Although gender is often acknowledged as a determinant of health, measuring its components, other than biological sex, is uncommon. The Bem Sex Role Inventory (BSRI) quantifies self-attribution of traits, indicative of gender roles. The BSRI has been used with participants across cultures and countries, but rarely in an older population in Brazil, as we have done in this study. Our primary objective was to determine whether the BSRI-12 can be used to explore gender in an older Brazilian population.
The BSRI was completed by volunteer participants, all community dwelling adults aged 65+ living in Natal, Brazil. Exploratory factor analysis was performed, followed by a varimax rotation (orthogonal solution) for iteration to examine the underlying gender roles of feminine, masculine, androgynous and undifferentiated, and to validate the BSRI in older adults in Brazil.
The 278 participants, (80 men, 198 women) were 65–99 years old (average 73.6 for men, 74.7 for women). Age difference between sexes was not significant (p = 0.22). A 12 item version of the BSRI (BSRI-12) previously validated among Spanish seniors was used and showed validity with 5 BSRI-12 items (Cronbach=0.66) loading as feminine, 6 items (Cronbach=0.51) loading onto masculine roles and neither overlapping with the category of biological sex of respondent.
Citation: Carver LF, Vafaei A, Guerra R, Freire A, Phillips SP (2013) Gender Differences: Examination of the 12-Item Bem Sex Role Inventory (BSRI-12) in an Older Brazilian Population. PLoS ONE 8(10): e76356. https://doi.org/10.1371/journal.pone.0076356
Editor: David W Dowdy, Johns Hopkins Bloomberg School of Public Health, United States of America
Received: August 8, 2012; Accepted: August 26, 2013; Published: October 2, 2013
Copyright: © 2013 Carver et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was funded by: CNPq - Conselho Nacional de Desenvolvimento Tecnológico (Website: www.cnpq.br), and FAPERN - Fundação de Apoio a Pesquisa do Rio Grande do Norte Website: www.fapern.rn.gov.br (Project Grant Number: 474728/2009-9.) The total funding for the project was $10,000. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Many researchers use the terms sex and gender interchangeably. However, sex is defined physiologically whereas gender is a cultural construct that includes social and psychological factors . Gender is linked to roles and behaviors expected of men and women in a particular culture at a specific time, and may be influenced by education and socioeconomic status. Sex and gender roles are “intertwined; they influence each other in a network of constant reciprocal changing processes” (p. 72) , however, it is possible to measure them separately. This research examines the ability of a short version of the Bem Sex Role Inventory (BSRI) to assess gender role in a sample of older Brazilian adults.
Defining Gender Roles
Gender is an understudied area in health research, particularly among older adults. When it is considered, the most commonly used and repeatedly validated measure of gender roles is the Bem Sex Role Inventory (BSRI) developed 4 decades ago . Sandra Bem categorized instrumental traits including: taking the lead, being aggressive, competitive, dominant, self-reliant, and athletic as masculine; while feminine role characteristics were considered expressive and included compassion, affection, sympathy, warmth, and being yielding.
Bem was the first to conceptualize gender roles as something other than exclusively masculine and feminine [3–5], defining a third role, androgyny, that combined both masculine and feminine traits and a fourth category, undifferentiated, describing people whose scores on both masculine and feminine traits were low . She hypothesized that “many individuals might be "androgynous"; that is, they might be both masculine and feminine, both assertive and yielding, both instrumental and expressive—depending on the situational appropriateness of these various behaviors” (p. 155) . Androgynous men and women were postulated to be adaptive and therefore more likely to have better mental health and higher competence . Some dispute the androgyny model [8,9] while others assert that androgyny exists, but reflects a developmental change .
The original BSRI included 60 dichotomous items divided into 3 subscales (Masculinity, Femininity, and Neutral) of 20 items each . A personality characteristic was categorized as feminine if it was independently judged, using a 7 point scale, by both females and males to be significantly more desirable for women than for men  and vice versa for masculine characteristics. Bem’s early studies  found alpha coefficients for masculinity of .86 and between .80 to .82 for femininity. Kamas and Preston  recently examined the BSRI with 310 college students determining that the scale remains valid.
Shorter versions of the BSRI are common. In 1981, Bem used factor loading to develop a 30 item scale, with 10 items per subscale , validated independently by others [13,14]. Following further confirmatory factor analysis Campbell, Gillaspy and Thompson  recommended using the shortened BSRI in future research due to its high reliability (αM = .82, αF = .89). A 12 item Spanish version of the BSRI (BSRI-12) was validated by Mateo and Fernandez  and included the items: gentle, sympathetic, leadership abilities, acts as a leader, dominant, tender, warm, affectionate, strong personality, defend own beliefs, sensitive to other’s need, and makes decision easily. This version demonstrated strong psychometric properties, in some cases better than the original 60 item BSRI [13,15,16].
Mateo and Fernandez  derived the 12 item BSRI (BSRI-12) from the original 60 item BSRI using a sample of Spanish university students. Students were told to evaluate the extent to which each item applied to him/her. A 7-point Likert scale was used: 1 signifying that the item was not applicable and 7 indicating the item totally applied. The BSRI-12 was translated into Spanish by one of the authors and then back-translated into English by two bilingual researchers, one of them was a U.S. citizen whose first language was English, the other was born and raised in Spain. Mateo and Fernandez  analyses resulted in a 12 item scale, shown in Figure 1, with six masculine/instrumental (M/I) items (defends own beliefs, strong personality, has leadership abilities, makes decisions easily, dominant, acts as a leader) and six feminine/expressive (F/E) items (affectionate, sympathetic, sensitive to needs of others, warm, tender, gentle).
Fernandez and Coello  used the BSRI-12 more recently, with students ranging in age from 12 to 15 years. They found reasonable reliability, using Cronbach’s alpha, on the anticipated Feminine/Expressive (F/E) and Masculine/Instrumental (M/I) scales (F/E scale α = .77; M/I scale α = .73) indicating internal consistency. They performed factor analysis using principal axis exploratory factor analysis (EFA) with oblimin rotation and δ = 0, revealed three factors explaining 58.11% of the total variance. Fernandez and Coello  described these factors as F1 (F/E) which loaded feminine/expressive items exclusively, explained 25.9% of total variance; F2 (M/I) loading exclusively masculine items and explaining 23.68% of the variance; and F3 (with both M/I items and F/E items) and accounting for 8.52% of variance. They also performed a principal components analysis which resulted in the same three factors, which were not significantly intercorrelated.
Gender Role Classification
Methods used to classify respondents’ scores into gender role vary and can impact the gender role attributed to a respondent [17,18]. Generally classification is done using participants’ scores on the masculine and feminine scales. Common methods for classifying scores on the BSRI into gender roles are to split the sample using either the medians from Bem’s original normative samples , the theoretical mean of the scale  or the sample medians [17–19].
Whether dividing the scores by theoretical mean or median (sample or normative) the technique is the same. For example, using the median split, Lenney  suggested that subjects should be classified as masculine (M) “if their M score is above the Median and their F score is below the F median; as feminine if the F score is above the F median, and their M score is below the M median; as androgynous if both their scores are above the respective scale medians; and as undifferentiated (or indeterminate) if both their scores are below the respective scale medians” (p. 579). Medians for the masculine and feminine scales are established using scores from the whole sample; however, Lenney  points out that some researchers use medians established in Bem’s normative studies. Caution must be used in considering the generalizability of the results of these methods. As Sedney  points out, using different sample medians to divide people into gender role classifications can result in “a subject who scored 5.00 on both the BSRI F and M scales might be called Androgynous at Stanford, Feminine at Harvard, Masculine at Michigan, and Undifferentiated at a liberal arts college in Rhode Island” (p. 219).
International use of BSRI
The full 60 item and the short versions of the BSRI appear to be valid across geography and culture. The BSRI has been used, for example, in Zimbabwe , Japan, and China  with good validity and reliability when a few items were removed to improve cultural fit. Using a short version of only instrumental items in a thirteen nation study with 6,543 participants, mean age 38.5 years old, Arrindell, van Well, Kolk, Barelds, Oei et al.  demonstrated internal consistency (Cronbach α = .83 averaged across the thirteen countries) and a strong correlation (r = .57, p < 0.05) with the Masculine Gender Role Stress (MGRS) scale. Research using both Spanish and English versions of the BSRI among middle aged women (mean age 44 years) has demonstrated reliability of the masculine (α = .78 in English and .76 in Spanish) and feminine (α = .88 in English and .86 in Spanish) scales . Although the Bem Sex Role Inventory has some history of reliability and validity in Brazil [24–26], to the best of our knowledge neither the full nor the 12 item BSRI has been evaluated among older Brazilians.
Gender and Aging
Gender is often overlooked in studies of aging populations. When it is considered, the BSRI appears to be the measure of choice [19,27]. For both men and women, gender identity is less rigid among older individuals [28,29]. In the few studies that looked at gender roles within aging populations, findings suggest that gender roles and biological sex may not be related. Twenge  found that with aging, women’s masculinity, but not their androgyny, scores increase. Huyck  also suggested that women embrace instrumental traits as they age. For both men and women, gender identity is less rigid among older individuals [28,29]. Finding that BSRI scores were not clearly masculine, feminine or androgynous among seniors (mean age of 70.6) Windle and Sinnott  hypothesized that this reflected complex gender roles rather than a dichotomous construct.
Gale-Ross et al.  examined gender role, life satisfaction and wellness in older Canadian women. Among their sample of older adults, 40% (n = 18) of male participants were classified by median split as masculine (instrumental), as compared to 11% (n = 6) of female participants. Thirty-six percent (n = 20) of women were classified as feminine (expressive) as compared to 16% of males (n = 7). Interestingly, 24% of males (n = 11) and 26% of females (n = 14) were classified as androgynous. More women (27%, n = 15) than men (20%, n = 9) were classified as undifferentiated. The variation between men and women in the gender classifications was not statistically significant. Gale-Ross et al.  concluded that their small sample size and lack of diversity may have played a role in their gender role results. They did find that androgyny was significantly associated with life satisfaction and general wellbeing.
Given the limited research on gender and older adults our primary objective was to determine whether the BSRI-12 can be used to explore gender in an older Brazilian population. Consistent with postulates underlying the BSRI, we hypothesized that men would be disproportionately represented in the masculine (instrumental) category while women would fall in the feminine (expressive) category on the BSRI.
Three hundred participants were randomly sampled from among community dwelling adults (older than 64 years) living in Natal, the capital of the province Rio Grande do Norte, Brazil. Two hundred and seventy-eight individuals (93%, 80 men, 198 women) agreed to participate in the study. Participants’ ages ranged from 65 to 99 with an average of 74.4 years. Marital status is given in Table 1. Fewer than half the participants (45%) were married; almost one third (32%) were widowed, the remainder were single or divorced. Men were, on average, younger than women (73.6 vs. 74.7), however the difference was not significant (t-test p value=0.22). Ninety-three percent of participants had fewer than six years of education (Table 2). Only 42% were satisfied with their level of income. Despite variations in socio-demographic factors within the groupings by sex, differences across sexes were not significant (Table 2).
|Married (%)||Single (%)||Widowed (%)||Divorced (%)||Total (%)|
|Men||38 (48)||12 (15)||26 (32)||4 (5)||80 (100)|
|Women||86 (43)||34 (17)||36 (32)||15 (8)||198 (100)|
Written informed consent was granted by all participants. This project was approved by the Committee on Ethics and Research of the University Hospital Onofre Lopes.
This study used the BSRI-12 (English items listed in Figure 1.) translated from Spanish  to Portuguese by two Brazilian translators familiar with the concept of gender roles. Both Brazilian and Canadian researchers performed back translation . As described above, BSRI-12 measures participants’ instrumental (masculine) and expressive (feminine) traits and thereby, gender roles. A 7-point Likert scale was used: 1 represented ‘not applicable’ and 7 represented ‘totally applicable.’ In the Spanish version of the BSRI-12 Mateo and Fernandez  found the coefficient of internal consistency  was .83 to .94.
To obtain a general overview of the population under study, descriptive statistics for each category of SES were calculated. Results are reported for the total population, as well as stratified by sex.
The BSRI-12 was subjected to an exploratory factor analysis to validate it as a measure of gender roles in this older population. The principal axis method was used to extract underlying factors, followed by a varimax rotation (orthogonal solution). The reason for choosing orthogonal solution over oblique solution was that gender constructs extracted in initial analyses were not correlated . Two standard methods were employed to determine the number of factors to retain: Kaiser Criterion, which retains factors with Eigenvalues greater than 1 , and Cattell’s  scree test, which involves an examination of a plot of the Eigenvalues for breaks or discontinuities.
In interpreting the rotated factor pattern, an item was said to load onto a given factor if the factor loading was greater than or equal to 0.25. Items which load on two factors similarly will not provide a meaningful interpretation . To establish a priori criteria, if the differences between loading on two or more factors were smaller than 0.10, the item was excluded from analysis . To assess the degree to which items included in each factor measured the same underlying concept (internal reliability, consistency), Cronbach’s Coefficient Alpha was computed for each of the retained factors .
Classification into Gender Roles.
In the literature there are multiple methods for classifying people into gender roles. The most common method uses the median split. This method was used in a recent study of an elderly population  and avoids methodological issues that occur when other approaches are used . Therefore the median split method was used to classify the gender roles of these participants. First the median for the whole sample was established for both the masculine and the feminine scales. Then individual scores for each participant on the femininity scale and the masculinity scale were calculated and compared to the median. Scores that fell at the median were classified as “high” rather than “low” scores. If the individual’s mean score was below the median on both the feminine and masculine scales, he/she was classified as undifferentiated. If the individual’s mean scores on both the masculine and feminine scales were equal to or above the median that individual was classified as androgynous. Those people who were equal to or higher than the median on the feminine scale and lower on the masculine scale were classified as feminine. Finally, those who were equal to or higher than the median on the masculine scale and lower on the feminine scale were classified as masculine (see Figure 2).
Prior to performing factor analysis the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy (KMO = .71) was used to ascertain the amount of item-variance shared among variables in the input matrix. Given that the KMO was greater than .70, the input matrix was deemed adequate in size . The Bartlett’s Test of Sphericity (χ2 = 357.8, df = 66, p < .0001) was statistically significant, demonstrating that the input correlation matrix was not equal to zero and therefore was suitable for factor analysis.
The exploratory factor analysis with varimax rotation resulted in factor loadings presented in Table 3. Results suggested retention of two factors based on our previously defined criteria. Factor 1 loaded feminine/expressive (F/E) items and accounted for 22% of the total variance. The second factor loaded masculine/instrumental (M/I) items and explained 12% of the variance. The inter-factor correlation (0.39) was below .50, the cut-off point recommended by Gorsuch  so factor analysis was deemed acceptable.
|Factor 1 (Feminine)||Loading||Factor 2 (Masculine)||Loading|
|Warm||0.62||Has leadership abilities||0.50|
|Affectionate||0.77||Acts as leader||0.40|
|Sensitive to other’s needs||0.36||Defends own beliefs||0.28|
|Tender*||-0.06||Makes decisions easily||0.28|
|Cronbach’s alpha||0.66||Cronbach’s alpha||0.51|
The five items loading onto factor 1 (F/E) were: warm (0.62); gentle (0.48); affectionate (0.77); sympathetic (0.39); sensitive to other’s needs (0.36). All were feminine items in the original inventory and showed reasonable internal consistency (Cronbach’s alpha=0.66). All six items loading onto factor 2 (M/I) were masculine items: leadership abilities (0.50); strong personality (0.49); acts as a leader (0.40); dominant (0.28); and defends own belief (0.28). The Cronbach’s alpha (α=0.51) suggested weak internal consistency among these items. One item, ‘tender,’ did not load onto either factor and was, therefore, excluded from further analysis.
Gender role groups based on the Bem androgyny model
The Bem androgyny model  was the conceptual base for constructing four gender role groups. Using a median split: among men, 18% (n=14) were categorized as masculine, 32% (n=26) were androgynous, that is, high in both masculine and feminine items, 16% (n=13) were categorized as feminine and 34% (n=27) were undifferentiated. For women, 20% (n=40) were classified as feminine, 19% (n=37) were masculine, 34% (n=67) were androgynous and 26% (n=52) were undifferentiated (Table 4). Most importantly, perceived gender roles did not differ significantly by sex (t(274) = -.737, p < .46), suggesting that biological sex is a different entity than gender.
|Gender role group||Male sex (%)||Female sex (%)|
|Undifferentiated||27 (34)||52 (26)|
|Masculine||14 (18)||37 (19)|
|Feminine||13 (16)||40 (20)|
|Androgynous||26 (32)||67 (34)|
|Total||80 (100)||198 (100)|
In summary, our factor analysis showed that, in this population, the BSRI-12 differentiated two factors, corresponding to feminine and masculine scales. Contrary to our predictions, gender role classification did not reflect biological sex. In fact a higher percentage of both males and females were classified as either androgynous or undifferentiated than those classified in traditional gender roles of masculine and feminine. Furthermore, the lack of association between biological sex and gender roles indicates that sex and gender roles are different entities in this population.
There is growing evidence that gender roles have an effect on health that is independent of biological sex itself, with masculinity likely conferring greater risks of illness for both men and women [40–44]. The most utilized measure of gender roles is the BSRI, developed almost 4 decades ago and validated repeatedly either as a 60 item inventory or an abbreviated form. Asking about, and including measures of gender roles among an older, non-English speaking population is surprisingly rare in health outcomes research. Perhaps this is because translating the concepts in the inventory is not simple and straightforward, but it may also speak to assumptions that seniors are either unaffected by gender roles or that they all ascribe to the same roles. If research is to acknowledge and account for gender as a social determinant of health then measuring gender roles is a key component of such research. Gender, however, is not fixed or static but varies across time and place making repeated validation of measures such as the BSRI necessary as characteristics of any population studied vary .
It would appear that the gender role differentiation power of BSRI-12 was not as strong statistically in our research as in a number of earlier studies. There are several possible explanations for this. Although the BSRI-12 could separate gender roles among this group, Cronbach’s alphas were below the traditionally acceptable limit of .70, perhaps because, as Streiner  explains, measures with fewer items result in reduced alpha’s. We also found different factor loadings than in previous studies using the BSRI-12 . Despite such variations in loadings and Cronbach’s alphas, other researchers’ results consistently list items on the BSRI-12 among the items on their first two factors . It may be that over the decades and with changes in gender roles the BSRI will require modification to better reflect social expectations of men and women. This is in keeping with the fluid and contextual nature of gender, itself.
Our original hypothesis was that a greater proportion of males would be classified in the masculine/instrumental category and a larger proportion of females would be classified as feminine/expressive. This was not the case. Similar to other researchers studying older adults , we found little overlap between biological sex and gender roles; that is, men were not significantly more likely to be masculine than were women. In fact, a smaller percentage of males and females fell into the traditional gender roles of masculine and feminine than into the androgynous and undifferentiated categories. In this sample of older Brazilians, unlike Twenge’s  finding that masculinity scores, rather than androgyny scores, increase in aging women, we found that more men (32%) and women (34%) were androgynous than masculine (men 18%; women 19%).
The lack of differentiation according to physiological sex between gender roles may have been due to the translation of the instrument, classification method or the low education and socioeconomic status of our participants. Perhaps the BSRI-12 does not describe masculinity and femininity in ways that fit the context of our study population. On the other hand, and in keeping with other research [19,26] this lack of congruency between biological sex and gender roles may indicate that each is an independent contributor to who a person is and, potentially, to well-being; and that measuring both constructs will add explanatory value in social determinants research. Despite frequent assumptions that the elderly are more traditional than younger people, the lack of overlap between sex and gender roles in this older population from a relatively traditional culture suggests that with aging, gender roles may actually become less stereotypic and rigid, even in a more traditional and less egalitarian society.
To deepen the meaning of biological sex and address how or whether aspects of gender are determinants of health requires including measures of gender equalities, constraints and expectations in quantitative research. The BSRI is one such measure that, despite its age, remains valid, and adds meaning not captured by the simple dichotomous classification of sex. It does require revalidation when used with new study populations.
Among a relatively socioeconomically deprived population of Brazilian seniors the BSRI-12 may have measured aspects of gender not encompassed by the construct of sex. Research on education, SES and gender may reveal the extent to which socioeconomic deprivation impacts gender. Using only sex in research automatically dichotomizes a population. Including the BSRI-12 as a measure of gender roles among this group of older Brazilian adults suggests convergence rather than division between men and women. And, given that fatigue is a common issue for older adults participating in research studies , using less time-consuming instruments, such as the BSRI-12, is an important strategy. The BSRI-12 appears to have meaning across contexts and age groups, and to be a valid measure of one aspect of gender among this older, low socioeconomic status, Brazilian population.
Conceived and designed the experiments: RG AF. Performed the experiments: AF RG. Analyzed the data: AV LFC SPP. Contributed reagents/materials/analysis tools: RG AF AV LFC SPP. Wrote the manuscript: LFC AV SPP.
- 1. Woodhill BM, Samuels CA (2003) Positive and negative androgyny and their relationship with psychological health and wellness. Sex Roles 48: 555–565. doi:https://doi.org/10.1023/A:1023531530272.
- 2. Schmitz S (2010) Sex, gender, and the brain – biological determinism versus socio-cultural constructivism. In: I. KlingeC. Wiesemann. Sex and gender in biomedicine: Theories, methodologies, results. Göttingen: University of Goettingen. pp. 57-72.
- 3. Bem SL (1974) The measurement of psychological androgyny. J Consult Clin Psychol 42: 155-162. doi:https://doi.org/10.1037/h0036215. PubMed: 4823550.
- 4. Spence JT, Helmreich R, Stapp J (1975) Ratings of self and peers on sex role attributes and their relation to self-esteem and conceptions of masculinity and femininity. J Pers Soc Psychol 32: 29-39. doi:https://doi.org/10.1037/h0076857. PubMed: 1206468.
- 5. Constantinople A (1973) Masculinity-femininity: An exception to a famous dictum? Psychol Bull 80: 389-407. doi:https://doi.org/10.1037/h0035334. PubMed: 4147970.
- 6. Hoffman DM, Fidell LS (1979) Characteristics of androgynous, undifferentiated, masculine, and feminine middle-class women. Sex Roles 5: 765-781. doi:https://doi.org/10.1007/BF00287937.
- 7. Bem SL (1977) On the utility of alternative procedures for assessing psychological androgyny. J Consult Clin Psychol 45: 196-205. doi:https://doi.org/10.1037/0022-006X.45.2.196. PubMed: 850004.
- 8. Flaherty J, Dusek J (1980) An investigation of the relationship between psychological androgyny and component of self-concept. J Pers Soc Psychol 38: 984-992. doi:https://doi.org/10.1037/0022-35184.108.40.2064.
- 9. Lott B (1981) A feminist critique of androgyny: Toward the elimination of gender attributions for learned behaviour. In: C. MayoNM Henley. Gender and non-verbal behavior. New York: Springer-Verlag. pp. 171-180.
- 10. Hyde JS, Phillis DE (1979) Androgyny across the life span. Dev Psychol 15: 334-336. doi:https://doi.org/10.1037/0012-16220.127.116.114.
- 11. Bem SL (1981) A manual for the Bem Sex Role Inventory. Palo Alto: Consulting Psychologist Press.
- 12. Kamas L, Preston A (2012) The importance of being confident; gender, career choice, and willingness to compete. J Econ Behav Organ 83: 82-97. doi:https://doi.org/10.1016/j.jebo.2011.06.013.
- 13. Campbell T, Gillaspy JA Jr, Thompson B (1997) The factor structure of the Bem Sex Role Inventory (BSRI): Confirmatory analysis of long and short forms. Educ Psychol Meas 57: 118-124. doi:https://doi.org/10.1177/0013164497057001008.
- 14. Martin HJ, Ramanaiah NV (1988) Confirmatory factor analysis of the Bem Sex-Role Inventory. Psychol Rep 62: 343-350. doi:https://doi.org/10.2466/pr0.1918.104.22.1683.
- 15. Mateo MA, Fernández J (1991) La dimensionalidad de los conceptos de masculinidad y feminidad. Investigaciones Psicológicas 9: 95-116.
- 16. Fernández J, Coello MT (2010) Do the BSRI and PAQ really measure masculinity and femininity? Span J Psychol 13: 1000-1009. doi:https://doi.org/10.1017/S113874160000264X. PubMed: 20977047.
- 17. Sedney MA (1981) Comments on median split procedures for scoring androgyny measures. Sex Roles 7: 217-218. doi:https://doi.org/10.1007/BF00287807.
- 18. Lenney E (1991) Sex role: The measurement of masculinity, femininity and androgyny. In: JP RobinsonPR ShaverLS Wrightsman. Measures of Personality and Social Psychological Attitudes, Vol. 1. San Diego: Academic Press. pp. 573-660.
- 19. Gale-Ross R, Baird A, Towson S (2009) Gender role, life satisfaction, and wellness: androgyny in a southwestern Ontario sample. Can J Aging 28: 135-146. doi:https://doi.org/10.1017/S0714980809090187. PubMed: 19860972.
- 20. Wilson D, McMaster J, Greenspan R, Mboyi L, Ncube T et al. (1990) Cross-cultural validation of the Bem Sex Role Inventory in Zimbabwe. Pers INDIV Differ 11: 651-656. doi:https://doi.org/10.1016/0191-8869(90)90249-Q.
- 21. Katsurada E, Sugihara Y (1999) A preliminary validation of the Bem Sex Role Inventory in Japanese culture. J Cross Cult Psychol 30: 641-645. doi:https://doi.org/10.1177/0022022199030005005.
- 22. Arrindell WA, van Well S, Kolk AM, Barelds DPH, Oei TS et al. (2012) Higher levels of masculine gender role stress in masculine than in feminine nations: A thirteen-nations study. Cross Cult Res 47: 51–67.
- 23. Sabina C, Cuevas CA, Schally JL (2012) The cultural influences on help-seeking among a sample of victimized Latino women. Am J Commun Psychol 49: 347–363. doi:https://doi.org/10.1007/s10464-011-9462-x.
- 24. Oliveira LSO (1983) Masculinidade feminilidade androginia. Rio de Janeiro: Achiamé.
- 25. Hutz C, Koller S (1992) A mensuração do gênero: uma readaptação do BSRI. Psicol Reflex Crit 5: 15-21.
- 26. Hernandez JAE (2009) Reavaliando o Bem Sex-Role Inventory. Estud Psicol 26: 73-83.
- 27. Windle M, Sinnott JD (1985) A psychometric study of the Bem Sex Role Inventory with an old sample. J Gerontol 40: 336-343. doi:https://doi.org/10.1093/geronj/40.3.336. PubMed: 3989248.
- 28. Sinnott JD, Shifren K (2001) Gender and aging: Gender differences and gender roles. In: JE BirrenKW Schaie. Handbook of the psychology of aging, 5th ed. San Diego: Academic Press. pp. 454–476.
- 29. Gutmann D (1987) Reclaimed powers: Toward a new psychology of men and women in later life. New York: Basic Books.
- 30. Twenge JM (1997) Changes in masculine and feminine traits over time: a meta-analysis. Sex Roles 36: 305-326. doi:https://doi.org/10.1007/BF02766650.
- 31. Huyck MH (1996) Continuities and discontinuities in gender identity. In: VL Bengston. Adulthood and aging: Research on continuities and discontinuities. New York: Springer Verlag. pp. 98–121.
- 32. Carmines EG, Zeller RA (1979) Reliability and Validity Assessment. Beverly Hills: Sage Publishing House.
- 33. Hatcher L (1994) A step-by-step approach to using the SAS system for factor analysis and structural equation modeling. Cary: SAS Institute Inc.
- 34. Kaiser HF (1960) The application of electronic computers to factor analysis. Educ Psychol Meas 20: 141-151. doi:https://doi.org/10.1177/001316446002000116.
- 35. Cattell RB (1966) The scree test for the number of factors. MULTIVAR BEHAV RES 1: 245-276.
- 36. Cronbach L (1951) Coefficient alpha and the internal structure of tests. Psychometrika 16: 297-334. doi:https://doi.org/10.1007/BF02310555.
- 37. Blackman S (1982) Comment on three methods of scoring androgyny as a continuous variable. Psychol Rep 51: 1100-1102. doi:https://doi.org/10.2466/pr0.1982.51.3f.1100.
- 38. Field AP (2009) Discovering statistics using SPSS. London: Solid Action on Globalization and Environment.
- 39. Gorsuch RL (1983) Factor analysis (2nd ed.). Hillsdale: Erlbaum.
- 40. McCartney G, Mahmood L, Leyland AH, Batty GD, Hunt K (2011) Contribution of smoking-related and alcohol-related deaths to the gender gap in mortality: evidence from 30 European countries. Tob Contr 20: 166-168. doi:https://doi.org/10.1136/tc.2010.037929. PubMed: 21228431.
- 41. Lyons AC, Willott SA (2008) Alcohol consumption, gender identities and women’s changing social positions. Sex Roles 59: 694-712. doi:https://doi.org/10.1007/s11199-008-9475-6.
- 42. de Visser R, Smith JA (2007) Alcohol consumption and masculine identity among young men. Psychol Health 22: 595-614. doi:https://doi.org/10.1080/14768320600941772.
- 43. Courtenay WH (2000) Constructions of masculinity and their influence on men’s well-being: a theory of gender and health. Soc Sci Med 50: 1385-1401. doi:https://doi.org/10.1016/S0277-9536(99)00390-1. PubMed: 10741575.
- 44. Hunt K, Hannah MK, West P (2004) Contextualising smoking: masculinity, femininity and class differences in smoking in men and women from three generation in the west of Scotland. Health Educ Res 19: 239-249. doi:https://doi.org/10.1093/her/cyg061. PubMed: 15140844.
- 45. Streiner DL (2003) Starting at the beginning: An introduction to coefficient alpha and internal consistency. J Pers Assess 80: 99-103. doi:https://doi.org/10.1207/S15327752JPA8001_18. PubMed: 12584072.
- 46. Avlund K, Pedersen AN, Schroll M (2003) Functional decline from age 80 to 85: influence of preceding changes in tiredness in daily activities. Psychosom Med 65: 771-777. doi:10.1097/01.PSY.0000082640.61645.BF. PubMed: 14508019.