Department of Social Medicine, Sahlgrenska Academy, Göteborg University, Box 453, SE-405 30 Göteborg, Sweden
Received 15 April 2002; in revised form 18 November 2002; accepted 9 December 2002
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Instruments, for example the Bems Sex Role Inventory (BSRI) or the Personal Attributes Questionnaire (PAQ), include self-assessment of masculinity and femininity as a way to operationalize gender identity (Bem, 1975; Spence et al., 1975
). In her earliest studies, Bem found that an androgynous sex-role orientation was associated with higher self-esteem and psychological well being (Bem, 1975
; Ashmore, 1990
; Sorell et al., 1993
). Androgynous persons scored highly on both the masculinity and the femininity scale, which was interpreted as being associated with a broad repertoire of behavioural and psychological coping strategies and personality traits. However, later studies have shown that the masculinity scale mainly can explain these positive effects on self-esteem and psychological well being (Sorell et al., 1993
). The role of femininity for psychological well being has been discussed. In a comprehensive review of 69 studies, Murnen and Smolak (1997)
found a small, but positive, relation between femininity and eating problems, especially anorexia nervosa.
In a review of earlier studies, Sorell et al. (1993) concluded that the results on the relation between alcoholism and sex-role orientation were contradictory. It was shown that women with alcoholism were ultra-feminine, overly masculine, con-fused about sex-role identity or had a masculine approach to life. Beckman (1978)
found in a study with BSRI that women with alcoholism were undifferentiated, i.e. they were low on both the masculinity and the femininity scale. Wilsnack et al.(1985)
found that androgynous sex-role orientation (according to PAQ) was associated with lower levels of alcohol consumption and of drinking-related problems. Finally, Sorell et al. (1993)
found that psychological masculinity (PAQ) was the most important factor in differentiating women with alcoholism from those without. In a study on a European sample, and using a different but related concept, Neve et al. (1997)
found that traditional gender role attitudes were associated with lower alcohol consumption and a higher proportion of abstainers among women.
Several authors have pointed to the need for further elaboration of the association between gender roles, gender identity and alcoholism (Sorell et al., 1993; Neve et al., 1997
). Room (1996)
concluded that Because gender roles and interactions are so often steeped in drinking and drug use, such studies can teach us something more general about the interaction of social forces and norms with mundane and private behavior. But the studies also offer the promise of finding ways forward in the practical task of reducing harm associated with alcohol and drug use (p. 237).
The aim of the present study was to analyse gender identity among women and its association to psychiatric disorders and alcohol consumption in a country that has undergone important changes in gender relations.
![]() |
SUBJECTS AND METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Overall study design
The overall study design of the screening phase is presented in Table 1. Three different cohorts were included in the project: (1) all women born in 1925, -35, -45, -55, -65, and registered for census purposes in the District West of Göteborg on 31 December 1985; (2) a sample of consecutive female visitors to clinical settings such as maternity units, GP units, acute wards and psychiatric clinics, between March and May 1986; (3) all women born in 1970 and 1975 and registered for census purposes in District West or the Central District of Göteborg on 1 January 1995.
|
Out of 1252 interviewed women, 836 fully completed the Masculinity/Femininity-Questionnaire (henceforth M/F-Q) (Bergman et al., 1988), and were thus included in the study.
Screening instrument and stratification procedure
A screening instrument, Screening Women and Alcohol in Göteborg (SWAG), was mailed to the home addresses of the population-based sample and handed out to consecutive arrivals in the clinical sample. SWAG has 13 items and is based on the Short-MAST, ML-MAST and the CAGE (Spak and Hällstöm, 1996). Since the prevalence of alcohol problems is low among women, the stratification method was designed to over-sample women with probable alcohol problems (Spak and Hällström, 1996
).
Interviews
The interviews were conducted after informed consent was obtained. The Ethics Committee of the Medical Faculty of the University of Göteborg has approved the study. Interviews were performed by seven clinicians (six women and one man), all with several years of clinical experience. They were all trained in DSM-III-R, and clinical examinations were performed until sufficient inter-rater reliability was achieved. One of the authors (F.S.) has been a consultant in diagnostic procedures. The interviews focused on socio-demographic characteristics, childhood conditions, family relations, work situation, physical and mental health, alcohol consumption and its adverse consequences, and personality factors (Wilsnack et al., 1991; Spak and Hällström, 1996
). Clinical psychiatric diagnoses were made according to DSM-III-R (axis I, IV and V) (American Psychiatric Association, 1987
). For diagnosis of alcohol dependence and abuse (ADA), the structured diagnostic instrument CIDI-SAM (Robins et al., 1986
) was also included in the protocol.
Masculinity and femininity questionnaire
As an indicator of gender identity in this study, the masculinity and femininity questionnaire was chosen. M/F-Q is based on BSRI and adapted for use in Sweden (Bergman et al., 1988).
The M/F-Q used in this study included 43 statements with four response alternatives: disagree completely, partly disagree, partly agree and agree completely, which are quantified to 1, 2, 3 and 4 points, respectively. Of the 43 statements, 17 items are considered to represent masculine dimensions and 17 items represent feminine dimensions, while nine items are seen as neutral.
All neutral questions were excluded from our analyses, since they are included in the questionnaire only to disturb the response pattern. Furthermore, the items I am masculine and I am feminine were excluded as recommended in earlier studies (Blanchard-Fields et al., 1994) and also because there was almost no variation over the response categories. Finally, a question on the use of swearing was excluded due to its strong correlation to age in this cohort. The final number of statements analysed was 31.
Psychometric analyses
Initially, we were interested in determining whether any effects of ageing were in evidence. Mean and P-values of paired differences (Wilcoxon matched-pairs signed-ranks test), item-specific and at the individual level, were therefore used in order to test for systematic changes between the two measurements (Table 2). All women who had answered the M/F-Q twice were included (n = 302). There were statistically significant changes in four of the 31 items (3, 16, 36 and 39), all related to the masculinity dimension. No statistically significant changes were found for any other item. Thus, there was considerable stability over time.
|
Face validity was discussed among the authors and we concluded that the items included in the M/F-Q represent aspects of identity and personality related to gender. A comparison has also been made between the items used in M/F-Q and a recent Dutch study (Visser, 1996) on contemporary notions on masculinity and femininity, and close resemblance was found.
Factor analysis
A factor analysis with varimax rotation was conducted in order to analyse the properties of the questionnaire in this cohort. Different solutions were tested. If no restriction was introduced, eight factors with an Eigen value above 1 were computed. In some of these, there were few items and the content was similar to other factors. A restriction was therefore introduced and a solution consisting of four factors was computed (Table 2), labelled leadership, caring, self-assertiveness and emotionality. The factors represent four dimensions of gender identity and, based on these dimensions, indices were created for the subsequent analyses. These indices were scored by summing the ratings (range 14) of the items included in each factor and for each person.
Psychiatric disorders and alcohol consumption
The study population was grouped into four categories with regard to psychiatric disorders (Table 3). Women with an alcohol diagnosis, irrespective of which one, were all included in the group Alcohol dependence and abuse. These women could also have other psychiatric disorders. However, in the other three groups, no women with alcohol diagnoses were included. Of the total number included in the study (n = 836), 445 (53%) had never met the criteria for a DSM diagnosis. The distribution of psychiatric disorders is based on a stratified sample and does not represent the distribution in the general population.
|
Logistic regression
Logistic regression was used to calculate odds ratios with 95% confidence intervals and to adjust for age in the analysis of the relation between gender identity and psychiatric disorder.
Attrition
Four types of attrition were noted. One external attrition consisted of those selected for screening who declined participation or did not answer the screening questionnaire. Another external attrition involved those selected for interview who declined further participation or were impossible to reach. The internal attrition consisted of those interviewed by telephone (the M/F-Q was not used) and those interviewed who did not fill in the M/F-Q.
Earlier studies on external attrition have not shown significantly higher proportions of women with ADA in the attrition (Midanik, 1982; Hasin et al., 1990
; Spak and Hällström, 1995
). Furthermore, telephone interviews conducted with women who were reluctant to participate indicated low alcohol consumption and shortage of time as the most common reasons for not participating. These findings do not indicate that the external attrition should differ substantially from the sample that filled in the M/F-Q. A medical record study of the women born in 1925, -35, -45, -55 and -65 did not show a higher proportion of ADA in the attrition group (Spak, 1996
). Thus, we believe our sample to be representative for an analysis of the association between gender identity, operationalized through the M/F-Q, and ADA and psychiatric disorders. We also believe that our findings are possible to generalize to the general population. No analyses were performed to test for differences in gender identity profiles between the study and attrition groups; however, we do not assume that gender identity influenced participation in this study.
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Comparisons between studies are difficult, since different concepts of gender identity were used, different instruments employed, alcohol outcomes were measured in different ways, and the associations probably also varied between time periods and cultures. For example, the study by Lara-Cantú et al. (1990) was performed in a country and time period when, according to the authors, there were double standards regarding drinking for women and men. There were strong restrictions against women drinking and a highly liberal attitude towards drinking by men. In contrast, our study was performed in a country that has undergone great changes in attitudes towards drinking in women, with a much more liberal attitude in recent decades. Such cultural differences influence gender identity according to recent theories and accordingly complicate comparisons over time and cultures (Ashmore, 1990
).
We found an association between bipolar disorders and severe anxiety disorders and low self-assertiveness, low leadership and high emotionality. These were the same dimensions of gender identity as those associated with ADA and, apart from low leadership, with HED. Earlier studies based on BSRI found an association between masculinity and better mental health (Spence et al., 1974; Bem, 1975
; Ashmore, 1990
; Blanchard-Fields et al., 1994
). Our findings are in line with these studies, since most items of the factors self-assertiveness and leadership come from the masculinity dimension of the M/F-Q. However, none of these indices, or the emotionality index, included items solely from the masculinity dimension. These findings speak in favour of using more descriptive components of the different gender dimensions than masculinity and femininity. Based on M/F-Q, factor analysis and theoretical considerations, we found that gender identity consisted of four dimensions, which we labelled leadership, caring, self-assertiveness and emotionality.
When gender identity is analysed in relation to psychiatric disorders, ADA and alcohol consumption, we suggest that a division into different diagnostic groups should be made and that several alcohol outcomes are used. As an example of more detailed information, we did not find any obvious pattern or association regarding gender identity and minor anxiety and depressive disorders, contrary to what we had expected. Neither did we find an association between any of the gender identity dimensions and high alcohol consumption.
As alcohol consumption measures often are used as the outcome variables in studies of alcohol problems, it is important to study the relation of gender identity to HED and HAC as well as to diagnostic concepts, e.g. ADA. Of the two alcohol consumption measures, the associations between HED and the different dimensions of gender identity showed the strongest similarities with those found for gender identity and ADA. HED may thus be a better predictor of risky outcome than HAC. However, women with HED had less extreme values on the different dimensions of gender identity, compared to women with ADA. We believe that these findings provide support for the use of strictly defined diagnoses such as ADA, rather than consumption measures, when trying to understand alcoholism clinically.
Methodological considerations
Construction of a mixed study sample. There are three main concerns over the inclusions of various samples in this study base, as well as collecting interviews from different occasions of the respondents participation. First, there was an over-sampling of women with alcohol problems in all three samples, and thus also of women with somatic illnesses or various other reasons for attending treatment settings. The effect of this over-sampling was somewhat reduced as the largest participant groups were recruited at the GP units and the maternity units, in connection with the observation that both of these groups contained comparatively few persons with ADA. Secondly, there is the problem of time of entry, i.e. at which state of participation the women were included in this study. They could be entered either from the first interview occasion or from the follow-up interview 5 years later. It is possible that the answers to the M/F-Q change with increasing age, albeit this hypothesis was not supported by our reliability test. Thus, it appears plausible to include subjects from various stages of the study. Thirdly, we feel that it was justified to include subjects from the different samples, as we studied associations, and not rates, in a certain population.
Self-reported data. Our results are based on self-perceived gender identity which might lead to a bias, since the role of social desirability might be stronger among women who were healthy, compared to those with psychiatric disorders. On the other hand, the majority of the women with psychiatric disorders had good social functioning and were well adapted to society (Hensing et al., 1997). Apart from the previously discussed possible differences between the age groups (for example, younger women might be more eager to present themselves as having leadership abilities), we do not believe that the effect of social desirability in answering the M/F-Q was different across the different stratification groups.
Discussion on causal directions is limited in a cross-sectional study. This study is cross-sectional and thus we do not know anything about the causal directions. For example, the analysis cannot tell us whether, say, lacking self-assertiveness leads to mental health problems/alcohol problems or vice versa. It might be assumed that women with psychiatric disorders become, as a consequence of their disorder, less gifted for leadership, low in self-assertiveness and become more emotional. However, more empirical research with a prospective and longitudinal design is needed regarding gender identity development and the course of alcoholism and psychiatric disorders respectively, as well as in combination.
In conclusion, we found that different dimensions of gender identity were associated with psychiatric disorders, ADA, HED and HAC in different ways. More effort should be made to study the concept of gender identity, preferably broken up in more detailed indices. Our findings need to be tested also in longitudinal studies to gain further knowledge on causal directions.
![]() |
ACKNOWLEDGEMENTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
FOOTNOTES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Ashmore, R. (1990) Sex, gender, and the individual. In Handbook of Personality, Pervin, L. ed. The Guilford Press, New York, NY.
Beckman, L. J. (1978) Self-esteem of women alcoholics. Journal of Studies on Alcohol 39, 491498.[ISI][Medline]
Bem, S. L. (1975) Sex role adaptability: one consequence of psychological androgyny. Journal of Personality and Social Psychology 4, 634643.
Bergman, H., Bergman, I., Engelbrektsson, K., Holm, L., Johannesson, K. and Lindberg, S. (1988) Psykologhandbok del 1 (In Swedish) (Manual for psychologists, part 1). Magnus Huss Klinik, Karolinska sjukhuset, Stockholm.
Blanchard-Fields, F., Suhrer-Roussel, L. and Hertzog, C. (1994) A confirmatory factor analysis of the Bem Sex Role Inventory: old questions, new answers. Sex Roles 5/6, 423457.
Hasin, D., Grant, B., Harford, T., Hilton, M. and Endicott, J. (1990) Multiple alcohol-related problems in the United States: on the rise? Journal of Studies on Alcohol 51, 485493.[ISI][Medline]
Hensing, G., Spak, F., Alexanderson, K. and Allebeck, P. (1997) Sick-leave among women and the role of psychiatric disorder. Scandinavian Journal of Social Medicine 3, 185192.
Knibbe, R. (1999) The comparability of alcohol consumption units and alcohol measures in European countries and possibilities for standardisation. In Alcohol Consumption and Alcohol Problems among Women in European Countries Project Final Report. Free University, Berlin.
Lara-Cantú, M. A., Medina-Mora, M. E. and Gutiérrez, C. E. (1990) Relationship between masculinity and femininity in drinking in alcohol-related behavior in a general population sample. Drug and Alcohol Dependence 26, 4554.[CrossRef][ISI][Medline]
Midanik, L. (1982) The validity of self-reported alcohol consumption and alcohol problems: a literature review. British Journal of Addiction 77, 357382.[ISI][Medline]
Murnen, S. A. and Smolak, L. (1997) Femininity, masculinity, and disordered eating. A meta-analytic review. International Journal of Eating Disorders 22, 231242.[CrossRef][ISI][Medline]
Neve, R. J. M., Lemmens, P. H. and Drop, M. J. (1997) Gender differences in alcohol use and alcohol problems: mediation by social roles and gender-role attitudes. Substance Use and Misuse 32, 14391459.[Medline]
Robins, L. N., Cottler, L. B. and Babor, T. F. (1986) The WHO/ ADAMHA CIDI-SAM Interview (revised 1987). WHO, Geneva.
Room, R. (1996) Gender roles and interactions in drinking and drug use. Journal of Substance Abuse 8, 227239.[CrossRef][ISI][Medline]
Sorell, G. T., Silvia, L. Y. and Busch-Rossnagel, N. A. (1993) Sex-role orientation and self-esteem in alcoholic and nonalcoholic women. Journal of Studies on Alcohol 54, 566573.[ISI][Medline]
Spak, F. (1996) To what extent is female alcohol dependence and abuse known in the health care system? The use of multi-source information in a Swedish population survey. Acta Psychiatrica Scandinavica 93, 8791.[ISI][Medline]
Spak, F. and Hällström, T. (1995) Prevalence of female alcohol dependence and abuse in Sweden. Addiction 90, 10771088.[CrossRef][ISI][Medline]
Spak, F. and Hällström, T. (1996) Screening for alcohol dependence and abuse in women: description, validation, and psychometric properties of a new screening instrument, SWAG, in a population study. Alcoholism: Clinical and Experimental Research 4, 723731.
Spence, J. T., Helmreich, R. L. and Holahan, C. K. (1974) The Personal Attributes Questionnaire: a measure of sex-role stereotypes and masculinity-femininity. JSAS: Catalog of Selected Documents in Psychology 43, Ms. No. 617.
Spence, J. T., Helmreich, R. and Stapp, J. (1975) Ratings of self and peers on sex role attributes and their relation to self-esteem and conceptions of masculinity and femininity. Journal of Personality and Social Psychology 32, 2939.[ISI][Medline]
Visser, I. (1996) The prototypicality of gender. Contemporary notions of masculine and feminine. Womens Studies International Forum 19, 589600.[CrossRef]
Wilsnack, S. C., Klassen, A. D. and Wright, S. I. (1985) Gender-role orientations and drinking among women in a U.S. national survey. In Proceedings of the Thirty-Fourth International Congress on Alcoholism and Drug Dependence, Calgary, Canada.
Wilsnack, S. C., Klassen, A. D., Schur, B. E. and Wilsnack, R. W. (1991) Predicting onset and chronicity of womens problem drinking: a five-year longitudinal analysis. American Journal of Public Health 3, 305318.