project's goals are ahead of us because they require following the tomboys into adolescence and beyond. Our purpose here has been to demonstrate that we have recruited a group of tomboys who are markedly different from other girls, and to characterize some of these differences. We hope that we have also conveyed some of the reasons behind our own enthusiasm for the Tomboy Project.
ACKNOWLEDGMENTS
Some of the work reported in this paper was completed by Kathleen Bechtold in partial fulfillment of the requirements for a Master's degree in Psychology at Southem Illinois University, Carbondale. The study was supported in part by National Institutes of Health Grant HD 19644. We thank the following people who contributed to the research: Kristina Korman coordinated the project and assisted in data processing; Corey Hillman, Sarah Mikach, and Peter Schreck tested subjects; Ellen Lin and Wendy Pillsbury scored videotapes; David DiLalla and Brenda Gilbert provided feedback as thesis committee members. Joan Linsenmeier provided feedback on an earlier version of this manuscript. We are especially grateful to the children and their parents for their enthusiastic participation in the study.
REFERENCES
Bailey, J. M., & Zucker, K. J. (1995). Childhood sex-typed behavior and sexual orientation: A conceptual analysis and quantitative review. Developmental Psychology, 31, 43-55.
Bechtold, K. T. (1996). Sex-segregation: Is it due to gender identity or toy and activity preferences? Unpublished master's thesis, Southern Illinois University, Carbondale.
Berenbaum, S. A. (1999). Effects of early androgens on sex-typed activities and interests in adolescents with congenital adrenal hyperplasia. Hormones and Behavior, 35, 102-110.
Berenbaum, S. A., & Bailey, J. M. (1998). Variation in female gender identity: Evidence from girls with congenital adrenal hyperplasia, tomboys, and typical girls. Unpublished manuscript.
Berenbaum, S. A., Bryk, K., Levine, S., Huttenlocher, J., & Bailey, J. M. (1999, April). Associations between sex-typed play and spatial
abilities in tomboys and typical girls. Poster presented at the biennial meeting of the Society for Research in Child Development, Albuquerque, NM.
Berenbaum, S. A., & Hines, M. (1992). Early androgens are related to childhood sex-typed toy preferences. Psychological Science, 3, 203-206.
Berenbaum, S. A., & Resnick, S. M. (1997). Early androgen effects on aggression in children and adults with congenital adrenal hyperplasia. Psychoneuroendocrinology, 22, 505-515.
Berenbaum, S. A., & Snyder, E. (1995). Early hormonal influences on childhood sex-typed activity and playmate preferences: Implications for the development of sexual orientation. Developmental Psychology, 31, 31-42.
Bussey, K., & Bandura, A. (1999). Social cognitive theory of gender development and differentiation. Psychological Review, 106, 676713.
Caspi, A., & Bem, D. J. (1990). Personality continuity and change across the life course. In L. A. Pervin (Ed.), Handbook of personality: Theory and research (pp. 549-575). New York: Guilford.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum.
Conley, J. J. (1984). Longitudinal consistency of adult personality: Selfreported psychological characteristics across 45 years. Journal of Personality and Social Psychology, 47, 1325-1333.
Conley, J. J. (1985). Longitudinal stability of personality traits: A multitrait-multimethod-multioccasion analysis. Journal of personality and Social Psychology, 49, 1266-1282.
Fridell, S. R., Zucker, K. J., Bradley, S. J., & Maing, D. M. (1996). Physical attractivenss of girls with gender identity disorder. Archives of Sexual Behavior, 25, 17-31.
Goy, R. W., Bercovitch, F. B., & McBrair, M. C. (1988). Behavioral maculinization is independent of genital masculinization in prenatally androgenized female rhesus macaques. Hormones and Behavior, 22, 552-571.
Green, R. (1987). The "sissy boy syndrome" and the development of homosexuality. New Haven, CT: Yale University Press.
Green, R., Williams, K., & Goodman, M. (1982). Ninety-nine "tomboys" and "non-tomboys": Behavioral contrasts and demographic similarities. Archives of Sexual Behavior, 11, 247-266.
Huston, A. C. (1983). Sex-typing. In E. M. Hetherington (Ed.), Handbook of child psychology: Socialization, personality, and social development (Vol. 4, pp. 388-467). New York: Wiley.
Hyde, J. S., Rosenberg, B. G., & Behrman, J. A. (1977). Tomboyism. Psychology of Women Quarterly, 2, 73-75.
Laumann, E. O., Gagnon, J. H., Michael, R. T., & Michaels, S. (1994). The social organization of sexuality: Sexual practices in the United States. Chicago: University of Chicago Press.
Linn, M. C.. & Peterson, A. C. (1985). Emergence and characterization of sex differences in spatial ability: A meta-analysis. Child Development, 56, 1479-1498.
Lippa, R. (1995). Do sex differences define gender-related individual differences within the sexes? Evidence from three studies. Personality and Social Psychology Bulletin, 21, 349-355.
Lippa, R., & Connelly, S. (1990). Gender diagnosticity: A new Bayesian approach to gender-related individual differences. Journal of Personality and Social Psychology, 59, 1051-1065.
Liss, M. B. (1981). Patterns of toy play: An analysis of sex differences. Sex Roles, 7, 1143-1150.
Maccoby, E. E. (1998). The two sexes: Growing up apart, coming together. Cambridge, MA: Harvard University Press.
Martin, C. L. (1990). Attitudes and expectations about children with nontraditional and traditional gender roles. Sex Roles, 22, 151-165. Martin, C. L. (1995). Stereotypes about children with traditional and nontraditional gender roles. Sex Roles, 33, 727-751.
Martin, C. L., & Halverson, C. F. (1981). A schematic processing model of sex typing and stereotyping in children. Child Development, 52, 1119-1134.
Merriam-Webster (Ed.). (1998). Merriam-Webster's collegiate dictionary (10th ed.). Springfield, MA: Merriam-Webster. Meyer-Bahlburg, H. F. L., Feldman, J. F., & Ehrhardt, A. A. (1985).
Questionnaires for the assessment of atypical gender role behavior: A methodological study. Journal of the American Academy of Child Psychiatry, 24, 695-701.
Morgan, B. L. (1998). A three generational study of tomboy behavior. Sex Roles, 39, 787-800.
Plumb, P., & Cowan, G. (1984). A developmental study of destereotyping and androgynous activity preferences of tomboys, nontomboys, and males. Sex Roles, 10, 703-712.
Reinisch, J. M. (1981). Prenatal exposure to synthetic progestins increases potential for aggression in humans. Science, 211, 11711173.
Reinisch, J. M., & Sanders, S. A. (1986). A test of sex differences in aggressive response to hypothetical conflict situations. Journal of Personality and Social Psychology, 50, 1045-1049.
Ruble, D. N., & Martin, C. L. (1998). Gender development. In W. Damon (Series Ed.) & N. Eisenberg (Vol. Ed.), Handbook of child psychology: Vol. 3. Social, emotional, and personality development (5th ed., pp. 933-1016). New York: Wiley.
Sandberg, D. E., & Meyer-Bahlburg, H. F. L. (1994). Variability in middle childhood play behavior: Effects of gender, age, and family background. Archives of Sexual Behavior, 23, 645-663.
Serbin, L. A., Zelkowitz, P., Doyle, A., Gold, D., & Wheaton, B. (1990). The socialization of sex-differentiated skills and academic performance: A mediational model. Sex Roles, 23, 613-628.
Swanson, J. L. (1999). Stability and change in vocational interests. In M. Savickas & A. Spokane (Eds.), Vocational interests: Meaning, measurement and counseling use (pp. 135-158). Palo Alto, CA: Davies-Black.
Thorne, B. (1993). Gender play: Girls and boys in school. New Brunswick, NJ: Rutgers University Press.
Udry, J. R. (2000). Biological limites of gender construction. American Sociological Review, 65, 443-457.
Williams, K., Goodman, M., & Green, R. (1985). Parent-child factors in gender role socialization in girls. Journal of the American Academy of Child Psychiatry, 26, 720-731.
Williams, K., Green, R., & Goodman, M. (1979). Patterns of sexual identity development: A preliminary report on the "tomboy." Research in Community and Mental Health, 1. 103-123.
Zucker, K. J., & Bradley, S. J. (1995). Gender identity disorder and psychosexual problems in children and adolescents. New York: Guilford.
Zucker, K. J., Bradley, S. J.. Sullivan, C. L., Kuksis, M., BirkenfeldAdams, A., & Mitchell, J. N. (1993). A gender identity interview for children. Journal of Personality Assessment, 61, 443-456.
Zuger, B. (1988). Is early effeminate behavior in boys early homosexuality? Comprehensive Psychiatry, 29, 509-519.
J. Michael Bailey, Ph.D., 1,6 Kathleen T. Bechtold, M.A., 2,3 and Sheri A. Berenbaum, Ph.D. 4,5
1Department of Psychology, Northwestern University, Evanston, Illinois.
2Department of Psychology, Southern Illinois University, Carbondale, Illinois.
3Present address: Department of Physical Medicine and Rehabilitation, Johns Hopkins University, Baltimore, Maryland.
4Department of Physiology, School of Medicine, Southern Illinois University, Carbondale, Illinois.
5Present address: Department of Psychology, Pennsylvania State University, University Park, Pennsylvania.
6To whom correspondence should be addressed at Department of Psychology, Northwestern University, 2029 Sheridan Road, Evanston, Illinois 60208; e-mail: [email protected].
Copyright Plenum Publishing Corporation Aug 2002