Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
P2: FYJ/FOP
QC: FLF
PP029-292515
19:37
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
530
19:37
Ward
consequences for gender identity and gender role socialization. Bem (1974)
originally argued that masculinity (M) and femininity (F) represented independent clusters of socially desirable instrumental and expressive traits,
and that it was possible, indeed preferable, for individuals to internalize
both masculine and feminine psychological attributes. The psychologically
androgynous person, that is, the individual who possesses comparable levels
of both masculine and feminine qualities, was perceived as relatively advantaged in comparison with sex-typed individuals in terms of behavioral
flexibility and psychological well-being (Bem, 1979). In her early research
Bem additionally demonstrated that androgynous individuals were more
comfortable and competent in performing a variety of tasks and provided
some evidence that androgyny (A) was linked to self-esteem (Bem, 1975,
1977; Bem & Lenney, 1976).
Androgyny research has weathered more than two decades of storms
and retains a prominent place in contemporary psychology despite both psychometric criticisms of the Bem Sex Role Inventory (BSRI) and theoretical
reservations about Bems original conceptualization of androgyny (see reviews by Cook, 1985, 1987; Lorenzo-Cioldi, 1996; Spence, 1984; Taylor &
Hall, 1982). Although researchers have acknowledged the BSRIs adequate
reliability and convergent and discriminant validity (e.g., Chung, 1996; Holt
& Ellis, 1998; Ramanaiah & Martin, 1984; Taylor & Hall, 1982; Vacha-Haase,
1998; Wilson & Cook, 1984), psychometric concerns have been voiced about
(1) the gender linkage and independence of the M and F subscales (BallardReisch & Elton, 1992; Lee & Scheurer, 1983; Marsh & Myers, 1986), (2) the
instability of the factor structures (Campbell, Gillaspy, & Thompson, 1997;
Gaudreau, 1977; Moreland, Gulanick, Montague, & Harren, 1978; Pedhazur
& Tetenbaum, 1979; Ruch, 1984; Thompson, 1989; Waters, Waters, & Pincus,
1977; Wilson & Cook, 1984), (3) the categorical versus continuous scoring
techniques (Blackman, 1982; Bryan, Coleman, & Ganong, 1981; Kalin, 1979;
Kelly, Furman, & Young, 1978; Spence, Helmreich, & Stapp, 1975; Strahan,
1975, 1981), and (4) the relative desirability of the male versus the female
subscale items (Taylor & Hall, 1982).
In addition, theoretical controversies have arisen about the relationship
between masculinity, femininity, androgyny and personality, self-concept,
and psychological well-being. In this instance three competing models of
androgyny have been advanced: (1) a masculinity model, which argues that
masculinity is responsible for the major variance in indicators of psychological and social well-being (Taylor & Hall, 1982; Whitley, 1983); (2) an
additive model, which suggests that both M and F contribute positively and
uniquely to the prediction of psychological adjustment (Spence, 1984); and
(3) an interactive model, which hypothesizes a M F interaction effect
on self-concept and proposes that androgynous persons are psychologically
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
Psychological Androgyny
19:37
531
advantaged compared with masculine, feminine, and undifferentiated individuals. Bems (Bem, 1974) original balance model of androgyny posited
M F interactions without significant main effects on psychological outcomes although more recently Hall and Taylor (1985) have discussed an
emergent interactive model, which suggests significant main effects along
with the interactions. Overall, the masculine model has received particularly strong support in the empirical literature (also see Lundy & Rosenberg, 1987; Markstrom-Adams, 1989; Williams & DAlessandro, 1994), but
research findings have by no means been conclusive (Francis & Wilcox,
1998; Hall & Taylor, 1985; Lubinski, Tellegen, & Butcher, 1983; Marsh, 1987;
Shifren & Bauserman, 1996).
Despite an abundance of research, methodological shortcomings of the
BSRI limit the conclusions that can be drawn about the most convincing
model of androgyny and the relationship between M, F, A, self-concept, and
psychological well-being. More specifically, the relatively greater desirability of the masculine versus feminine traits remains problematic. Feminine
traits are, on the whole, evaluated as less desirable than masculine qualities,
and the tendency for masculine subscale scores to account for the substantial
proportion of the variance in self-esteem measurements and adjustment indicators has been attributed by some to these evaluative differences (Silvern
& Ryan, 1979). In many instances the differential valuing of masculine and
feminine traits has prompted claims of measurement artifact. Taylor and
Hall (1982), however, have countered these claims and have argued that if
traits associated with men are more highly valued than those associated with
women, then this should be reflected in the assessment technique and that
such differences are not artifactual. Regarding the issue of social desirability, they conclude artificially creating a socially desirable femininity scale
in order to demonstrate the positive effects of femininity would make the
whole research enterprise misleading (Taylor & Hall, 1982, p. 361).
Although the nature of masculinity, femininity, and social desirability
in the United States remains open to debate, extension of androgyny research to more culturally diverse settings may provide a natural avenue for
balancing the relative desirability rating of gender-linked traits in the measurement of psychological androgyny and for unraveling confounding variables observed in monocultural research. More specifically, it might allow
the more in-depth examination of the relative contributions of masculinity
and femininity to self-concept and self-esteem, teasing out the influence of
content domain (instrumentality and expressiveness) from overall desirability appraisals. Consequently, this paper describes a cross-cultural extension
and refinement of psychological androgyny research. It reports a series of
five studies including the construction of the Singapore Androgyny Inventory (SAI) and the exploration of the prominent theoretical controversies
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
19:37
532
Ward
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
19:37
Psychological Androgyny
533
of M and F should contain a mixture of positive and negative traits, the approach adopted here
was explicitly modeled on Bems work for comparative purposes. One tailed significance testing
was used for the masculine and feminine traits.
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
19:37
534
Ward
Table I. SAI Items
Feminine
Masculine
languagea
2. Active
5. Adventurous
8. Assertivea
11. Clever
14. Daring
17. Decisivea
20. Dynamic
23. Enterprising
26. Forcefula
29. Independenta
32. Intelligent
35. Masculinea
38. Powerful
41. Self-confident
44. Willing to take risksa
Neutral
3. Accommodating
6. Altruistic
9. Charismatic
12. Cool-headed
15. Easy-going
18. Ethical
21. Flexible
24. Generous
27. Humane
30. Intuitive
33. Lighthearted
36. Poised
39. Self-sacrificing
42. Thoughtful
45. Truthfula
items.
criteria, subscale items with item-total correlations <.3 were initially deleted.
Bearing in mind some of the frequent criticisms of Bems inventory, it was
then important to ensure that (a) the included traits were indeed desirable
and (b) that the desirability ratings of the composite masculine and feminine
subscales were not significantly different. For these criteria reference was
made to the previous study, and only those items with gender appropriate
(for masculine and feminine items) or overall (for neutral items) desirability
ratings of greater than 4 were included. Balancing the social desirability ratings and examining the item content for redundancies resulted in the selection of 15 masculine, 15 feminine, and 15 gender-neutral items (Table I). The
mean gender-appropriate social desirability rating was 5.3 (SD = 0.71) for
the feminine subscale and 5.1 (SD = 0 .81) for the masculine subscale; these
evaluations were not significantly different, t(168) = 1.77, ns. The overall
social desirability mean for the gender-neutral subscale was 4.7 (SD = 0.74).
Although only M and F scales are required for the assessment of androgyny, a socially desirable gender-neutral scale was retained. This served
two purposes: (1) the 15 items function as fillers to disguise the nature of
the inventory and (2) the neutral subscale can be used as a test of divergent
validity for the assessments of masculinity and femininity.
Scoring. The masculine, feminine, and gender-neutral scales are scored
unidirectionally by calculating the mean M, F, and N scores over the 15
corresponding rating scales. The scoring of androgyny may be calculated in
a number of ways. Bem (1974) originally used FM as a base androgyny
score and t-ratios to classify respondents as masculine, feminine, or androgynous. This was later abandoned with a change to a median split technique
to classify respondents with the further distinction of an undifferentiated
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
Psychological Androgyny
19:37
535
category (Bem, 1977). There has been some debate, however, as to whether
the categorization technique is preferable to the scoring of androgyny as a
continuous variable. For the continuity approach, Kalin (1979; A =[(M +
F) |(M F)|]/2), Bryan, Coleman, and Ganong (1981; A(G) = MF),
and Strahan (1975; A [S] = M + F) have formulated alternatives.4
Norms and Selected Statistics. Using the median split technique the following classifications emerged: androgynous (33.8% men and 16.0% women),
feminine (11.0% men and 39.6% women), masculine (35.7% men and 13.9%
women), and undifferentiated (19.5% men and 30.6% women). Total sample means for the Femininity, Masculinity, and the Gender-neutral subscales
were 4.38 (SD = .73), 4.32 (SD = .87), and 4.7 (SD = .61), respectively. Tests
for homogeneity of related variances indicated that the M scale variance was
greater than that of the F scale, t(309) = 3.20, p < .01.
Reliability. Subscales were found to have good internal consistency.
Cronbach alphas for the Femininity, Masculinity, and Gender-neutral traits
subscales were .80, .90, and .80, respectively.
Validity. Supporting Bems contention that masculinity and femininity
are independent clusters of traits, the correlation between these subscales
was .02. However, both masculinity and femininity were significantly correlated with the socially desirable gender-neutral items (between .53 and .59).
In keeping with Bems scoring of androgyny (FM) there was no significant
relationship between androgyny and gender-neutral desirability (.10), although Kalins (androgyny), Bryan et al.s (androgyny G), and Strahans (androgyny S) methods yielded significant correlations. As expected, both masculinity and femininity related to the scoring of androgyny as a continuous
measure; however, it is worth noting that the correlations between masculinity and androgyny were on average slightly higher (between .58 and .84) than
the correlations between femininity and androgyny (.49.72; see Table II).
T-tests revealed gender differences in mean masculinity and femininity
scores for men and women. Men (M = 4.6) scored higher in masculinity
than did women, M = 4.0; t(309) = 7.25, p < .0001, and women (M = 4.5)
scored higher in femininity than did men, M = 4.2, t(309) = 3.97, p < .0001.
The nature of the M and F subscales of the BSRI have caused various controversies in the measurement of psychological androgyny. On one
particular count it has been argued that the masculine and feminine
clusters of traits merely reflect instrumental versus expressive domains, but
that these qualities are not specifically gender-linked. An important concern, therefore, pertains to the relationship between masculinity/femininity
and the associated groupings of traits. To address this issue correlations were
performed between the masculine and feminine items of the M and F
4 In
Bryan et al.s formula G refers to geometric mean. In Strahans formula S refers to sum.
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
19:37
536
Ward
Table II. Intercorrelations among Masculine, Feminine, and
Neutral Subscales and Androgyny Scores
Scales
1. Masculinity
All respondents
Men
Women
2. Femininity
All respondents
Men
Women
3. Gender-neutral traits
All respondents
Men
Women
4. Androgyny
All respondents
Men
Women
5. Androgyny (G)
All respondents
Men
Women
6. Androgyny (S)
p
.02
.14
.10
.58
.59
.56
.70
.58
.84
.78
.79
.79
.77
.80
.76
.53
.58
.59
.49
.71
.38
.64
.71
.68
.65
.70
.72
.62
.59
.64
.77
.76
.77
.78
.77
.77
.89
.89
.88
.84
.85
.84
.99
.99
.99
< .001.
subscales with total scores from the remaining 14 subscale items. Intercorrelations of .33 ( p < .0005) for F and .51 ( p < .0005) for M resulted, suggesting a moderately strong association between expressivity and femininity
and instrumentality and masculinity. Within-gender analyses also demonstrated significant correlations; however, the patterns were slightly different.
For men the intercorrelations were .21 ( p < .01) for F and . 51 ( p < .0005)
for M whereas for women these intercorrelations were .56 ( p < .0005) and
.37 ( p < .001), respectively. Although instrumentality and expressivity do
appear to be independent domains, the masculine and feminine scale
items were negatively correlated with each other in the total sample (.61,
p < .001).
Factor analysis of the 30 gender-linked items lent support to independent masculine and feminine factors. Two robust factors emerged, the first
with an eigenvalue of 6.94, accounting for 23.1% of the variance, and a second factor with an eigenvalue of 4.43, accounting for 14.8% of the variance.
All masculine qualities loaded positively and significantly on Factor I (.52
.74). The feminine traits likewise loaded positively on Factor II (.36.69).
Four of the feminine traits (kind, loving, nice, pleasant) loaded in the range
of .38.49 on Factor I, whereas their loadings on Factor II varied between
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
19:37
Psychological Androgyny
537
Factor I
Factor II
Active
Adventurous
Assertive
Clever
Daring
Decisive
Dynamic
Enterprising
Forceful
Independent
Intelligent
Masculine
Powerful
Self-confident
Willing to take risks
Does not use harsh language
Domestic
Eager to soothe hurt feelings
Easily expresses tender emotions
Feminine
Gentle
Graceful
Innocent
Kind
Loving
Nice
Pleasant
Sensitive
Soft-spoken
Tender
.60
.60
.53
.65
.69
.64
.74
.56
.62
.58
.67
.52
.72
.71
.57
.02
.01
.22
.14
.16
.27
.09
.06
.40
.38
.43
.48
.17
.10
.19
.22
.28
.25
.05
.16
.17
.03
.06
.18
.15
.14
.38
.13
.16
.24
.39
.36
.47
.43
.52
.63
.62
.39
.53
.55
.54
.49
.39
.48
.69
.49 and .55. One of the masculine traits (masculine, .38) loaded negatively
on Factor II. (See Table III.)
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
19:37
538
Ward
Results
The subscales demonstrated good temporal stability. Test-retest correlations were .84 for Femininity, .89 for Masculinity, and .76 for the Genderneutral subscale. M and F were unrelated; intercorrelations were .04 on
the first testing and .06 on the second.
MODELS OF ANDROGYNY
Study 4: Construct Validity and Models of Psychological Androgyny
Method
Participants. The sample was composed of 205 students (91 men and
114 women). Mean age was 21.16 years (SD = 2.0). The sample was predominantly Chinese (89.8%) with smaller proportions of Malays (6.3%)
and others, including Indians and Eurasians (3.9%).
Materials and Procedure. In addition to personal information and the
45-item SAI, the questionnaire included Mehrabians Need for Achievement
(NAch) scale (Mehrabian & Bank, 1978) and Need for Affiliation (NAff)
scale (Mehrabian, 1970), the social (SSC) and personal (PSC) subscales of
the Tennessee Self-Concept Scale (TSCS; Fitts, 1965), and a 25-item Attitudes toward Women Scale (AWS) adapted from Spence and Helmreich
(1972).
NAch taps domains such as pride and optimism in work, tendency to
work hard, future orientation, doing more than is required, leadership, and
ambition. As such, it would be expected to relate to psychological masculinity. NAff, by contrast, is focused on importance of and desire for friendship, social contacts, and cooperation and would be expected to be associated with psychological femininity. The measurement of these needs assists
in the construct validation of the SAI. It should be noted that NAch and
NAff scales have previously demonstrated good reliability and validity with
Singaporeans in a study by Ang (1994).
The TSCS and AWS were included to consider the various models of
androgyny. The TSCS has been used successfully in previous unpublished research with Singaporean respondents (Cheong et al., 1986). The adaptation
of the AWS for use in Singapore, along with reliability and validity indices,
is described by Ward (1985) and subsequently used in Wards (Ward, 1988)
research.
Minor alterations were made on the various scales to ensure cultural and
linguistic appropriateness of items. All instruments utilized a Likert format;
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
19:37
Psychological Androgyny
539
higher scores represented greater need for achievement and affiliation, more
positive self-concept, and more liberal attitudes toward women, respectively.
Questionnaires were distributed in undergraduate classes; participation
was anonymous and voluntary.
Results
Preliminary analysis entailed reliability checks for the measurement instruments. Scales appeared internally consistent with the following Cronbach
alphas: NAff (.75), NAch (.86), SSC (.74), PSC (.76), AWS (.84), F (.79),
M (.90), and N (.83). The intercorrelations among the scales are presented in
Table IV. This includes the three scoring techniques for the measurement of
androgyny as a continuous variable, which are presented for cross-technique
and cross-sample comparability.
As can be noted from Table IV, femininity is significantly correlated
with the need for affiliation whereas masculinity is significantly related to
the need for achievement. Masculinity, femininity, and androgyny are significantly correlated with self-concept measures but are not linked to attitudes toward womens roles. Correlations were also undertaken for men and
women separately. These results followed a similar pattern, except that for
Table IV. Correlations of Masculinity, Femininity, and Androgyny
with Personality and Attitude Measures
NAch
Masculinity
All respondents
Men
Women
Femininity
All respondents
Men
Women
Androgyny
All respondents
Men
Women
Androgyny (G)
All respondents
Men
Women
Androgyny (S)
All respondents
Men
Women
p
.56
.46
.63
NAff
.00
.00
.05
PSC
SSC
AWS
.47
.50
.41
.40
.49
.34
.00
.14
.15
.07
.15
.05
.24
.20
.26
.22
.29
.21
.35
.41
.32
.08
.17
.06
.49
.39
.56
.05
.09
.06
.43
.47
.36
.40
.47
.35
.02
.16
.09
.48
.42
.50
.09
.11
.12
.48
.53
.42
.50
.58
.43
.05
.20
.07
.46
.41
.48
.09
.11
.12
.48
.53
.42
.50
.59
.44
.05
.20
.07
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
540
19:37
Ward
men small but significant relationships were found between femininity (.17,
p < .05), androgyny G, and androgyny S (.20, p < .03) and liberal attitudes
toward women.
Although M and F scales were unrelated (r = .11), it is worth noting
that on the whole the correlations between masculinity and other variables
were greater in magnitude than parallel associations between those variables
and femininity. Less variance was obtained on the femininity subscale (M =
4.29, SD = .68) compared to the masculinity subscale (M = 4.07, SD = .88),
which is congruent with the findings in Study 2; t(203) = 1.96, p < .05. It is
also worth noting that the three androgyny scoring techniques were highly
intercorrelated. Androgyny G and androgyny S intercorrelated at .99 while
their correlations with the third androgyny scoring technique were .90 and
.86, respectively.
In addition to scoring androgyny as a continuous variable, Bems median (weighted for gender) split technique was utilized to categorize participants as high or low in masculinity and femininity. In line with Taylor
and Halls recommendations (Taylor & Hall, 1982), analysis of variance
was then used to compare need for achievement, need for affiliation, attitudes toward womens roles, and self-concept scores; more specifically, a
2 (M) 2 (F) 2 (participant gender) analysis of variance was undertaken.
As expected M, F(1, 197) = 51.2, p < .0001, but not F, F(1, 197) = .06, ns,
produced a main effect for need for achievement. Similarly, F, F(1, 197) =
12.1, p < .001, but not M, F(1, 197) = 0.4, ns, produced a main effect for
need for affiliation. Neither main effects for participant gender (Fs < 1)
nor interaction effects (Fs < 1.3) were observed in these analyses. For AWS
scores only participant gender produced a significant effect with women
(M = 74.3, SD = 11.6) having more liberal attitudes toward womens roles
than men (M = 65.7, SD = 12.9); F(1, 197) = 23.8, p < .001. The means are
presented in Table V.
Analyses were then undertaken to investigate the masculine, additive, and interactive models of psychological androgyny and its relation
to self-concept. To reiterate, the masculine model contends that masculinity is responsible for most of the variance in indicators of self-concept.
In terms of analysis by ANOVA, this model would predict main effects
only for M. In contrast, the additive model proposes that both M and F
contribute positively and uniquely to the prediction of psychological wellbeing; consequently, it would predict main effects for both M and F on selfconcept. Finally, interactive models of androgyny argue that psychologically
androgynous individuals are relatively advantaged compared with masculine, feminine, and undifferentiated persons. The distinguishing feature of
this model is that it predicts a significant interaction effect of M and F on selfconcept. The balance-interaction model would expect this interaction without underlying main effects whereas the emergent-interaction model would
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
19:37
Psychological Androgyny
541
High
Low
46.8
(7.4)
48.5
(6.5)
45.9
(7.8)
44.4
(7.9)
45.9
(7.7)
42.8
(7.8)
49.0
(6.6)
49.3
(6.7)
48.4
(6.4)
42.6
(7.5)
42.8
(6.6)
42.6
(7.9)
46.8
(6.7)
47.5
(6.0)
46.4
(7.1)
43.5
(7.3)
44.2
(7.1)
42.7
(7.5)
47.2
(6.7)
47.2
(6.1)
47.1
(7.5)
43.3
(7.2)
42.7
(7.2)
43.6
(7.2)
119.7
(22.3)
124.3
(24.5)
117.1
(20.7)
118.9
(22.5)
121.9
(23.8)
115.7
(20.8)
130.7
(21.3)
129.1
(25.0)
133.2
(13.9)
109.7
(18.4)
112.4
(18.2)
108.4
(18.5)
93.6
(14.3)
92.8
(18.3)
94.1
(11.6)
85.8
(16.1)
84.6
(15.4)
87.1
(16.9)
88.9
(16.8)
88.4
(17.9)
89.6
(15.2)
90.2
(14.8)
86.9
(15.6)
91.6
(14.3)
71.8
(11.2)
67.7
(13.1)
74.2
(9.3)
69.2
(14.2)
64.3
(12.7)
74.4
(14.1)
70.1
(13.1)
66.5
(13.4)
75.6
(10.5)
70.8
(12.8)
64.3
(12.1)
73.6
(12.1)
a Significant
Masculinity
High
Low
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
19:37
542
Ward
self-concept. F(1, 197) = 12.4, p < .001, for masculinity and F(1, 197) =
9.8, p < .002, for femininity. In both cases those high in M or F evinced
more positive self-concepts. M F interaction effects were not significant;
F(1, 197) = 3.1, ns, for personal self-concept and F(1, 197) = 2.4, ns, for social self-concept. Neither main nor interaction effects for participant gender
on self-concept measures were observed (Fs < 1). This overall pattern of
results clearly supports the additive model of androgyny.
Results
Preliminary analysis revealed that all scales retained good internal consistency. Cronbach alphas were: M (.89), F (.82), N (.80), well-being (.86),
and self-acceptance (.71). The intercorrelations among scales are reported
in Table VI. As expected, M and F scales were unrelated (.02); however, unlike Study 4, correlations among the subscales on the androgyny inventory
and measures of mental health differed between men and women. For men,
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
19:37
Psychological Androgyny
543
Masculinity
All respondents
Men
Women
Femininity
All respondents
Men
Women
Androgyny
All respondents
Men
Women
Androgyny (G)
All respondents
Men
Women
Androgyny (S)
All respondents
Men
Women
p
Self-Acceptance
Psychological Well-Being
.62
.58
.67
.24
.41
.15
.03
.17
.06
.05
.00
.08
.46
.19
.62
.19
.29
.14
.45
.30
.55
.18
.32
.09
.44
.33
.50
.15
.31
.05
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
19:37
544
Ward
Table VII. Mean Scores on Self-Acceptance and Psychological Well-Being
Femininity
Scale
Self-Acceptancea
All respondents
Men
Women
Psychological Well-Beinga
All respondents
Men
Women
Masculinity
High
Low
High
Low
53.8
(10.8)
54.4
(10.2)
53.5
(11.1)
55.0
(10.7)
55.8
(10.5)
54.0
(10.9)
60.2
(9.4)
59.0
(9.3)
61.6
(9.5)
48.6
(8.5)
49.1
(8.6)
48.4
(8.5)
163.5
(25.0)
165.8
(27.8)
162.5
(23.7)
163.4
(19.8)
162.0
(18.2)
165.0
(21.4)
168.3
(19.8)
168.2
(21.7)
168.4
(17.4)
158.0
(23.4)
153.6
(19.9)
159.8
(24.5)
a Significant
p < .0001. No other main effects (1 < Fs < 2.8) or interaction effects (Fs
1) occurred. See Table VII for accompanying mean scores.
DISCUSSION
Taylor and Hall (1982) have maintained that androgyny researchers
face two important tasks: establishing the construct validity of their measurement instruments and assessing empirical support for their theoretical
formulations. Three studies presented here have described the construction
of a reliable and valid culture-specific instrument designed to measure psychological masculinity, femininity, and androgyny in multiethnic, Englishspeaking Singaporean samples.5 A further two studies have considered the
masculine, additive, and interactive models of androgyny in relation to selfconcept and mental health.
5 Despite
the robust psychometric properties of the SAI, limitations of the instrument should
also be noted. The scale was constructed in English for research with university students.
English usage is extensive and increasing in Singapore; it is considered the dominant working
language in Singapore and is utilized along with Malay as the primary means of interethnic
communication (Kuo, 1979, 1985). As English is one of Singapores four official languages
and the medium of instruction in schools and universities, it is the most appropriate language
choice for a multiethnic student sample. Malay and Mandarin versions of the scale could be
explored for use with older or less educated respondents or both in Singapore, and this work is
in progress. It is emphasized, however, that the SAI is a culture-specific test and that no claims
are made about its cross-cultural construct validity.
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
19:37
Psychological Androgyny
545
Psychometric Issues
Although the construction of the SAI was modeled on work by Bem
(1974), the author has eliminated many of the criticisms aimed at the instruments predecessor. The SAI demonstrated good reliability as evidenced by
assessments of the internal consistency and temporal stability of the M, F, and
N scales. The orthogonal nature of the MF subscales was corroborated, and
the unidimensional structure of the M and F factors was confirmed. In addition, the SAI includes M and F items, which are both desirable and equivalent
in terms of gender appropriate social desirability. This circumvents the biases
introduced in the BSRI by the use of relatively undesirable feminine traits.
The F and M subscales mirror expressive and instrumental domains
postulated by Spence (1984) as being the core of psychological masculinity
and femininity and as having transcultural generality. This is corroborated
by tests of convergent and discriminant validity. As expected, higher scores
were produced on the F scale by women whereas higher scores on the M scale
were found in men; in addition, significant relationships emerged between
the F scale and need for affiliation and the M scale and need for achievement. Despite the masculine/instrumental and feminine/expressive linkage,
the moderate correlations between femininity and the remaining expressive items on the F scale and between masculinity and instrumental traits
on the M scale, as well as the negative relationship between femininity and
masculinity, suggest that the gender constructs are considerably broader
and more complex and cannot be simply reduced to nurturant and agentic
domains. In this sense the Singaporean data concur with Spences position
(Spence, 1984) that masculinity and femininity constructs are multifaceted
but that expressiveness and instrumentality are major components of their
definitive features. In this case the breadth and complexity of the constructs
may permit the criticism of the SAI, like the BSRI, for limitations in content
domain and suggest certain implications for theoretical formulations concerning the relationship between masculinity, femininity, and self-concept.
Despite this potential limitation, the emphasis on instrumental/expressive
distinctions was retained in the SAI to make the cross-cultural work on models of androgyny directly comparable to research with the BSRI. This instrument has been the most widely used assessment of masculinity, femininity,
and androgyny and has furnished the empirical foundations for the theoretical models of androgyny and self-concept and psychological well-being.
Models of Psychological Androgyny
After the construction of a culturally appropriate measurement, the research was extended to investigate the relationship between psychological
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
546
19:37
Ward
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
19:37
Psychological Androgyny
547
appears to rely heavily on items that tap both neuroticism (e.g., I have nightmares every few nights; much of the time my head seems to hurt all over) and
instrumentality (e.g., I usually expect to succeed in the things I do; any person who is willing to work hard has a good chance of succeeding). Previous
research has corroborated that M is associated with lowered neuroticism
(OHeron & Orlofsy, 1990; Zeldow, Clark, & Daugherty, 1985).6 Furthermore, as masculinity itself is defined by an instrumental-agentic domain, it
is not surprising that it is strongly associated with psychological well-being.
Femininity, by contrast, is essentially defined by expressive qualities and
does not significantly contribute to the variance in psychological well-being.
Self-acceptance has been defined by Gough (1956) as a sense of self
worth and capacity for independent thought and action. This suggests that
self-acceptance is similar to self-esteem, which has primarily been associated with masculinity in the North American literature (e.g., Whitley, 1983).
However, examination of the scale items reveals that the instrument has a
decided agentic flavor, encompassing statements such as I doubt whether
I would make a good leader, and At times I have worn myself out by undertaking too much. Given the instrumental bend to self-acceptance, again
it is not surprising that masculinity, but not femininity, is associated with
increased psychological adjustment.
Theoretical and Methodological Issues Revisited
The cross-cultural extension reported here has offered a broader perspective on theoretical and methodological controversies in androgyny research. It has expanded the investigation of masculine, additive, and interactive models of androgyny to a new cultural milieu and has generated an
original, culturally appropriate instrument for the assessment of masculinity, femininity, and psychological androgyny. The approach has provided
the opportunity to tease out the influences of extraneous variables that have
plagued monocultural researchers. In this context a major advantage of these
studies is found in their capacity to assess critically the contention that differential valuing of masculine and feminine attributes is primarily responsible
for the more positive and robust effect of masculinity on self-concept and
mental health.
Research in Singapore has provided a natural avenue for the construction of measurement scales that incorporate equivalent desirability ratings of
feminine and masculine traits. This circumvents the allegations of measurement artifact leveled at the American BSRI research that has highlighted
6 Subsequent
research in Singapore with the SAI has also linked M to lowered levels of neuroticism (Ang & Ward, 1994).
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
548
19:37
Ward
the significance of masculinity for self-concept and mental health. But the
method of scale construction in this cultural context has also avoided counter
criticisms pertaining to the consequences of artificially inflating the relative
desirability of feminine traits for androgyny theory and research (Taylor &
Hall, 1982). Consequently, the research allows the unbiased examination
of the association between instrumental and expressive traits and personality and mental health. In this context findings with the SAI support a
differentiated additive model of androgyny. Both femininity and masculinity contribute positively and significantly to personal and social self-concept;
however, masculinity alone retains an influence on psychological well-being
and self-acceptance. Given both theoretical and methodological considerations, then, the differential outcomes of M and F cannot be reduced to
desirability ratings.
Measurement artifact due to social desirability of M and F scales is not
the only psychometric issue that has been raised by critics of the BSRI. Taylor
and Hall (1982) have also argued that variability, not social desirability, is
the most plausible psychometric explanation for the dominant masculinity effect on psychological well-being. Unfortunately, this research cannot
adequately address the issue of variance and its contribution to the more
robust masculinity effects. As with the BSRI, the SAI M scale consistently
produces more variance than the F scale, and correlations of greater magnitude between masculinity, compared to femininity, and self-concept are
observed. Although it is likely that scale variance influences the relationship
between femininity and measures of self-concept and mental health to some
degree, overall, the theoretical, rather than psychometric, rationale for F
effects appears more compelling.
Despite the merits of a cross-cultural replication and refinement of psychological androgyny research and the persuasive empirical evidence supporting the differentiated additive model of androgyny, there are broader
conceptual criticisms about androgyny research in general that should be
acknowledged. First, the majority of investigations of androgyny have essentially examined the impact of gender-linked instrumental and expressive
personality traits. Second, psychological advantages that accrue to androgynous individuals occur because of the additive influences of these instrumental and expressive qualities. As expressiveness and instrumentality represent single dimensions of masculinity and femininity, and as measures of
self-concept are sometimes affected by the conjoint influence of expressivity and instrumentality, it may be argued that the term androgyny is both
inaccurate and superfluous. Indeed, Taylor and Hall (1982) make this point
disclaiming the existence of an independent entity termed androgyny. It
may be that future research should be geared toward studies of instrumentality and expressivity rather than masculinity and femininity. Either way,
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
19:37
Psychological Androgyny
549
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
550
19:37
Ward
Blackman, S. (1982). Comments on three methods of scoring androgyny as a continuous variable. Psychological Reports, 51, 11001102.
Bryan, L., Coleman, M., & Ganong, L. (1981). Geometric mean as a continuous measure of
androgyny. Psychological Reports, 48, 691694.
Campbell, T., Gillaspy, J. A., & Thompson, B. (1997). The factor structure of the Bem Sex Role
Inventory: Confirmatory analysis of long and short forms. Educational and Psychological
Measurement, 57, 118124.
Cheong, C., Chow, C.-Y., Chua, G., Mohammed Dawood, K., Kwek, A., Loh, J., & Woo,
M.-L. (1986). Psychological adjustment of married and unmarried graduates. Unpublished
manuscript, National University of Singapore.
Chung, B. (1996). The construct validity of the Bem Sex Role Inventory for heterosexual and
gay men. Journal of Homosexuality, 30, 8797.
Clammer, J. (1982). The institutionalization of ethnicity: The culture of ethnicity in Singapore.
Ethnic and Racial Studies, 5, 127139.
Cook, E. P. (1985). Psychological androgyny. New York: Pergamon.
Cook, E. P. (1987). Psychological androgyny: A review of the research. The Counseling Psychologist, 15, 471513.
Deutsch, C. J., & Gilbert, L. A. (1976). Sex role stereotypes: Effect on perceptions of self and
others and on personal adjustment. Journal of Counseling Psychology, 23, 373379.
Erdwins, C., Small, A., & Gross, R. (1980). The relationship of sex role to self-concept. Journal
of Clinical Psychology, 36, 111115.
Fitts, W. (1965). Tennessee Self-Concept manual. Nashville: Counselor Recordings and Tests.
Flaherty, J. F., & Dusek, J. B. (1980). An investigation of the relationship between psychological
androgyny and components of self-concept. Journal of Personality and Social Psychology,
38, 984992.
Francis, L. J., & Wilcox, C. (1998). The relationship between Eysencks personality dimensions
and Bems masculinity and femininity scales revisited. Personality and Individual Differences, 25, 683687.
Gaudreau, P. (1977). Factor analysis and the Bem Sex Role Inventory. Journal of Consulting
and Clinical Psychology, 45, 299302.
Gough, H. G. (1956). California Psychological Inventory manual. Palo Alto, CA: Consulting
Psychologists Press.
Hall, J. A., & Taylor, M. C. (1985). Psychological androgyny and the masculinity femininity
interaction. Journal of Personality and Social Psychology, 49, 429435.
Holt, C. L., & Ellis, J. B. (1998). Assessing the current validity of the Bem Sex Role Inventory.
Sex Roles, 39, 929941.
Kalin, R. (1979). Method for scoring androgyny as a continuous variable. Psychological Reports,
44, 12051206.
Kelly, J. A., Furman, W., & Young, V. (1978). Problems associated with the typological measurement of sex roles and androgyny. Journal of Consulting and Clinical Psychology, 46,
15741576.
Kimlicka, T., Cross, H., & Tarnai, J. (1983). A comparison of androgynous, feminine, masculine and undifferentiated women on self-esteem, body satisfaction and sexual satisfaction.
Psychology of Women Quarterly, 7, 291294.
Kuo, E. C. Y. (1979). Measuring communicativity in multilingual societies: The cases of Singapore and West Malaysia. Anthropological Linguistics, 21, 328340.
Kuo, E. C. Y. (1985). Language identity: The case of Chinese in Singapore. In W. Tseng & D.
Wu (Eds.), Chinese culture and mental health (pp. 181192) New York: Academic Press.
Lai, A.-E. (1995). Meanings of multiethnicity: A case study of ethnicity and ethnic relations in
Singapore. Kuala Lumpur: Oxford University Press.
Lamke, L. K. (1982). The impact of sex-role orientation on self-esteem in early adolescence.
Child Development, 53, 15301535.
Lee, A. G., & Scheurer, V. L. (1983). Psychological androgyny and aspects of self-image in
women and men. Sex Roles, 9, 289306.
Lorenzo-Cioldi, F. (1996). Psychological androgyny: A concept in search of lesser substance.
Journal for Theory of Social Behavior, 26, 137155.
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
Psychological Androgyny
19:37
551
Lubinski, D., Tellegen, A., & Butcher, J. N. (1983). Masculinity, femininity and androgyny
viewed and assessed as distinct concepts. Journal of Personality and Social Psychology, 44,
428439.
Lundy, A., & Rosenberg, J. A. (1987). Androgyny, masculinity and self-esteem. Social Behavior
and Personality, 15, 9195.
Markstrom-Adams, C. (1989). Androgyny and its relationship to adolescent psychosocial wellbeing: A review of the literature. Sex Roles, 21, 325340.
Marsh, H. W. (1987). Masculinity, femininity and androgyny: Their relations with multiple
dimensions of self-concept. Multivariate Behavioral Research, 22, 91118.
Marsh, H. W., & Byrne, B. (1991). Differentiated additive androgyny model: Relations between
masculinity, femininity, and multiple dimensions of self-concept. Journal of Personality and
Social Psychology, 61, 811828.
Marsh, H. W., & Myers, M. R. (1986). Masculinity, femininity and androgyny: A methodological
and theoretical critique. Sex Roles, 14, 397430.
Mehrabian, A. (1970). The development and validation of measures of affiliative tendency and
sensitivity to rejection. Educational and Psychological Measurement, 30, 417428.
Mehrabian, A., & Bank, L. (1978). A questionnaire measure of individual differences in achieving tendency. Educational and Psychological Measurement, 38, 475478.
Moreland, J., Gulanick, N., Montague, E., & Harren, V. (1978). Some psychometric properties of the Bem Sex Role Inventory. Applied Psychological Measurement, 2, 249
256.
OHeron, C. A., & Orlofsy, J. L. (1990). Stereotypic and nonstereotypic sex role trait and
behavior orientations, gender identity and psychological adjustment. Journal of Personality
and Social Psychology, 58, 134143.
Pedhazur, E. J., & Tetenbaum, T. J. (1979). The Bem Sex-Role Inventory: A theoretical and
methodological critique. Journal of Personality and Social Psychology, 37, 9961016.
Ramanaiah, N. V., & Martin, H. J. (1984). Convergent and discriminant validity of selected
masculinity and femininity scales. Sex Roles, 10, 493504.
Ruch, L. O. (1984). Dimensionality of the Bem Sex Role Inventory: A multi-dimensional analysis. Sex Roles, 10, 99117.
Shifren, K., & Bauserman, E. (1996). The relationship between instrumental and expressive traits, health behaviors and perceived physical health. Sex Roles, 34, 841
864.
Silvern, L. E., & Ryan, V. L. (1979). Self-rated adjustment and sex-typing on the Bem Sex Role
Inventory: Is masculinity the primary predictor of adjustment? Sex Roles, 5, 739763.
Spence, J. T. (1984). Masculinity, femininity and gender related traits: A conceptual analysis and
critique of current research. In B. A. Maher & W. B. Maher (Eds.), Progress in experimental
personality research (Vol. 13), (pp. 197). New York: Academic Press.
Spence, J. T., & Helmreich, R. (1972). The Attitudes toward Women Scale: An objective instrument to measure the rights and roles of women in contemporary society. Journal Supplement Abstract Service, 2, 6667.
Spence, J. T., Helmreich, R., & Stapp, J. (1974). The Personal Attributes Questionnaire: A
measure of sex role stereotypes and masculinity-femininity. Journal Supplement Abstract
Service, 4, 4344.
Spence, J. T., Helmreich, R. L., & 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.
Strahan, R. F. (1975). Remarks on Bems measurement of psychological androgyny: Alternative
methods and a supplementary analysis. Journal of Consulting and Clinical Psychology, 43,
568571.
Strahan, R. F. (1981). Some remarks on scoring androgyny as a continuous variable. Psychological Reports, 49, 887890.
Taylor, M. C., & Hall, J. A. (1982). Psychological androgyny: A review and reformulation of
theories, methods and conclusions. Psychological Bulletin, 92, 347366.
Thompson, B. (1989). Meta-analysis of factor structure studies: A case example with Bems
androgyny measure. Journal of Experimental Education, 57, 187197.
P1: vendor/LOV
P2: FYJ/FOP
QC: FLF
PP029-292515
552
19:37
Ward