Psychological Self-Perception in Male Transsexuals, Homosexuals, and Heterosexuals1 Candice Skrapec, MSc.,2 and K. R. MacKenzie, M.D., F.R.C.P.(C)2,3
Gender-related aspects of self-perception were explored for 24 anatomical males consisting of three matched groups of transsexuals, heterosexuals, and homosexuals. MacKenzie's Diagnostic Criteria Scale ratings were used to confirm group membership. Instruments used were the Repertory Grid Technique, the Rosenberg Self-Esteem Scale, Bem Self Role Inventory, Maferr Inventory of Masculine Values, and Derogatis Sexual Functioning Inventory. Results for transsexual subjects reflected lowest self-esteem. Of the three groups, they perceived themselves to be the most like females and the most unlike other males. Transsexual subjects tended to describe themselves in nonmasculine terms and valued their gender role orientation. Homosexual subjects, on the other hand, reported the highest self-esteem. Additionally, they saw themselves as the most similar to males and the most dissimilar to females. As a group, homosexual subjects described themselves in comparatively strong masculine-stereotyped fashion and valued this posture. However, they believed that women do not value the self-orientation which these subjects endorsed for themselves. Heterosexual subjects scored moderately high in terms of global self-esteem. They described themselves as somewhat masculine to androgynous and reported valuing their gender role orientation. Results on the Repertory Grid and the Derogatis Sexual Functioning Inventory showed heterosexual subjects to be the least polarized in their gender-related self-descriptions. This group generally appeared to be least concerned with traditional gender role referents. The importance of a cognitive approach to clinical assessment of the transsexual individual is stressed. KEY WORDS: transsexualism; identification (psychological); gender identity; self-concept; self-perception; self-assessment; gender role; sexual identity. 1Some of the material in this article was presented at the 28th Annual Meeting of the Canadian Psychiatric Association, Halifax, 1978 and the Sixth International Symposium on Gender Dysphoria, San Diego, 1979. 2Department of Psychiatry, Faculty of Medicine, The University of Calgary, Calgary, Alberta, Canada T2N IN4. 3To whom correspondence should be addressed. INTRODUCTION The University of Calgary has an established Gender Dysphoria Clinic with a sizable caseload of applicants for sex reassignment surgery. The general approach of the clinic, along with results of its assessment and treatment procedures, has been presented elsewhere (MacKenzie, 1973; MacKenzie et al., 1977). This investigation measures selected gender-related variables forming part of the complex network of self-referent dimensions subsumed under the general heading of self-concept. The theoretical orientation of this study views self-concept as a composite of hierarchically organized interacting subsystems (Epstein, 1973). A core of self-esteem is taken as the focus of one primary subsystem (MacKenzie, 1979). That is, a basic sense of self- evaluation both affects and is affected by interaction of the self with specific others in various contexts. The outcome of this interaction, what one observes as behavior, will depend on the particular situation. Core self- esteem is thus a global conception providing higher order organization. With cognitive dissonance overriding the interaction, core self-esteem tends to have a peremptory quality in evaluation of contextual meaning. Similarly, core gender identity is considered the focal point of another primary subsystem (Stoller, 1968; Green, 1974; Money and Ehrhardt, 1972). It interacts with gender-related dimensions such as gender role orientation, sex object choice, and the sexual implications of physical characteristics and psychological constitution. The resultant gender-related behaviors observed are thus a product of boundary resolution between the higher order sense of core gender identity and the lower order specific manifestations. It is assumed that individuals who are clear about their gender identification draw referents from an organized system of beliefs as to the psychosexual meaning of being a male or female. That is, being a male or female implies a set of beliefs about appropriate gender roles, sexual preference, psychological makeup, and physical appearance. As a developmental process, individuals incorporate the resultant composite into their sense of self. Evaluative processes accompany this integration. The transsexual individual perceives his or her gender identity to be incongruous with the apparent anatomical reality, thus giving rise to boundary stress between core gender identity and physical characteristics. The purpose of this study was to explore how gender issues impact on several dimensions of self-perception METHODOLOGY Subjects Eight male transsexuals requesting sex reassignment surgery to female were selected from the Gender Dysphoria Clinic of the University of Calgary. Histories and assessment summaries were reviewed in order to rule out clear indications of psychopathology. Specifically, information was derived from available psychological testing reports (including the MMPI and social and psychiatric history questionnaires) and clinical interview summaries. Eight volunteer heterosexual and eight volunteer homosexual males matched for age and socioeconomic status with the transsexuals were also studied. Sample characteristics are summarized in Table I. Table I. Sample Characteristics Transsexual Heterosexual Homosexual Age range 22-32 21-30 25-35 Mean age 26.8 23.6 29.6 Social classa frequencies I 1 2 1 II 0 0 1 III 3 1 3 IV 2 4 2 V 2 1 1 aFrom Hollingshead's Two Factor Index of Social Position (Hollingshead, 1957); a brief class description is as follows: I. High social prestige (wealth). II. High managerial positions ("well-to-do"). III. White-collar status. IV. Semi-skilled positions. V. Unskilled labor (poor). None of the 24 subjects gave a history of involvement with organizations which might be expected to render a major selection bias to this study (e.g., homophile societies). Additionally, all appeared to be responsible members of the community, maintaining stable employment without attracting political or personal notoriety. It should be noted that data for one homosexual subject were incomplete. Analyses were adjusted accordingly. Instruments and Techniques 1. MacKenzie's (1978) Diagnostic Criteria Scales are based on clinical histories. They are used to quantify aspects of sexual history along the following dimensions: (1) childhood cross-gender behavior; (2) adult cross- gender behavior; (3) homophile association; (4) heterophile association; (5) genital erotic focus; and (6) transvestite behavior. 2. Repertory Grid Technique offers an intriguing approach to the operationalization of gender identification as an aspect of self- identification. The technique is derived from Kelly's (1955) theory of personal constructs. The element components used are summarized in Table II. These comprise names of actual people from the subject's life as well as 10 idealized figures. There were 25 constructs obtained in the usual fashion by eliciting descriptors used to differentiate triads of element figures. Each element was then rated along a 5-point scale on each construct. The resultant grid was analyzed in terms of the interelement distances. This calculation constitutes a measure of similarity between elements (Lederman, 1954; Giles and Rychlak, 1965; Ryle, 1975). Table II. Nature of Grid Elements Element number 1 \ 2 \ (7 significant males from subject's personal life) (7 significant females from subject's personal life) 13 / 14 / 15 ideal example of a woman 16 poor example of a woman 17 ideal example of a man 18 poor example of a man 19 actual self 20 ideal self 21 women's perception of self as female 22 men's perception of self as female 23 men's perception of self as male 24 women's perception of self as male 3. The Rosenberg Self-Esteem Scale is a 10-item Guttman scale to measure self-evaluation in global terms (Rosenberg, 1965). 4. The Bem Sex Role Inventory was administered to determine subjects' self-perceptions according to gender role stereotypes. This scale measures an individual's affinity for various constellations of gender role attributes. Masculinity and femininity are treated as independent dimensions. This permits characterization of individuals as masculine- or feminine-gender typed; androgynous (representing coexistence of masculine and feminine attributes), or undifferentiated (implying gender role postures low in both masculinity and femininity) (Bem, 1974, 1977). 5. The Maferr Inventory of Masculine Values was used as a means of quantifying attitudes toward masculine roles (Steinmann et al., 1968). Subjects described "self" and "ideal man" perceptions according to family- versus self-oriented values. For the family-oriented man, family responsibilities take precedence over personal endeavors. A self-oriented man, on the other hand, considers personal satisfactions as more important than those of family. The family- versus self-orientation distinction parallels an expressive/instrumental dimension which is generally taken to distinguish gender role stereotypes. 6. The Derogatis Sexual Functioning Inventory provides a composite psychological profile with particular relevance for sexual behavior (Derogatis, 1976). This instrument has specific subscales related to sexual information, experience, drive, attitudes, psychological symptoms, affect, gender role definition, sexual fantasy, body image, and general sexual satisfaction. Procedure All subjects were seen individually on two occasions separated by approximately three weeks. They were told at the outset that the purpose of the study was to explore attitudes about sexuality and how these might be incorporated into a perception of self. Confidentiality of information was stressed. It was made clear to members of the transsexual group that results would in no way influence gender program decisions. Except for Repertory Grid Technique administration, which occupied both sessions, the order in which instruments were presented across the two sessions was randomly determined. RESULTS Results of the Diagnostic Criteria Scale ratings are presented in Table III. Pairwise comparisons revealed statistically significant differences across subject groups in expected directions. Repertory Grids for each subject were analyzed using Slater's INGRID 72 program (Slater, 1977). The output includes distances between elements in terms of their similarity or dissimilarity to each other, according to the elicited constructs. A comparison was made between a subject's description of himself and mean descriptions for both males and females. That is, did he describe himself more like "a man" or more like "a woman"? Table IV presents these results. Heterosexual subjects saw themselves as equally similar to males and females. According to the subjects' own descriptions of males and females, transsexual subjects describe themselves as more like females, whereas, homosexual subjects saw themselves most like males. Table III. Comparison of Diagnostic Criteria Scale Ratings across Groups: Obtained Student's t Values Crossgender Homo- Hetero- Genital Trans- behavior phile phile Erotic vestite Child Adult index index focus behavior Transsexual vs. Heterosexual 3.97b 13.05b 3.13b 3.31b 2.15a 1.89a Transsexual vs Homosexual 2.71b 10.21b - 1.74a -1.06 -6.63b 1.89a Homosexual vs. heterosexual 2.14a 1.34 7.94b 2.05a 3.26b 0 ap < 0.05. bp < 0.01. Table IV. Distance between Elementsa Significance (Student's t, Self vs Females Self vs Males within groups) Transsexual subjects 0.898 1.002 -> p < 0.05 Heterosexual \ subjects 0.908 0.985 N.S. -> p < 0.05b Homosexual / subjects 1.012 0.910 -> p < 0.01 aResults are expressed as mean standardized element-element distances. Values less than 1.0 imply similarity between elements, and values greater than 1.0 reflect dissimilarity. bBetween groups. Table V. Self-Evaluation Indices Mean Rosenberg Scorea Actual-ideal distanceb Transsexual subjects / 1.38 \ / 0.650 p < 0.01 \ N.S. Homosexual \ \ \ subjects / 1.13 p < 0.01 / 0.683 N.S. p < 0.01 (0.86) / N.S. (t-test) Heterosexual \ / \ subjects 0.75 / 0.765 aLow scores reflect high global self-esteem. bStandardized element-element distances; values less than 1.0 imply similarity between elements. An index of self-evaluation was obtained in two ways: (1) using the Rosenberg Self-Esteem Scale as a global measure; (2) using the concepts of actual versus ideal self-descriptions relative to the self-generated construct systems of the Repertory Grid. Results are shown in Table V. Scores on the Rosenberg Self-Esteem Scale clearly discriminated transsexual subjects from heterosexual and homosexual ones. Homosexuals thought highest of themselves in global terms, and transsexuals scored lowest; heterosexuals fell in the middle. All comparisons reached statistical significance. However, sample size was small, and one heterosexual individual scored near maximum in terms of self-dissatisfaction on the Rosenberg Scale. If his score is removed (as indicated in parentheses in Table V), then heterosexual and homosexual groups are not significantly different from each other. No significant differences were found between actual - ideal self - ratings on the Repertory Grid. Thus, these two measures, both purporting to tap self-evaluation, appeared to give different results. To analyze this discrepancy, Rosenberg Self-Esteem Scale scores were correlated with actual-ideal-self grid ratings. When scores for all three groups were collapsed, a Pearson product moment correlation coefficient of 0.46 was obtained. This indicated that 80% of the total variance between the two sets of scores could not be accounted for by a linear relationship. Each subject group was then looked at through separate comparisons. For heterosexual subjects, a significant correlation was found between Rosenberg scores and actual-ideal-self distances (r = 0.88, p < 0.01). This difference remained even when data for the extreme scores on the Rosenberg Scale were omitted from analysis. No such relationship was revealed for transsexual (r = 0.10) or homosexual (r = 0.06) subjects. That is, transsexual and homosexual self-descriptions on the Repertory Grid tended not to match their more global self-evaluations on the Rosenberg Self-Esteem Scale. Both the original t-ratio scoring and the more recently recommended median- split procedures were used to determine Bem Sex Role Inventory results (Bem, 1974, 1977). Using t-ratio scoring, transsexual subjects described themselves either as gender typed in a feminine direction or as androgynous (mean = 0.58, standard deviation = 1.18). Only one of these subject's scores implied slight masculinity. The mean of the heterosexual group was in the androgynous range (mean = 0.65, standard deviation = 2.08). One heterosexual subject scored in a highly feminine direction. More than half of the homosexual subjects scored in a masculine direction, with two describing themselves in more feminine terms (mean = 1.53, standard deviation = 3.82). Scores for homosexual subjects revealed the greatest variability. When the same data were analyzed using a median-split procedure, heterosexual subjects tended to score low in both masculine and feminine characteristics (undifferentiated) relative to the overall sample. In contrast, five of the eight transsexual subjects scored high on both masculine and feminine characteristics, implying androgyny. Only one transsexual individual could be classified as undifferentiated. Homosexual subjects' results revealed even more marked polarity, with six of the seven scoring in a masculine-gender-typed direction and the seventh subject high in both directions. Table VI shows the classification of subjects by gender role endorsement according to the median-split procedure. Subjects' attitudes toward masculine roles were examined by analyzing the data from the Maferr Inventory of Masculine Values. The same questions based on family- versus self-oriented values were answered for (1) self-description; (2) subject's perception of the values held by the ideal man; and (3) subject's perception of the values women associate with the ideal man. Subjects were first compared on the basis of their own personal values. Homosexuals were more self-oriented than heterosexuals, who, in turn, were more self-oriented than transsexual subjects. Using the same instrument, subjects described their conceptions of the ideal man. Homosexual subjects felt that the ideal man values self-orientation more than did heterosexual or transsexual subjects. Finally, subjects were asked how they thought women would describe values of the ideal man. Transsexuals again presented the ideal man in a family-oriented direction. Heterosexuals rated him as moderately self-oriented. However, results for homosexuals were markedly different; they perceived women's ideal man to be highly family oriented. Group means are presented in Table VII. Table VI. Bem Sex Role Inventory Results: Median-Split Procedure (Number of Subjects) Masculine Feminine Androgynous Undifferentiated Transsexuals 0 2 5 1 Heterosexuals 3 1 0 4 Homosexuals 3 3 1 0 Table VII. Maferr Inventory of Masculine Values Results, Mean Role-Value Scoresa Self-description Subjects' Subjects' ideal perception of man women's ideal man Transsexuals +0.13 - 0.63 - 7.00 Heterosexuals +5.25 p < 0.05 +1.38 +1.25 (t-test) p < 0.05 Homosexuals +9.43 +7.43 -10.43 (t-test) at-Test significance probabilities for transsexual compared to homosexual groups. Table VIII. Derogatis Sexual Functioning Inventory: Mean T-Scores Subscale Transsexuals Heterosexuals Homosexuals I. Information 48 43 47 II. Experience 42 41 47 III. Drive 42 51 64 IV. Attitude 39 48 54 V. Symptoms 64 64 69 VI. Affect balance 49 48 42 VII. Gender role definition 36 49 56 VIII. Fantasy 46 54 61 IX. Body image 29 43 38 X. Satisfaction 44 45 49 Maferr Inventory of Masculine Values self-description results were consistent with Bem Sex Role Inventory data. The Pearson product moment correlation coefficient for transsexual subjects was -0.65 (p < 0.05), for heterosexual subjects, -0.50 (N.S.), and for homosexual subjects, -0.63 (p < 0.05). Results for subject groups on the Derogatis Sexual Functioning Inventory are presented in Table VIII. On seven of the ten subscales, heterosexual subjects scored between transsexual and homosexual subjects. Summarizing these data, homosexual subjects reported higher sexual drive levels than did transsexual subjects, with heterosexuals in the middle. Homosexual subjects were most liberal-minded, and transsexuals, least. Homosexuals most strongly endorsed masculine traits for themselves, in contrast with transsexuals who revealed the most femininity in terms of gender role definition. Homosexual subjects reported entertaining the widest range of fantasy activities, and transsexual subjects, the most limited. Body images of transsexuals were poorest. This group also scored lowest on general sexual satisfaction. DISCUSSION By clinical history and the Diagnostic Criteria Scales, the three groups appeared to be relatively uncontaminated with each other. Differences between groups across all six scales were consistent with clinical expectations. In particular, profiles for transsexual subjects were compatible with a classic transsexual history. This group is therefore representative only of one subcategory of patient presenting at gender dysphoria clinics (Bentler, 1976). In looking at the results across instruments, it becomes apparent that there was a considerable degree of consistency for each of the subject groups. From Repertory Grid data, it was found that transsexual subjects saw themselves as more like females than males. It may be that they have incorporated a female set of attitudes regarding themselves. On the Rosenberg Self-Esteem Scale, these individuals reported having the lowest global self- esteem. However, there was no linear relationship between Rosenberg Scale scores and actual-ideal-self distances on the grid. Thus, transsexuals globally evaluated themselves more critically than would be predicted from their self-ratings. This suggests that in seeking sex reassignment surgery, they are responding to an abstract idea of gender, more than to actual self- perception. The presenting problem of the transsexual is one of sexual dysphoria more than of a particular self-description. Results suggest that the basis for transsexualism is a higher order, abstracted sense of gender, rather than a function of actual behavioral descriptions. This is compatible with Stoller's conceptualization of core gender identity which goes beyond behavioral characteristics. Such an abstract sense of maleness/femaleness may serve as the self-evaluative determinant for the transsexual. Moreover, Maferr instrument results reflected that transsexuals subjects valued their chosen gender role orientation. On the Bem Sex Role Inventory, these subjects used extremes in their endorsement of masculine and/or feminine characteristics for themselves. On the Derogatis Sexual Functioning Inventory, they scored as the least liberal-minded group. Their endorsement of rather polarized conservative values is in keeping with their endorsement of traditional gender role referents. In general, the level of involvement in and satisfaction with sexual behavior tended to be lowest for this group. As would be expected, transsexuals had the poorest body image. Results for heterosexuals tended to contrast with those for transsexuals, and homosexuals. Referring to Repertory Grid distances, these subjects saw themselves as equally similar to male and female others. They reported moderately high global self-esteem. Further, their global self-evaluations were congruent with actual versus ideal self-ratings on the Grid. For these individuals, responses to the two measures of self-evaluation would seem to have been derived from a single cognitive impression of self. The correlation between Rosenberg Scale scores and actual-ideal-self Grid ratings for this group approximates that determined by Silber and Tippett (1965) for college students. Relative to transsexual and homosexual subjects, heterosexuals described themselves in comparatively less polarized terms according to androgyny scores. The consistency of value ratings between self and ideal-man perceptions would indicate that heterosexual subjects valued their own gender role orientation. Results on the Derogatis instrument presented these individuals as the least polarized on seven of the ten subscales. Overall, general sexual satisfaction was highest in this group. Heterosexual subjects appeared to be moderately liberal in attitude. Therefore, the fact that they presented as androgynous and not "masculine" on the Bem Scale scores should not evoke psychosexual conflict. As a group, homosexuals sketched an interesting but puzzling picture of themselves. In terms of self versus other grid distances, it was expected that they would see themselves somewhat further removed from their descriptions of other males than would heterosexual subjects, particularly since a considerable proportion of the male elements rated by homosexual subjects were heterosexual figures. Instead, homosexual subjects described themselves as being more like other males than heterosexual subjects did. Homosexual subjects reported feeling better about themselves in a global sense than their self-ratings from the grid would suggest. They seem to have described themselves using different referents than they used in evaluating themselves. As for gender role orientation, homosexual androgyny scores were the most polarized and displayed the greatest variability. They reported valuing self- over family-orientation and felt this was also true of the ideal man. However, they perceived women's ideal man to strongly value a family orientation. This contrasted sharply with the self-oriented terms endorsed in describing their own values and those of their ideal man. As such, there was a discrepancy between the gender role behavior they acknowledged and valued for themselves, and those they believed were valued by women. This suggests a major incongruity in their perception of male versus female expectations regarding male behavior. Cognitive dissonance could be evoked in situations where social expectations challenge self-perception. In particular, such incongruity could conceivably have negative implications for psychosexual stability. It certainly could shape the behavior of the homosexual male as he interacts with women. In terms of sexual functioning, homosexual subjects presented a profile suggestive of considerable involvement and satisfaction. They were the most liberal-minded regarding sexual attitudes. By adopting a liberal attitudinal set, they may be able to reduce any cognitive dissonance generated by their gender role orientation and sex object choice. Certainly, the sample size of 24 imposes important restrictions on the generalizability of results to other members of the three populations represented in this study. Nonetheless, the three groups were matched for age, sex of birth, and socioeconomic status, and statistically significant differences were found. One might also note that groups did not differ in terms of affect balance or psychological symptoms on the Derogatis Scales. Moreover, patterns of obtained results were compatible with logical extensions to psychosexual stability. To summarize, the results indicated that transsexuals tend to describe themselves in nonmasculine terms and value their gender role orientation. As sex-reassigned females, it would be reasonable to expect smooth psychosexual adjustment. Indeed, sex reassignment surgery may be viewed as a means of enhancing the consistency of the gender component of self-concept. Heterosexuals were least polarized in terms of gender-related measures. They appeared least concerned with traditional frames of reference and, further, endorsed substantial balance in their gender role. For them, integration of gender component elements into the self-concept system is not likely to meet with great conflict. Homosexuals described themselves in more rigidly stereotypic fashion. One might suspect that the gender identification component of self-concept was most fragile in this group of individuals. The data suggests that they introduced a systematic distortion into their perceptions of maleness and male roles. One possible explanation could be the systematic use of denial and reaction formation. The results could be explained by using the idea of compensatory masculine responding, where masculinity is defensively exaggerated in the face of gender role "threat" (Babl, 1979). Such a response would be understandable in situations where the homosexual's wish for enduring interpersonal relationships is met by a reality of brief, more superficial encounters. In a sense then, his global self- evaluation is "over-determined," with some kind of compensation operating at a cognitive level. If further research supports these results, gender dysphoria clinics may derive additional means for discriminating between effeminate homosexual and transsexual candidates. It should be noted that, to our knowledge, the Repertory Grid Technique has not been systematically used to assess gender issues. Our results, in which comparison is made with three established instruments, may be taken as supporting the construct validity of this approach. To conclude, it is felt that the importance of cognitive mediating variables in relation to the interpersonal construct system used for describing and evaluating self and others must be stressed. In terms of a self-concept system, the presenting problem of the transsexual may be appreciated as a symptom of the need for boundary resolution within this system. The theoretical framework underlying this research would suggest that psychosexual stability is determined by the efficacy of such resolutions. ACKNOWLEDGMENT We thank Dr. Patrick Slater and the British Medical Research Council for services in providing the INGRID 72 program. We are particularly grateful to Dr. Slater for his helpful correspondence. REFERENCES Babl, J. D. (1979). Compensatory masculine responding as a function of sex role. J. Consult. Clin. Psychol. 47: 252-257. Bem, S. L. (1974). The measurement of psychological androgyny. J. Consult. Clin. Psychol 42: 155-162. Bem, S. L., and Watson, C. (1976). Revised Scoring Packet: Bem Sex Role Inventory. Available from first author. Bem, S. L. (1977). On the utility of alternative procedures for assessing psychological androgyny. J. Consult. Clin. Psychol. (1976). 45: 196-205. Bender, M. P. (1976). Does construing people as similar involve similar behavior towards them?: A subjective and objective replication. Brit. J. Soc. Clin. Psychol. 15: 93-95. Bentler, P. M. (1976). A typology of transsexualism: gender identity theory and data. Arch. Sex. Behav. 5: 567-584. Derogatis, L. R. (1976). Psychological assessment of the sexual disabilities. In Meyer, J. K. (ed.), Clinical Management of Sexual Disorders, Baltimore, Williams & Wilkins. Derogatis, L. R., Meyer, J. K., and Vazquez, N. (1978). A psychological profile of the transsexual: I. The male. J. Nerv. Ment. Dis. 166: 234-254. Epstein, S. (1973). The self-concept revised: Or a theory of a theory. Amer. Psychol. May: 28: 404-416. Giles, P. G., and Rychlak, J. R. (1965). The validity of the role construct repertory test as a measure of sexual identification. J. Proj. Tech. 29: 7-11. Green, R. (1974). Sexual Identity Conflict in Children and Adults. New York, Basic Books. London, Gerald Duckworth. Hollingshead, A. B. (1957). Two Factor Index of Social Position. New Haven, Connecticut, August B. Hollingshead. Kelly, G. A. (1955). The Psychology of Personal Constructs. New York, W. W. Norton. Lederman, D. G. (1954). Sexual identification on the role construct repertory test. Unpublished doctoral dissertation. Ohio State University. MacKenzie, K. R. (1973). The family context of a case of gender dysphoria. Paper presented at the Third International Symposium on Gender Identity, Dubrovnik, Yugoslavia. MacKenzie, K. R. (1978). Gender dysphoria syndrome: Toward standardized diagnostic criteria. Arch. Sex. Behav. 7: 251-262. MacKenzie, K. R. (1979). The measurement of self concept in psychiatric outpatients: A factor analytic study of the Tennessee Self Concept Scale. Paper presented at the American Group Psychotherapy Association Conference, New York. MacKenzie, K. R., Bala, M., Angus, D., and Love, E. (1977). Endocrine abnormalities in gender dysphoric patients. Paper presented at the Fifth International Symposium on Gender Dysphoria, Norfolk. Money, J., and Ehrhardt, A. A. (1972). Man and Woman, Boy and Girl: The Differentiation and Dimorphism of Gender Identity from Conception to Maturity, Baltimore, The Johns Hopkins University Press. Rosenberg, M. (1965). Society and the Adolescent Self Image, Princeton, Princeton University Press. Rosenberg, M. (1979). Conceiving the Self, New York, Basic Books. Ryle, A. (1975). Frames and Cages: The Repertory Grid Approach to Human Under standing, New York, International Universities Press. Silber, E., and Tippett, J. S. (1965). Self-esteem: Clinical assessment and measurement validation. Psychol. Rep. 16: 1017-1071. Slater, P. (ed.). (1977). The Measurement of Intrapersonal Space by Grid Technique, Volume 2, Dimensions of Intrapersonal Space, New York, John Wiley. Steinmann, A., Fox, D., and Farkas, R. L. (1968). Male and female perceptions of male sex roles. Proceedings of the 76th Annual American Psychological Association Convention, San Francisco, American Psychological Association. Stoller, R. J. (1968). Sex and Gender (Vol. 1), New York, Science House. Wylie, R. C. (1974). The Self-Concept: A Review of Methodological Considerations and Measuring Instruments (Vol. 1, rev. ed.), Lincoln, University of Nebraska Press.