Research and discussion paper

Some thoughts on cross-cultural research into transgender

Sam Winter, Division of Learning, Development and Diversity, Faculty of Education, University of Hong Kong, Hong Kong

Copyright Sam Winter to whom requests for reproduction and dissemination falling under copyright laws must be made

Contact author

uploaded 2/5/2002 

Gender diversity, like gender development itself, has many faces. In a comprehensive review of the literature on gender development, Ruble and Martin (1998) conceive of the development of gender-typing in terms of a matrix, displayed at the foot of this page. The matrix incorporates four constructs ( ( a ) concepts or beliefs, ( b ) identity or self-perception, ( c ) preferences, and ( d ) behavioural enactment / adoption) each operating upon six content areas ( ( a ) biological / categorical sex, ( b ) activities and interests, ( c ) personal-social attributes, ( d ) gender-based social relationships, ( e ) styles and symbols, and ( f ) gender-related values. The resulting 24 cells represent the range of what we think of as gender-typing. And what a broad range it is! Only two of the cells concerns itself with gender identity, one’s sense of being male, female or indeed a third gender (Cell B1), or desire to be so (Cell B2). Many of the other cells concern themselves with what may be termed gender-related affinities; tendencies to take on the beliefs, attitudes, traits, interests, styles or behaviours generally associated with one or another gender category.

It’s worth remembering that, while transgendered individuals have a cross-gendered identity, many or most of them will also display very strong cross-gendered affinities. Many may retain at least some of the values, beliefs, attitudes, traits, interests, styles or behaviours differentially associated with their original gender (Skrapec and MacKenzie, 1981; Fleming et al., 1984; Cole et al., 1997; Winter and Udomsak, 2002). To that extent these two aspects – identity and affinities - should be considered separate phenomena. Incidentally, Bartlett et al (2000) make a similar point in an article questioning the status of childhood GID as a disorder (Endnote 1).

Since 1999 my own research with Thai transgenders has focused upon a particular aspect of gender-related affinity. My efforts, drawing on a background in child behavioural psychology, have been fuelled by a desire to examine in a relatively systematic, objective and quantitative fashion how TGs perceive themselves as human beings – what traits do they possess, or desire to have, and how do these traits relate to those stereotypically associated with men and women. Seen in terms of Ruble and Martin’s matrix, the research focuses on Cells A3, B3 and C3.

Whenever one gets into this sort of research one comes face to face with a simple yet unavoidable difficulty: to understand a person’s gender-related affinities one must know what are the salient gender-typed beliefs, attitudes, traits, interests, styles and behaviours for that culture and at the present time. Simply going in with an instrument developed elsewhere and at another time just may not work; ideas about gender-typing in the culture and time of origin may not be gender-typed in the culture and time of application.

A nice example of this is provided by the Bem Sex-Role Inventory (Bem, 1974), at one time the most widely used gender-typing instrument, and one of the five most frequently used psychological tests (Robinson et al., 1991). The Bem presents requires that participants survey a list of personality traits, rating each according to the degree to which it describes them. The traits have been carefully chosen to be sterotypically masculine (20 traits seen as particularly desirable for men), feminine (20 viewed as particularly desirable for women) and gender-neutral. From a person’s responses it is possible to assess the degree of masculinity and femininity in that person’s self-concept. You’ll notice that the Bem focuses very specifically on Cell B3 of the Ruble and Martin matrix.

Recent research suggests that the Bem may fail to address the essence of masculinity and femininity in other cultures. Working with a Japanese sample Sugihara and Katsurada (1999) found only four significant sex differences for the 20 masculine items. One of them was in a contrary direction, with women rating themselves higher on ‘analytical’ (a ‘masculine’ item) than do men. There was only one significant sex difference on feminine items. Overall, scores for men and women failed to differ on either the masculine or feminine scales. Should we conclude that Japanese men and women see either themselves or men and women generally as having similar characteristics. Probably not. Using other methods Williams and Best have obtained findings that suggest otherwise (1990a, p.123 and 1990b, p251). Rather, it is likely that notions of what it is in to be masculine or feminine that apply in the US simply do not apply in Japan.

Again, Zhang et al. (2001), this time working with a Chinese sample, and employing a loose standard for significance (p<0.05), found significant sex differences on only six of the 20 ‘masculine’ items, and only nine of the feminine items. Strikingly, four of these nine differences were in an unexpected direction; that is women scored lower than men on ‘affectionate’, ‘flatterable’, ‘warm’ and ‘loves children’, despite the fact that these were all nominally feminine items! We can therefore conclude that, whatever the basis for masculinity and femininity may be in China, the Bem does not tap it.

Incidentally, the cross-cultural diversity that plagues the Bem is also evident where other instruments are employed. See the work of Williams and Best (1990a and b) employing the Adjective Check List (ACL) and Sex Stereotype Measure (SSM). For example, I have done a close analysis of Williams and Best’s international data on gender-trait stereotyping (for gender-normative persons, please note, and displayed in 1990b, Appendix A). The analysis is summarised in Table 2. You will notice from the correlations that Thai gender-trait stereotypes are less than identical to those found in the U.S.A.. Indeed, to the extent that the pooled international data correlates rather less with the Thai sample than with the US sample, we can see that, at least compared with the USA, Thailand deviates to a greater degree from whatever frail international norms exist.

All this ACL data merely serves to reinforce the point that cross-cultural differences exist, that they need to be taken account of in the very way that one approaches the task of doing cross-cultural research, and that if not these differences can disrupt the operation of even the most carefully designed psychological instrument; as certainly as humidity can disrupt a video camera.

So much for cross-cultural differences. What about the effects that time can wring? To return to the Bem, it seems that the instrument may not even be appropriate for US culture any more! Zhang et al. (op.cit.) found significant sex differences on only seven of the male items and thirteen of the female items (albeit that all the differences are in the right direction). Whatever masculinity and femininity may mean in US culture, the Bem seems less able to identify it. This conclusion is echoed by Auster and Ohm (2000). They looked at the desirability of the Bem traits; recall that differential desirability for men and women being the basis for trait selection when the Bem was designed. They found that differential desirability remained for eighteen of the feminine items, but only for eight of the masculine items. We can conclude that the ‘masculinity’ ideal has changed a lot over the last quarter century.

Similarly, in our own research we have found that Thai TGs and their contemporary gender-normative compatriots correlate much higher on gender-trait stereotypes than do those same Thai TGs with their compatriots of the eighties reported in Williams and Best (1990b). The figures are **** and 0.767 respectively. Once again we find evidence of changing ideas about gender-type.

How then can a cross-cultural researcher avoid the problems posed by cross-cultural differences. By ensuring that, as well as collecting data on an individual’s gender-related affinities, he/she also identifies what are the salient characteristics contemporarily found in that culture. Our own research into gender-typing in self-concept represents an attempt to do that.

We have made use of the bank of 300 personality traits contained in Gough and Heilbrunn’s (1980) Adjective Check List (ACL). We rather like the instrument. It is versatile, easy to administer, and there is already a bank of international research into gender-trait stereotype that has employed it as a research tool (Williams and Best, 1990b).

Our own approach has been to use the checklist three times with each TG respondent, first asking her to check traits that describe her as she is, next that describe her as she would like to be, and finally according to whether she believes the traits are more closely associated with males or females. Note again that this sort of study focuses on cells 3B (actual self-concept), 3C (ideal self-concept) and gender-trait stereotypes (3A) in Ruble and Martin’s matrix.

The advantage of this approach (concurrent collection of data on respondents’ self-concept and gender-trait stereotypes) has made possible the study of self in the context of the respondents’ own gender-trait stereotypes, and to do so to a degree that would not be possible were we simply applying an instrument developed for other populations in other places at other times.

Incidentally, our research has proved highly informative. It has revealed that most Thai TGs display either an androgynous or non-differentiated self-concept. Very few display (or pursue) a stereotypically female self. Indeed, almost all aspired towards stereotypically ‘male’ qualities of resourcefulness and dependability and away from stereotypically ‘female’ qualities of caring and social harmony. Quite clearly the slavish pursuit of the female stereotype is very rare among our TGs indeed. Instead, they appear to pursue some carefully considered goals for personal growth that transcend common conceptions of MtF gender transition, and indeed gender itself.

More information about our research is available in Winter and Udomsak (2002, submitted for publication, and in preparation)

Arguably the approach I have described is still flawed. After all, the ACL was compiled in the West some time ago, and for a purpose other than the identification of gender-trait stereotype. Notwithstanding that some traits may discriminate between male and female stereotypes quite well for Thailand, there may be even more salient ones (traits that are considered important aspects of ‘femaleness’, ‘maleness’ or indeed of humanity itself) that are absent from the checklist (full version or any sort-form used) and which therefore do not enter the research.

I think it pretty unlikely, given number and range of traits that are included in the ACL, that important traits might be missing. However, for the researcher concerned about this the way forward may be first of all to obtain from respondents themselves (or their peers in a pilot study) a list of those traits that they consider important characteristics of males, females or indeed humanity, and then to use them as the basis for the research instrument. The essential thing is that the instrument (whichever it is and however it was developed) is used in a way that enables us to examine respondents’ beliefs about themselves in relation to their own beliefs about men and women, not those of people in a far off place and (in today’s terms) long ago. That is the main message of this paper.

End notes

( 1 ) Bartlett et al argue that cross-gender behaviours and cross-gender identity are two entirely different things. They point out that the overlap between these two aspects of gender-typing is, at best, moderate. In a paper entitled ‘ Is Gender Identity Disorder in Children a Mental Disorder?’, they question whether cross-gender behaviours either ( a ) cause current distress or disability, ( b ) lead to increased risk of future suffering, death, pain, disability, or loss of freedom, or ( c ) represent deviance as distinct from a conflict between the child and society. They bemoan the fact that, under DSM-IV guidelines, a child can, given the display of sufficient cross-gender behaviours, be diagnosed as GID without ever himself / herself stating that he / she would like to be a member of another gender. They strongly recommend a revision of the diagnostic criteria for childhood GID.


Auster, C. and Ohm, C. (2000). Masculinity and Femininity in Contemporary American Society: A re-evaluation using the Bem Sex-Role Inventory. Sex Roles, 43, 7-8, 499-528.

Bem, S. (1974). The Measurement of Psychological Androgyny. Journal of Consulting and Clinical Psychology, 42, 2, 155-162.

Cole, C.M., O’Boyle, M., Emory, L.E. and Meyer, W.J. (1997). Comorbidity of gender dysphoria and other major psychiatric diagnoses. Archives of Sexual Behavior, 26, 13-26.

Bartlett, N., Vasey, P. and Bukowksi, W. (2000). Is gender identity disorder in children a mental disorder? Sex Roles, 43, 11/12, 753-785.

Fleming, M.Z., MacGowan, B.R. and Salt, P. (1984). Female-to-male transsexualism and sex roles: self and spouse ratings on the PAQ. Archives of Sexual Behavior, 13, 51-57

Gough, H.G. and Heilbrunn, A.B. (1980). The Adjective Check List Manual. Palo alto, CA: Consulting Psychologists Press

Robinson, J., Shaver, P. and Wrightsman, L. (1991). Measures of Personality and Social Psychological Attitudes. San Diego, USA: Academic Press.

Ruble, D., and Martin, C. (1998). Gender development. In Damon, W. and Eisenberg, N. (Eds.) Handbook of Child Psychology, Vol. 3 (Social, Emotional and Personality Development), New York: Wiley,

Sugihara, Y. and Katsurada, E. (1999). Masculinity and Femininity in Japanese Culture: A pilot study. Sex Roles, 40, 7-8, 635-646

Skrapec, C., and MacKenzie, K.R. (1981). Psychological self-perception in male transsexuals, homosexuals and heterosexuals. Archives of Sexual Behavior, 10, 357-370

Williams, J.E. and Best, D.L. (1990a). Sex and psyche: Gender and self-concepts viewed cross-culturally. Newbury Park, CA: Sage Publications.

Williams, J and Best, D. (1990b). Measuring Sex Stereotypes: A Multi-nation study. Newbury Park, CA: Sage Publications

Winter, S. and Udomsak, N. (2002). Male, Female and Transgender : Stereotypes and Self in Thailand. International Journal of Transgenderism, 6, 1.  link to resource

Winter, S. and Udomsak, N. (submitted for publication) Gender Stereotype and Self among Transgenders: Underlying Elements.

Winter, S. and Udomsak, N. (in preparation) Heterogeneity in Transgender: a cluster analysis of a Thai sample.

Zhang J., Norvilitis, J. and Jin,S.H. (2001). Measuring Gender Orientation with the Bem Sex Role Inventory in Chinese Culture. Sex roles, 44, 3–4, 237-251


Table one: Ruble and Martin’s matrix of gender-Typing: constructs by content (all entries are examples)




Concepts or beliefs


Identity or self-perception

C. Preferences

D. Behavioural Enactment, Adoption


Biological / Categorical Sex

A1. Gender labeling and constancy

B1. Inner sense of maleness or femaleness, or self-perception of masculinity of femininity

C1: Wish to be male or female

D1. Displaying bodily attributes of one’s gender (e.g. clothing, body type, hair), transvestism, transsexualism


Activities and interests: toys, play activities, occupations, household roles, tasks

A2: Knowledge of gender stereotypes or concepts about toys, activities etc

B2: Self-perception of interests

C2: Preference for toys, games etc.

D2. Engaging in play, activities, occupations, or achievement tasks that are gender typed.


Personal-social attributes: personality traits, social behaviours, and abilities.

A3. Concepts about gender stereotypes of personality or role-appropriate social behaviour

B3: Perception of own traits and abilities (e.g. on self rating questionnaires)

C3. Preference or wish to have certain attributes

D3. Displaying gender-typed traits (e.g. aggression, dependence) and abilities (e.g. math)


Gender-based social relationships: sex of peers, friends, lovers, preferred parent, models

A4: Concepts about norms for gender-based relations

B4: Self-perception of own patterns of friendships, relationships or sexual orientation.

C4. Preference for friends,. Parents, and models, or judgments of popularity based on gender

D4. Engaging in social activity with others on the bias of sex or gender (e.g. same sex peer play)


Styles and symbols:

Gestures, speech patterns, play styles, fantasy

A5: Awareness of gender-related symbols or styles

B5: Self-perceptions of non-verbal, stylistic characteristics

C5. Preference for symbolic or stylistic objects or personal attributes

D5. Manifesting gender-typed verbal and nonverbal behaviour or fantasy

6. Gender-related values.

A6: Knowledge of greater value attached to one sex of gender role than the other

B6: Biased self-perceptions associated with group identification

C6. In group/out-group biases: prejudice, attitudes towards egalitarian roles

D6. In-group/out-group identification

Table two. Correlations between gender-trait stereotypes for the U.S., for Thailand and for pooled international data (Williams and Best, 1990b data)

U.S. v Thailand                                         0.872

U.S. v pooled international data                 0.975

Thailand v pooled international data            0.892