Вы находитесь на странице: 1из 16

See

discussions, stats, and author profiles for this publication at:


https://www.researchgate.net/publication/232476252

Confirmatory Factor Analysis of the


Conformity to Masculine Norms
Inventory and Development of the
Conformity to Mas....

Article in Psychology of Men & Masculinity · July 2009


DOI: 10.1037/a0015481

CITATIONS READS

87 2,719

2 authors, including:

Mike C. Parent
University of Texas at Austin
39 PUBLICATIONS 584 CITATIONS

SEE PROFILE

All content following this page was uploaded by Mike C. Parent on 22 June 2015.

The user has requested enhancement of the downloaded file.


Psychology of Men & Masculinity © 2009 American Psychological Association
2009, Vol. 10, No. 3, 175–189 1524-9220/09/$12.00 DOI: 10.1037/a0015481

Confirmatory Factor Analysis of the Conformity to Masculine


Norms Inventory and Development of the Conformity to Masculine
Norms Inventory-46
Mike C. Parent and Bonnie Moradi
University of Florida

The Conformity to Masculine Norms Inventory (CMNI; Mahalik et al., 2003) has
proven to be an important tool in advancing the study of masculinity. The authors of
the present study conducted the first factor analysis of the CMNI since its creation and
also developed a short form of the measure to facilitate its broader utility. Confirmatory
factor analysis of data from 229 men supported nine of the originally proposed CMNI
factors, but also suggested the removal of two factors (Dominance and Pursuit of
Status) that demonstrated relatively poor construct specificity, low factor loadings, and
weak reliability coefficients in this study and in prior research. The data also supported
elimination of low-loading items from other factors, resulting in an abbreviated version
of the CMNI with improved model fit, acceptable reliability coefficients, and high
correlations with corresponding original-form subscales. The CMNI-46 is discussed as
an efficient tool for assessing men’s conformity to masculine norms.

Keywords: masculine norms, masculine gender conformity, gender role attitudes,


reliability and structural validity, confirmatory factor analysis

The study of gender roles—the “patterns of tudes (Wilkinson, 2004), poor dietary decisions
behavior that are culturally expected of ‘nor- (Levi, Chan, & Pence, 2006), and, in children,
mal’ men and women” (Brehm, Miller, Per- propensity to bully other children (Gini & Poz-
lman, & Campbell, 2002, p. 21)— has pro- zoli, 2006). Such findings highlight the impor-
gressed from focusing on gender differences to tance of assessing conformity to masculine
examining more complex relations of gender- norms in psychological research and practice.
related ideologies with psychological well- Some measures of masculine gender role
being and distress (Smiler, 2006). Within this constructs focus on stress and conflict associ-
context, the study of masculinity has flourished, ated with adherence to traditional masculine
and conceptual models and instruments have norms (e.g., Masculine Gender Role Stress
been advanced to capture various aspects of Scale [MGRS], Eisler & Skidmore, 1987; Gen-
men’s masculine gender role conformity. In- der Role Conflict Scale [GRCS], O’Neil,
deed, masculine gender role-related constructs Helms, Gable, David, & Wrightsman, 1986).
have been linked to a host of psychological and Measures that focus on the strains of masculin-
physical issues for men, including reluctance to ity differ from measures that assess endorse-
seek psychological help (Berger, Levant, ment of masculine ideology (for review, see
McMillan, Kelleher, & Sellers, 2005; Pender- Smiler, 2004). Indeed, Mahalik et al. (2003)
son & Vogel, 2007), expressions of hostility observed that conformity to masculine norms
(Jakupak, Tull, & Roemer, 2005), antigay atti- may not be negative in all contexts, and could
be adaptive in some situations. As such, a mea-
sure of conformity to a range of masculine
norms would be a useful compliment to existing
Mike C. Parent and Bonnie Moradi, Department of Psy- strain measures. Also, therapeutic approaches,
chology, Counseling Psychology Program, University of such as gender role analysis, call for helping
Florida. clients to examine adaptive and maladaptive
Correspondence concerning this article should be ad-
dressed to Mike C. Parent, Department of Psychology, Box consequences of gender role conformity and
112250, University of Florida, Gainesville, FL 32611. nonconformity (Brown, 1990; Philpot, Brook-
E-mail: michael.parent@ufl.edu ers, Lusterman, & Nutt, 1997). Thus, a measure
175
176 PARENT AND MORADI

that assesses level of conformity to a broad viduals’ experiences of these gender role norms
range of masculine norms would have clinical and their level of conformity to such norms are
and research utility. shaped by individual and group factors (e.g.,
To address the need for such a measure, personality, race/ethnicity), and there are costs
Mahalik et al. (2003) developed the Conformity and benefits for conforming and not conforming
to Masculine Norms Inventory (CMNI). The to gender role norms. The CMNI assesses con-
CMNI assesses level of conformity to mascu- formity to 11 masculine norms (see Table 1);
line norms and, as such, differs from measures these norms emerged from a series of rational
of the conflict or stress associated with such instrument construction methods and psycho-
norms (e.g., MGRS, GRCS). Furthermore, the metric evaluation. Specifically, Mahalik et al.
CMNI subsumes norms reflected in other mas- (2003) reviewed prior literature on traditional
culine norm conformity measures—mainly masculinity to identify dominant cultural mas-
those assessed by Thompson and Pleck’s (1986) culinity norms, reasoning that “the expectations
Male Role Norms Scale (i.e., status, toughness, of masculinity as constructed by Caucasian,
antifemininity) and Levant et al.’s (1992) Male middle- and upper-class heterosexuals should
Role Norms Inventory (i.e., avoidance of fem- affect members of that groups and every other
ininity, restrictive emotionality, achievement/ male in U.S. society who is held up to those
status, aggression, self-reliance, homophobia, standards” (p. 5). These norms were discussed
nonrelational hypersexuality)—and, as dis- over an eight-month period in weekly focus
cussed next, it assesses additional norms as well groups of counseling psychology graduate stu-
(i.e., power over women, risk taking, primacy of dents (three European American men, three Euro-
work). pean American women, one Haitian Canadian
woman, and one Asian American man). The
Development and Use of the CMNI focus groups refined the norms, ultimately
agreed on 12 norms, and developed items to
The CMNI is grounded in Mahalik’s (de- assess those norms. Mahalik et al. (2003) pi-
scribed in Mahalik et al., 2003) gender role loted these items with small samples to ensure
norms model, which posits that socially domi- their interpretability.
nant groups shape the gender role norms that are These item development steps were followed
communicated to individuals in a society. Indi- by psychometric evaluation with data from a

Table 1
Masculine Norms Assessed by the Conformity to Masculine Norms Inventory (CMNI)
Subscale name Description Sample item
Emotional Control Emotional restriction and suppression “I tend to keep my feelings to myself”
Winning Drive to win “In general, I will do anything to win”
Playboy Desire for multiple or noncommitted sexual “If I could, I would frequently change sexual
relationships and emotional distance partners”
from sex partners
Violence Proclivity for physical confrontations “Sometimes violent action is necessary”
Self-reliance Aversion to asking for assistance “I hate asking for help”
Risk-taking Penchant for high-risk behaviors “I frequently put myself in risky situations”
Power Over Women Perceived control over women at both “In general, I control the women in my life”
personal and social levels
Dominance General desire to have personal control “In general, I must get my way”
over situations
Primacy of Work Viewing work as a major focus of life “My work is the most important part of my
life”
Pursuit of Status Being pleased with being thought of as “It feels good to be important”
important
Disdain for Aversion to the prospect of being gay, or “I would be furious if someone thought I
Homosexualsa being thought of as gay was gay”
a
Considerations regarding renaming the “Disdain for Homosexuals” subscale to “Heterosexual Self-presentation” are
described in the Discussion section.
FACTOR STRUCTURE OF THE CMNI 177

predominantly White college student sample tive behaviors such as exercising to cope with
(Mahalik et al., 2003). Specifically, an initial depression and greater independence in daily
exploratory factor analysis (EFA) resulted in living tasks following hospitalization for brain
retention of 11 of the 12 masculine norms or spinal injury (Good et al., 2006; Mahalik &
(Physical Toughness was not supported, but as- Rochlen, 2006). These findings highlight that
pects of Physical Toughness are reflected in multidimensional assessment of masculine
retained factors such as Risk-taking, Emotional norms can reveal both adaptive and maladaptive
Control, and Violence). Subsequent EFAs of correlates of such norms.
the same data resulted in reduction of the item
pool from 144 to 94 items. Scores on the 11 Need for Reexamination of the Structure
masculine norm subscales yielded mostly low of CMNI Data
to moderate intercorrelations, supporting the
multidimensionality of the CMNI. Also, CMNI In light of its demonstrated promise, further
total scores were correlated moderately with refinement of the CMNI to advance its psycho-
GRCS and MGRS total scores, supporting the metric strengths and increase its practical utility
distinctiveness of the constructs assessed by is important for future research and practice
these measures. Cronbach’s alphas for CMNI with men. Specifically, the factor structure of
subscale items ranged from .72 to .91, and it CMNI data has not been examined beyond the
was .92 for all items. Two- to three-week test– EFA conducted during its initial development.
retest reliability coefficients for CMNI sub- Further attention to the structural properties of
scales were variable, ranging from .51 to .96, the CMNI is needed for three reasons. First,
with a median value of .80. Thus, initial reli- following EFA, confirmatory factor analysis
ability and validity evidence was promising in a (CFA) is recommended for assessing cross-
primarily White student sample. sample structural stability of data produced by a
Since its development, the CMNI has been measure in order to guide further theoretical and
used to study adaptive and maladaptive corre- measurement refinement (Worthington & Whit-
lates of men’s adherence to masculine norms. taker, 2006). Second, the 94-item length of the
Across samples of predominantly White men, CMNI may represent a response burden that
CMNI dimensions have been linked with rape prohibits its use in research and practice. In-
myth acceptance, health-risk behaviors such as deed, study-specific strategies for shortening the
smoking and problem drinking, relationship CMNI are emerging, with researchers using the
dissatisfaction, negative attitudes toward help- one or two highest-loading items of each CMNI
seeking, greater propensity to cope with depres- subscale to generate an abbreviated CMNI total
sion through drinking, and lower propensity to score (e.g., Mahalik, Burns, et al., 2007; Roch-
cope with depression through talking with a len, McKelley, Suizzo, & Scaringi, 2008). But
wife, partner, or mental health professional these strategies lose the CMNI’s multidimen-
(Burn & Ward, 2005; Good et al., 2006, 2008; sionality and have yielded lower-than-accept-
Locke & Mahalik, 2005; Mahalik, Burns, & able reliability coefficients. These low reliabil-
Syzdek, 2007; Mahalik & Rochlen, 2006). ity coefficients are likely a result of collapsing
Links of CMNI scores with health risk behav- conceptually distinct dimensions of masculinity
iors, such as substance use, also have been that, as discussed previously, have low to mod-
replicated with Asian American, Australian, erate intercorrelations. Thus, a short form of the
and Kenyan men (Liu & Iwamoto, 2007; CMNI that retains the empirical strengths of the
Mahalik, Lagan, & Morrison, 2006; Mahalik, original version would be a useful addition to
Levi-Minzi, & Walker, 2007). Beyond such the literature.
maladaptive correlates, in predominantly White Relatedly, CMNI subscales vary widely in
samples, CMNI dimensions have been linked length, ranging from 4 to 12 items. The overall
with college students’ vocational interests length and variability in subscale length of the
(Mahalik, Perry, Coonerty-Femiano, Catraio, & CMNI resulted from retaining items in the ini-
Land, 2006), interacted with type of cancer tial EFA that met absolute cutoffs for factor
treatment to predict self-reported physical well- loadings (greater than .40) and cross-loadings
being of men with prostate cancer (Burns & (less than .30; Mahalik et al., 2003). Most items
Mahalik, 2008), and were correlated with adap- met these criteria, but retained items varied in
178 PARENT AND MORADI

the magnitude of their factor loadings (.40 to Levi-Minzi, et al., 2007; Mahalik et al., 2003;
.79). Thus, strong and distinct factor indicators Mahalik & Rochlen, 2006; Tager & Good,
were retained along with relatively weaker and 2005), although Smiler (2006) reported an alpha
structurally more ambiguous indicators. CFA of greater than .70 (but did not report the exact
the CMNI can guide retention of the strongest value). Similar concerns are evident about Pur-
and most clear factor indicators to yield sub- suit of Status as well. Specifically, in the instru-
scales that are more clear and parsimonious ment development study, 4 of the 12 original
reflections of the underlying constructs. Such Pursuit of Status items loaded onto other factors
instrument length optimization can prevent pro- or did not load onto any factor. Also, across
liferation of study-specific abbreviations of the samples, Pursuit of Status items have yielded
CMNI and provide a standard short form that low reliabilities, with Cronbach’s alphas in the
retains the depth and breadth of the constructs low .60s and .70s in two studies (Mahalik et al.,
assessed. 2003; Smiler, 2006), but an alpha of .80 in a
Finally, CFA of the CMNI data can guide third study (Tager & Good, 2005). Thus, con-
refinements to address conceptual and psycho- sistent with the validity problems for Domi-
metric limitations evident in prior research. nance and Pursuit of Status, structure and reli-
Specifically, Mahalik et al. (2003) observed ability problems challenge the interpretability
problems with the Dominance and Pursuit of of these two subscales as cohesive and distinc-
Status subscales. First, Dominance and Pursuit tive masculine norms.
of Status scores were not correlated signifi- In contrast to the conceptual and psychomet-
cantly with social dominance orientation, which ric concerns about Dominance and Pursuit of
was a convergent validity indicator. Second, Status, for the other nine CMNI subscales, items
Pursuit of Status scores were generally uncor- loaded more cleanly onto their intended factors
related with scores on another measure of mas- in the instrument development sample, and
culine norms; by contrast, other CMNI subscale Cronbach’s alphas have typically been in the
scores were correlated more consistently with .80s and .90s across samples (Liu & Iwamoto,
scores on this convergent validity indicator 2007; Mahalik, Levi-Minzi, et al., 2007;
(Mahalik et al., 2003). Third, across four age Mahalik et al., 2003; Mahalik & Rochlen, 2006;
cohorts (undergraduates to older adults), Pursuit Tager & Good, 2005). In terms of validity,
of Status was the only subscale that did not CMNI subscale scores, other than Pursuit of
differentiate between women and men for any Status, were correlated with total scores and
age group (Smiler, 2006). Thus, tests of the with scores on multiple conceptually relevant
validity of Dominance and Pursuit of Status subscales of the masculine norms convergent
challenge their conceptualization as specific di- validity indicator (Mahalik et al., 2003). Thus,
mensions of masculine norms. structural and convergent validity evidence
These conceptual and validity problems with from the instrument development sample, as
Dominance and Pursuit of Status were com- well as reliability evidence across samples, sup-
pounded with structural and reliability prob- ports most of the CMNI subscales. But, the
lems. Specifically, in Mahalik et al.’s (2003) aforementioned psychometric and conceptual
measurement development study, 8 of the 12 considerations suggest the need to reevaluate
items intended for the Dominance subscale the Dominance and Pursuit of Status subscales
loaded primarily onto other factors, cross- and reexamine the structure of the CMNI.
loaded to an unacceptable degree, or did not
meet the factor loading cutoff for any factor. Summary and Purpose of the
This pattern suggests that Dominance was dif- Present Study
fuse and subsumed into other CMNI factors. In
fact, only 4 of the original 12 Dominance items The CMNI is a promising tool for assessing
were retained, and these retained items had fac- masculine norm conformity. Empirically
tor loadings that were lower (.40 –.43) than grounded measurement refinement that can re-
those of items for other factors. Low reliability tain the depth and breadth of the constructs
is also a concern across samples for Dominance assessed by the CMNI, and also optimize its
items, with Cronbach’s alphas generally in the reliability and structural properties, can facili-
.50s and .60s (Liu & Iwamoto, 2007; Mahalik, tate continued attention to masculine norms in
FACTOR STRUCTURE OF THE CMNI 179

research and practice with men. The present cated that CMNI means in the present sample
study addresses these aims through two specific were comparable to those in Mahalik et al.’s
objectives. First, CFA is used to investigate the (2003) sample of men. Specifically, of the 12
replicability of Mahalik et al.’s (2003) factor subscale and total scale scores, only two dif-
structure for the CMNI. Second, findings are fered significantly across samples; Mahalik et
used to inform model and instrument refinement al.’s (2003) sample’s mean for Playboy
and to examine the prospect of reducing the was 1.39 points (of 36 possible points) lower,
overall length of the CMNI to make it more t(979) ⫽ ⫺3.09, p ⬍ .05, d ⫽ .24, and its mean
versatile and suitable for use in research and for Power over Women was 1.97 points (of 27
practice. possible points) higher, t(979) ⫽ 6.09, p ⬍
.001, d ⫽ .48, than that of the present sample.
Method Thus, means were comparable across the two
samples (see Table 2).
Participants
Procedure
Data from 229 undergraduate men from a
large Canadian university were analyzed in the A total of 249 men were recruited for this
present study. Participants ranged in age study and received credit in their introduction to
from 18 to 45 years (M ⫽ 19.95, SD ⫽ 3.23, psychology course in exchange for participa-
Mdn ⫽ 19.00). Participants identified as White tion. Before beginning the survey, participants
(59%), Asian or Asian American (23%), Afri- received and signed consent forms indicating
can or African American (4%), Hispanic or that they would be completing a questionnaire
Hispanic American (1%), Biracial or Multira- about masculinity. Administrations occurred
cial (4%), other race/ethnicity (7%), or did not with 15 to 25 people at a time and in a room
report their identification (3%).1 Sexual orien- large enough to place each person at a separate
tation was assessed using a free-response ques- table. To provide fair opportunity for earning
tion, allowing for open-ended expression of course credit, women were offered a different
identification; 35 participants chose not to dis- questionnaire. After completing the survey, par-
close a sexual orientation, and 17 provided re- ticipants received a debriefing form and the
sponses that were not codeable either because of researchers’ contact information for follow-up
apparent misunderstanding of the question (e.g., questions. Based on preliminary data screening
“only with my wife”) or idiosyncrasy of re- (detailed next), 20 cases were removed, result-
sponse (e.g., “sexual”). Of the participants who ing in a sample size of 229 for the present
disclosed a codeable sexual orientation, 92% analyses.
identified as heterosexual or attracted to
women, 5% identified as attracted to both sexes
Results
but primarily to women, 2% identified as bisex-
ual, less than 1% identified as attracted to both Suitability for Confirmatory
sexes but primarily to men, and 1% identified as Factor Analysis
gay or attracted to men.
We first examined the data for missing val-
Instruments ues. Five participants missing more than 20% of
data on a subscale were dropped from analyses.
CMNI. Participants completed the CMNI For the remaining participants, the individual
(Mahalik et al., 2003). The 94 CMNI items are participant’s subscale mean was substituted for
rated on a 4-point scale (0 ⫽ strongly disagree missing data in the corresponding subscale.
to 3 ⫽ strongly agree). Appropriate items were This procedure is acceptable when the volume
reverse-scored, and item ratings were added to
obtain subscale and total scale scores. Cron- 1
bach’s alphas with the present sample were For this Canadian sample “American” refers to North
American. We recognize that some Canadian individuals
similar to those that Mahalik et al. (2003) ob- might prefer the modifier “Canadian” to “American” but we
tained (see Table 2). Bonferroni-corrected t chose to use the identity labels reflected in the response
tests, with alpha adjusted to .004 (.05/12), indi- choices on the demographic questionnaire.
180 PARENT AND MORADI

of missing data is low and reliability is good

0.91
0.88
0.88
0.84
0.85
0.82
0.87
0.76
0.90
0.73
0.72
0.94

Mahalik et al.
(Schafer & Graham, 2002); advantages of this

0.89 14.89 5.66


0.86 16.91 5.10
0.85 12.06 6.05
0.86 12.38 3.96
6.63 2.81
0.85 16.58 3.61
0.83 10.59 4.46
8.97 3.28
0.91 17.74 6.65
5.84 1.88
0.69 11.85 2.43
0.92 134.45 24.64
procedure over other methods are also demon-
(2003)
SD
strated (Dodeen, 2003). One data point was
replaced for 20 participants, two data points
M

were replaced for four participants, and three


data points were replaced for three participants.
0.85

0.76

0.63

The data met guidelines for univariate nor-


Current study

mality as outlined by Weston and Gore (2006).


15.54 4.93
16.54 4.68
13.45 5.65
12.05 4.17
6.95 3.01
16.32 3.99
8.62 3.65
9.15 3.29
18.71 6.22
6.15 1.67
11.77 2.17
135.24 22.27
SD

In terms of multivariate normality, 15 cases had


Mahalanobis distances that were significant at
p ⬍ .001, suggesting that they were multivariate
M

outliers (Tabachnick & Fidell, 1996); these 15


0.38ⴱⴱ cases were removed from the data prior to
11

analyses.
The final sample size of 229 exceeded
0.44ⴱⴱ
0.50ⴱⴱ

Kline’s (2005) recommendation of at least


10
Descriptive Statistics and Correlations for the 94-Item Conformity to Masculine Norms Inventory (CMNI)

200 cases for CFA but did not meet Bentler


and Chou’s (1987) recommended minimum
0.13 0.27ⴱⴱ

0.14ⴱ 0.59ⴱⴱ
0.12 0.10

of five cases per parameter estimated. Never-


9

theless, other researchers have averred that


0.30ⴱⴱ ⫺0.11

sample sizes of 200 are adequate for CFA


8

(e.g., Quintana & Maxwell, 1999). Also,


models with greater degrees of freedom re-
0.31ⴱⴱ

0.61ⴱⴱ
0.10

0.08

quire smaller sample sizes to achieve higher


7

power than do models with fewer degrees of


0.18ⴱⴱ

0.21ⴱⴱ
0.18ⴱⴱ
0.26ⴱⴱ
0.48ⴱⴱ

freedom (MacCallum, Browne, & Sugawara,


⫺0.07
6

1996). Indeed, MacCallum et al. (1996) re-


ported that as degrees of freedom approach
0.31ⴱⴱ
0.14ⴱ

0.24ⴱⴱ ⫺0.16ⴱ

100, with a sample size of 200, power is


0.01
0.08
0.00

0.05
5

between .87 and .96. The degrees of freedom


for the least complex model tested in the
0.32ⴱⴱ
0.33ⴱⴱ

0.23ⴱⴱ
0.27ⴱⴱ

0.60ⴱⴱ
0.15ⴱ

present study (df ⫽ 953) was substantially


⫺0.10
4

higher than 100, thus the present sample size


was adequate for the analyses.
0.28ⴱⴱ

0.20ⴱⴱ
0.41ⴱⴱ

0.56ⴱⴱ
0.16ⴱ
0.16ⴱ
0.14ⴱ ⫺0.02

0.03
0.08
3

Confirmatory Factor Analysis


0.22ⴱⴱ
0.32ⴱⴱ

0.31ⴱⴱ
0.27ⴱⴱ

0.37ⴱⴱ
0.41ⴱⴱ
0.36ⴱⴱ
0.67ⴱⴱ
0.04
2

Maximum likelihood estimation was used to


conduct CFAs. To evaluate model fit, we fol-
0.23ⴱⴱ
0.23ⴱⴱ
0.25ⴱⴱ
0.36ⴱⴱ

0.22ⴱⴱ

0.21ⴱⴱ

0.54ⴱⴱ

lowed recommendations to report the ␹2 statis-


0.03

0.00

0.13
0.01
1

tic with degrees of freedom and significance


level, the root mean square error of approxima-
Disdain for Homosexuals

tion (RMSEA) with 90% confidence interval


Power Over Women

p ⬍ .01

(CI), the standardized root-mean-square resid-


Emotional Control

Primacy of Work

Pursuit of Status

ual (SRMR), and the comparative fit index


Subscale

Self-reliance

(CFI; e.g., Kline, 2005; Worthington & Whit-


CMNI total
Risk-taking

Dominance

ⴱⴱ
Violence

taker, 2006). For sample sizes of less than 500,


Winning
Playboy

p ⬍ .05.

RMSEA and SRMR values of less than .10 and


Table 2

CFI values greater than .90 suggest an accept-


1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.

able fit (Weston & Gore, 2006).



FACTOR STRUCTURE OF THE CMNI 181

CFA of the Original Structure of items were eliminated. But, the number of items
the CMNI was still variable across subscales in this
trimmed measure. Specifically, Emotional Con-
The first model tested included all CMNI trol retained nine items and Disdain for Homo-
items, loading onto their intended latent con- sexuals retained seven items, whereas the other
struct according to Mahalik et al.’s (2003) 11- subscales retained between four and six items.
factor model, with latent factors allowed to cor- Because we were interested in optimal subscale
relate. The chi-square statistic for this model length and Emotional Control and Disdain for
was ␹2(4222, N ⫽ 229) ⫽ 7089.30, p ⬍ .001. Homosexuals each had more high-loading items
Acceptable data-model fit was suggested by the than did the other subscales, we eliminated
RMSEA (.055, 90% CI ⫽ .052, .057) and three additional lowest-loading Emotional Con-
SRMR (.0774), but unacceptable fit was sug- trol items (items 19, 27, and 93, with loadings
gested by the CFI (.71). The relatively poorer of .63, .63, and .62, respectively) and one addi-
CFI is consistent with simulation study findings tional lowest-loading Disdain for Homosexuals
that, in correctly specified models, CFI tends to item (item 63, with a loading of .64). These item
decline whereas absolute fit indices (e.g., eliminations resulted in more uniform subscale
RMSEA, SRMR) tend to improve with in- lengths of four to six items and a total of 46
creases in the number of variables in the model items for the CMNI-46.
(Kenny & McCoach, 2003). Still, the low CFI Table 5 presents subscale correlations,
value challenges the fit of the original 11-factor means, standard deviations, and Cronbach’s al-
CMNI model. Table 3 presents item factor load- phas for the CMNI-46, as well as the correlation
ings and uniquenesses, and Table 4 presents between corresponding original and short form
latent variable intercorrelations. Items loaded subscales. This shortening of the measure resulted
significantly onto their intended factors, but in acceptable data-model fit on all fit indexes,
Dominance and Pursuit of Status items gener- ␹2(953, N ⫽ 229) ⫽ 1414.30, p ⬍ .001;
ally had lower loadings than did items loading RMSEA ⫽ .046, 90% CI ⫽ .041, .051;
onto other factors. These lower loadings of SRMR ⫽ .0591; CFI ⫽ .90. Factor loadings and
Dominance and Pursuit of Status items were uniquenesses for items on the CMNI-46 are
consistent with the lower Cronbach’s alphas presented in Table 6, and latent variable inter-
(see Table 1) for items on these subscales (.63 correlations are presented in Table 4.
and .69) relative to that for other subscale items
(mostly in the .80s). Evaluating Item Elimination Decisions
Against Mahalik et al.’s (2003) Findings
Exploring a Short Form of the CMNI
To evaluate the stability of our factor load-
Consistent with Worthington and Whittaker’s ings and resultant item elimination decisions,
(2006) suggestion to examine factor loadings as we compared our factor loadings against the
a criterion for item retention in instrument loadings obtained by Mahalik et al. (2003) (de-
length optimization, we considered as candi- tails obtained from J. R. Mahalik, personal com-
dates for deletion those items that had the low- munication, April 16, 2008). Some variability in
est loadings on their intended factor. Specifi- factor loadings is expected across our CFA of
cally, we retained items with loadings of .60 or the 94-item CMNI and Mahalik et al.’s (2003)
greater on their intended factors. This criterion EFA, because the factor loadings that Mahalik
was chosen because it resulted in retention of et al. (2003) reported were from an EFA of 132
the strongest indicators of the latent constructs items (that is, before low-loading items were
of interest while, for most subscales, also satis- deleted). As such, factor loadings for the 94
fied conventional guidelines for a sufficient retained CMNI items might have shifted
number of items (i.e., three or more; Worthing- slightly after Mahalik et al. (2003) eliminated
ton & Whittaker, 2006). Exceptions were that poorer items. Nevertheless, consistency be-
this step resulted in the elimination of the Dom- tween Mahalik et al.’s (2003) findings and our
inance and Pursuit of Status subscales because findings would provide additional support for
most of their items did not meet the .60 cutoff. our item elimination decisions to form the
Overall, a total of 44 relatively lower-loading CMNI-46. There was substantial consistency in
182 PARENT AND MORADI

Table 3
Eleven-Factor Solution for the 94-Item Conformity to Masculine Norms Inventory (CMNI), Proposed by
Mahalik et al. (2003)
Mahalik Mahalik Mahalik
et al. Current study et al. Current study et al.
Current study
Item Item Item
number Loading Loading Uniqueness number Loading Loading Uniqueness number Loading Loading Uniqueness
Emotional Control items Violence items Dominance items
52 ⫺0.79 0.79 0.37 44 0.65 0.80 0.37 18 ⫺0.43 0.60 0.64
77 ⫺0.77 0.76 0.42 79 0.64 0.79 0.37 94 ⫺0.43 0.56 0.68
65 ⫺0.74 0.74 0.45 62 0.57 0.73 0.47 6 ⫺0.41 0.56 0.68
43 ⫺0.55 0.68 0.53 25 0.49 0.65 0.57 69 ⫺0.40 0.48 0.77
36 ⫺0.61 0.68 0.54 15 0.67 0.63 0.60 Primacy of Work items
88 ⫺0.63 0.66 0.57 68 0.64 0.62 0.62 30 0.64 0.72 0.49
19 ⫺0.63 0.63 0.61 34 0.55 0.55 0.70 64 0.56 0.69 0.53
27 ⫺0.73 0.63 0.61 4 0.53 0.50 0.75 76 0.57 0.68 0.54
93 ⫺0.61 0.62 0.62 Self-reliance items 49 0.66 0.66 0.56
1 ⫺0.50 0.56 0.69 85 0.62 0.80 0.36 84 0.50 0.45 0.79
14 ⫺0.54 0.40 0.84 10 0.62 0.76 0.42 56 0.41 0.39 0.84
Winning items 53 0.62 0.70 0.51 20 0.48 0.39 0.85
48 0.70 0.73 0.46 74 0.51 0.68 0.54 8 0.46 0.34 0.88
54 0.71 0.69 0.52 29 0.53 0.67 0.55 Pursuit of Status items
39 0.68 0.68 0.53 45 0.62 0.56 0.69 50 0.42 0.66 0.56
22 0.61 0.65 0.58 Risk-taking items 26 0.51 0.58 0.66
67 0.68 0.65 0.58 24 0.77 0.78 0.39 87 0.44 0.51 0.74
2 0.61 0.62 0.62 40 0.64 0.72 0.48 7 0.49 0.51 0.74
89 0.57 0.59 0.66 17 0.65 0.70 0.51 59 0.43 0.44 0.81
75 0.53 0.58 0.66 60 0.60 0.70 0.52 41 0.55 0.43 0.81
31 0.73 0.55 0.69 70 0.47 0.67 0.56 Disdain for Homosexuals items
12 0.56 0.54 0.71 82 0.42 0.55 0.70 51 ⫺0.77 0.89 0.20
Playboy items 55 0.44 0.53 0.72 37 ⫺0.70 0.87 0.24
47 0.67 0.80 0.36 32 0.52 0.52 0.73 42 ⫺0.66 0.85 0.27
3 0.77 0.79 0.38 11 0.41 0.49 0.76 73 ⫺0.70 0.73 0.47
72 0.65 0.79 0.38 92 0.55 0.49 0.76 16 ⫺0.67 0.71 0.49
33 0.64 0.62 0.62 Power Over Women items 91 ⫺0.63 0.68 0.54
28 0.60 0.59 0.65 81 0.48 0.73 0.46 63 ⫺0.60 0.64 0.59
58 0.57 0.58 0.66 86 0.53 0.73 0.47 5 ⫺0.75 0.58 0.66
38 0.50 0.58 0.67 46 0.47 0.62 0.62 80 ⫺0.49 0.49 0.76
83 0.45 0.55 0.70 61 0.58 0.62 0.62 23 ⫺0.60 0.48 0.77
90 0.57 0.44 0.81 35 0.57 0.58 0.67
78 0.45 0.35 0.88 57 0.58 0.57 0.68
66 0.45 0.34 0.88 71 0.49 0.54 0.71
13 0.55 0.27 0.93 21 0.49 0.53 0.72
9 0.50 0.47 0.78
Note. All factor loadings for the current data are significant at p ⬍ .01. Item numbers reflect numbering in 94-item CMNI.
Fit indexes: ␹2(4222, N ⫽ 229) ⫽ 7089.30, p ⬍ .001; root mean square error of approximation ⫽ .055, 90% confidence
interval ⫽ .052, .057; standardized root-mean-square residual ⫽ .0774; comparative fit index ⫽ .71.

Mahalik et al.’s (2003) factor loadings and the correspondence suggests that items eliminated
loadings obtained with the present sample (see based on the present CFA were largely those
Table 3). As in our data, Mahalik et al.’s (2003) that were also the lowest loaders in Mahalik et
factor loadings for Dominance and Pursuit of al.’s (2003) EFA. There were three cases
Status were in the magnitude of the lowest where a relatively high loading item in Ma-
loadings for the other factors. Also noteworthy halik et al.’s findings was eliminated based on
is the near perfect correspondence across the a relatively low loading in our sample (i.e.,
two samples in the within-subscale rank order items 27, 31, 5). Evaluation of these items
of items based on loading magnitudes. This indicated that their content was captured by
FACTOR STRUCTURE OF THE CMNI 183

Table 4
Latent Factor Intercorrelations for the 94-Item Conformity to Masculine Norms Inventory (CMNI) and the
CMNI-46
Subscale 1 2 3 4 5 6 7 8 9 10
ⴱⴱ ⴱⴱ ⴱⴱ ⴱ ⴱ
1. Emotional Control 0.23 0.12 0.21 0.40 ⫺0.01 0.17 ⫺0.02 0.19
2. Winning 0.25ⴱⴱ 0.26ⴱⴱ 0.34ⴱⴱ 0.16ⴱ 0.24ⴱⴱ 0.33ⴱⴱ 0.04 0.38ⴱⴱ
3. Playboy 0.18ⴱ 0.25 ⴱⴱ
0.31ⴱⴱ 0.00 0.27ⴱⴱ 0.44ⴱⴱ 0.07 0.10
4. Violence 0.26ⴱⴱ 0.35ⴱⴱ 0.31ⴱⴱ 0.14 0.32ⴱⴱ 0.34ⴱⴱ ⫺0.18ⴱ 0.16ⴱ
5. Self-reliance 0.40ⴱⴱ 0.16ⴱ ⫺0.02 0.17ⴱ 0.04 0.05 ⫺0.04 0.14
6. Risk-taking 0.02 0.31ⴱⴱ 0.26ⴱⴱ 0.35ⴱⴱ 0.02 0.22ⴱⴱ ⫺0.06 0.16ⴱ
7. Power Over Women 0.23ⴱⴱ 0.33ⴱⴱ 0.46ⴱⴱ 0.37ⴱⴱ 0.08 0.20ⴱ 0.17ⴱ 0.38ⴱⴱ
8. Primacy of Work ⫺0.01 0.04 0.07 ⫺0.15 ⫺0.03 ⫺0.10 0.15 ⫺0.13
9. Disdain for Homosexuals 0.23ⴱⴱ 0.40ⴱⴱ 0.09 0.20ⴱⴱ 0.15 0.18ⴱ 0.36ⴱⴱ ⫺0.12
10. Dominance 0.15 0.54ⴱⴱ 0.26ⴱⴱ 0.33ⴱⴱ 0.09 0.25ⴱⴱ 0.43ⴱⴱ 0.23ⴱ 0.32ⴱⴱ
ⴱⴱ ⴱ
11. Pursuit of status 0.00 0.48 0.22 0.30 ⫺0.20ⴱ
ⴱⴱ
0.31ⴱⴱ 0.08 0.14 0.06 0.64ⴱⴱ
Note. Values below the diagonal are for the 94-item CMNI, values above the diagonal are for the CMNI-46.

p ⬍ .05. ⴱⴱ p ⬍ .01.

retained items. Specifically, the content of Comparability of the CMNI-Original and


item 27, “I love to explore my feelings with CMNI-46
others,” was reflected in other Emotional Con-
trol items such as item 52, “I like to talk about We evaluated the comparability of the
my feelings”; the content of item 31, “Winning CMNI-46 with the original form by examining
isn’t everything, it’s the only thing,” was re- consistency of Cronbach’s alphas, conceptual
flected in other Winning items such as item 48, content of subscales, and magnitudes of corre-
“It is important for me to win”; and the content lations between corresponding subscales of the
of item 5, “It is important to me that people two forms. As indicated in Tables 2 and 5,
think I am heterosexual,” was reflected in other Cronbach’s alphas remained stable after item
Disdain for Homosexuals items such as item 91, reduction, with differences between the original
“I try to avoid being perceived as gay.” Thus, in and short form alphas ranging from ⫺.05 to
each of these cases, the item content was re- ⫹.01. According to Ponterotto and Ruckde-
tained in the shortened CMNI-46 subscale. schel’s (2007) matrix for estimating adequacy

Table 5
Descriptive Statistics and Correlations for the Conformity to Masculine Norms Inventory (CMNI)-46
Subscale 1 2 3 4 5 6 7 8 9 M SD ␣ ra
1.
Emotional Control 8.65 3.06 .86 .96
2.
Winning 0.20ⴱⴱ 9.99 2.98 .83 .95
3.
Playboy 0.12 0.22ⴱⴱ 5.06 2.86 .84 .89
4.
Violence 0.18ⴱⴱ 0.30ⴱⴱ 0.28ⴱⴱ 10.14 3.41 .86 .97
5.
Self-reliance 0.36ⴱⴱ 0.13 ⫺0.01 0.10 6.19 2.65 .84 .98
6.
Risk-taking ⫺0.02 0.22ⴱⴱ 0.25ⴱⴱ 0.26ⴱⴱ 0.02 7.44 2.36 .84 .95
7.
Power Over
Women 0.16ⴱ 0.28ⴱⴱ 0.37ⴱⴱ 0.28ⴱⴱ 0.03 0.21ⴱⴱ 3.77 1.95 .78 .90
8. Primacy of Work ⫺0.02 0.02 0.02 ⫺0.16ⴱ ⫺0.04 ⫺0.05 0.13 4.26 2.04 .77 .93
9. Disdain for
Homosexuals 0.18ⴱⴱ 0.34ⴱⴱ 0.08 0.16ⴱ 0.12 0.16ⴱ 0.33ⴱⴱ ⫺0.10 11.05 4.28 .91 .96
10. CMNI 46 S total 0.51ⴱⴱ 0.63ⴱⴱ 0.51ⴱⴱ 0.58ⴱⴱ 0.39ⴱⴱ 0.44ⴱⴱ 0.59ⴱⴱ 0.09 0.60ⴱⴱ 66.55 12.81 .88 .96
a
Values reflect correlations between corresponding original and short form subscales.

p ⬍ .05. ⴱⴱ p ⬍ .01.
184 PARENT AND MORADI

Table 6
Nine Factor Solution for the Conformity to Masculine Norms Inventory (CMNI)-46
Item Item Item
number Loading Uniqueness number Loading Uniqueness number Loading Uniqueness
Emotional Control items Violence items Power Over Women items
52 0.79 0.38 44 0.82 0.33 86 0.79 0.38
77 0.76 0.42 79 0.78 0.40 81 0.72 0.48
65 0.75 0.43 62 0.71 0.49 61 0.62 0.61
36 0.69 0.53 15 0.65 0.57 46 0.62 0.62
43 0.68 0.54 25 0.65 0.58 Primacy of Work items
88 0.66 0.57 68 0.63 0.60 64 0.72 0.49
Winning items Self-reliance items 30 0.71 0.50
54 0.76 0.43 85 0.81 0.35 76 0.67 0.55
48 0.75 0.44 10 0.77 0.41 49 0.62 0.62
39 0.73 0.47 53 0.69 0.52 Disdain for Homosexuals items
67 0.61 0.62 29 0.68 0.54 51 0.91 0.17
22 0.61 0.63 74 0.66 0.56 37 0.88 0.23
2 0.58 0.67 Risk-taking items 42 0.85 0.28
Playboy items 24 0.80 0.36 73 0.72 0.49
47 0.84 0.29 40 0.71 0.49 16 0.71 0.50
72 0.81 0.34 60 0.70 0.51 91 0.67 0.55
3 0.80 0.37 17 0.69 0.52
33 0.61 0.63 70 0.67 0.55
Note. All factor loadings are significant at p ⬍ .01. Item numbers reflect numbering in 94-item CMNI.
Fit indexes: ␹2(953, N ⫽ 229) ⫽ 1414.30, p ⬍ .001; root mean square error of approximation ⫽ .046, 90% confidence
interval ⫽ .041, .051; standardized root-mean-square residual ⫽ .0591; comparative fit index ⫽ .90.

of Cronbach’s alpha, alphas for the CMNI-46 literature on masculinity by providing the first
subscale items were all good or excellent. We CFA of the CMNI to evaluate the replicability
also reviewed the content of eliminated items of its structure. Overall, the CFA provided ten-
against that of items retained in the CMNI-46 to tative support for the multidimensional struc-
see whether item eliminations resulted in loss of ture of the CMNI proposed by Mahalik et al.
unique item content. This review confirmed that (2003). The data also pointed to areas for im-
the conceptual content of all eliminated items provement that addressed weaknesses observed
was subsumed in the retained items. Thus, the in previous studies, and these improvements led
original conceptual meaning of the nine re- to creating a more parsimonious short form of
tained subscales was preserved in the CMNI-46. the CMNI that can facilitate its future use.
This conceptual consistency between the origi- The present CFA results supported the via-
nal CMNI and the CMNI-46 was also supported bility of nine of the 11 original CMNI factors.
by the magnitudes of observed correlations be- But, relatively lower factor loadings and lower
tween the original form and corresponding Cronbach’s alphas in the present study raised
CMNI-46 subscale scores (r ⫽ .89 to .98; see questions about retaining the Dominance and
Table 5). The conceptual consistency and very Pursuit of Status factors. These concerns are
high correlations between the two forms sug- consistent with loading and cross-loading prob-
gest that previously obtained validity data for lems for Dominance and Pursuit of Status items
the original CMNI subscales are applicable to in the instrument development study (Mahalik
the CMNI-46 subscales. et al., 2003) and with low Cronbach’s alphas
found for these subscale items in subsequent
Discussion samples (e.g., Liu & Iwamoto, 2007; Mahalik,
Levi-Minzi, et al., 2007; Mahalik et al., 2003;
The CMNI is a promising tool for research Mahalik & Rochlen, 2006; Tager & Good,
and practice aiming to understand the role of 2005).
multiple dimensions of masculine norms in Low alphas, low factor loadings, misload-
men’s lives. The present study contributes to the ings, and cross-loadings for Dominance and
FACTOR STRUCTURE OF THE CMNI 185

Pursuit of Status items across samples are con- In addition to informing refinement of
sistent with a number of validity and conceptual the conceptual and empirical framework for the
concerns about these two subscales as well. In CMNI, the present study also yielded the
particular, Dominance and Pursuit of Status CMNI-46. We were able to reduce the length of
may be too broad to constitute unique masculine the CMNI while retaining the depth and breadth
norms, and other CMNI subscales may capture of multiple subscales. Specifically, the short
their masculine norm-specific manifestations. form retained 9 of the 11 original CMNI sub-
For example, Power over Women involves ex- scales (all but Dominance and Pursuit of Status)
erting dominance over women, Violence in- and reduced overall length of the measure by
volves physical dominance, and Winning in- nearly 60%. The CMNI-46 subscale items also
volves competitive dominance and pursuit of retained acceptable reliability in the present
status. Consistent with such conceptual overlap, sample; this is particularly noteworthy given
the highest latent factor intercorrelations in the that Cronbach’s alpha is associated positively
full CMNI model (see Table 4) reflected overlap with number of items. Furthermore, in the
between Dominance and Pursuit of Status (r ⫽ present sample, the CMNI-46 yielded superior
.64), and overlap of Winning with Pursuit of model-data fit than did the original form of the
Status (r ⫽ .48) and Dominance (r ⫽ .54). As CMNI. Importantly, the CMNI-46 retains the
such, dominance and pursuit of status may re- intended conceptual and empirical multidimen-
flect broad substrates of specific masculine sionality of the original CMNI, a feature that is
norms that are reflected in other CMNI sub- lost in short forms that aggregate selected
scales. This interpretation is further supported CMNI items to compute total scores (e.g.,
by the fact that Dominance and Pursuit of Status Mahalik, Burns, et al., 2007; Rochlen et al.,
did not emerge as clear constructs and most of 2008). Thus, at less than half the length of the
their items were subsumed into other factors in original form, with the present sample, the
CMNI-46 yielded acceptable structural validity
instrument development. Thus, domain specific
and reliability, and also retained the multidi-
manifestations of dominance and pursuit of sta-
mensionality of the original CMNI.
tus may be reflected in other CMNI subscales.
The present findings also indicated very high
Additional validity concerns about Domi-
correlations (in the .90s) between corresponding
nance and Pursuit of Status also are important to CMNI- 46 and original form scores. These high
consider. Specifically, contrary to expectation, correlations imply consistency in the constructs
neither Dominance nor Pursuit of Status scores assessed across the two forms and in the relative
were correlated significantly with social domi- rank order of participants along CMNI and
nance orientation, a convergent validity indica- CMNI-46 scores. Such high correlations are
tor (Mahalik et al., 2003). Also, Pursuit of Sta- consistent with our conceptual review of
tus was generally uncorrelated with subscale dropped items which indicated that their content
scores on the measure of masculinity norms was reflected in retained items. Also, at the
selected to assess convergent validity (Mahalik subscale level, the items deleted from the short
et al., 2003). Finally, across age cohorts ranging form corresponded well with low-loading items
from undergraduate to older adult, Pursuit of in the original CMNI development study. These
Status did not differentiate between women and correspondences between the two forms suggest
men, raising questions about its validity as a that validity evidence accumulated for the orig-
masculine norm (Smiler, 2006). Taken together, inal CMNI is applicable to the CMNI-46.
these validity concerns across samples, along Still, the present study did not examine con-
with present and prior data highlighting struc- vergent validity beyond the high correlations
ture and reliability problems with Dominance between CMNI-46 and original form scores,
and Pursuit of Status, suggest that these two and further research is needed to evaluate the
factors should be removed, or at a minimum, validity of CMNI-46 scores across diverse sam-
interpreted with caution. Eliminating these two ples. For instance, tests of CMNI-46 scores’
factors seems reasonable given that the under- independence from impression management
lying constructs tapped by these two subscales and convergence with other measures of mas-
overlap substantially with retained subscales culine ideology (for examples, see Smiler,
(e.g., Winning). 2004) can evaluate discriminant and convergent
186 PARENT AND MORADI

validity. Also, qualitative exploration of mascu- plore how gender norms impact their function-
line norms across cultural groups and examina- ing and presenting problems (Brown, 1990;
tion of links of acculturation and enculturation Philpot et al., 1997). Given its brevity, the
with CMNI-46 scores could test the cultural CMNI-46 might be particularly advantageous to
specificity or transferability of the norms cap- use in such interventions. The CMNI-46 also
tured by this measure. Similarly, longitudinal can be used to evaluate the extent to which
data on the links of CMNI-46 scores with ac- gender role analysis interventions are effective
culturation and enculturation, particularly in im- in modifying conformity to masculine norms,
migrant populations, could reveal developmen- and the extent to which modifications in con-
tal processes in acquisition of U.S. masculinity formity to masculine norms are related to inter-
norms. Such research to test the validity of the vention outcomes (e.g., reduced distress).
CMNI-46 and its underlying conceptual frame-
work across diverse populations is important for Limitations and Future Directions
elucidating how masculine norms are experi-
enced by men of diverse backgrounds (e.g., The present findings can inform research and
race/ethnicity, sexual orientation, socioeco- practice in a number of ways, but these findings
nomic status, age), and for identifying culture- should be interpreted in light of some limita-
specific masculine norms that could be incorpo- tions. First, the present data were collected from
rated to broaden the utility of the CMNI-46 mostly White Canadian college students and are
across groups. As Mahalik et al. (2003) noted, not immediately generalizable outside of this
the masculine norms measured by the CMNI, population. Previous findings suggest similari-
and consequently those assessed with the ties in mean CMNI scores across predominantly
CMNI-46, may not be an exhaustive list of such White men of varying ages (Smiler, 2006).
norms. Researchers may do well to identify, Building on such findings, multiple cohort and
validate, and incorporate other categories of longitudinal studies can explore the structural
masculine norms into the CMNI-46. stability of the CMNI-46 across cohorts and
Overall, the present findings establish prelim- time. In addition, the present Canadian sample’s
inary support for the CMNI-46 as an efficient means were generally similar to means from
tool for assessing and understanding the role of Mahalik et al.’s (2003) U.S. sample. In another
conformity to masculine norms in men’s lives. study, a Kenyan sample’s mean CMNI scores
Specifically, use of the CMNI-46 can help min- also were similar to that of Mahalik et al.’s
imize length of questionnaire packages and re- (2003) sample (Mahalik et al., 2006); but, cross-
sponse burden in order to prevent acquiescence, cultural differences in CMNI means emerged
boredom, or loss of motivation for clients or between a U.S. and Italian sample (Tager &
research participants. Given the importance of Good, 2005). Further cross-cultural research is
efficiency in clinical assessment, the CMNI-46 needed to explore the structural stability of the
may prove particularly useful in practice. As CMNI-46 with different racial/ethnic, sexual
prior studies have shown, masculine gender orientation, and other cultural groups. Given
conformity is linked to such counseling issues that most prior research with the CMNI has
as treatment outcomes (Good et al., 2006), strat- involved samples of predominantly White, pre-
egies for coping with depression (Mahalik & sumably heterosexual, and middle to upper-
Rochlen, 2006), attitudes toward help-seeking middle class men, establishing structural stabil-
(Berger et al., 2005), and vocational interests ity across groups is important for informing use
(Mahalik et al., 2006). Thus, knowledge about a of the CMNI with diverse client and research
client’s adherence to masculine norms can aid populations.
case conceptualization and treatment planning. Another limitation is that the present sample
To this end, Mahalik, Talmadge, Locke, and size met minimal standards for the analyses, but
Scott (2005) discussed how the CMNI can be further research is needed to evaluate support
used with clients to explore level of conformity for the present factor structure across larger
to various masculine norms, as well as the costs samples. Larger samples might also facilitate
and benefits of such conformity in different life examination of structural stability across racial/
contexts. Also, gender role analysis has been ethnic, sexual orientation, age, and other sub-
highlighted as important to helping clients ex- groups to the extent that sufficient diversity on
FACTOR STRUCTURE OF THE CMNI 187

these dimensions is represented within the sam- In sum, we hope that the CMNI-46 can facil-
ple. Also, despite the fact that all fit index itate continuing attention to masculine norms
values in the present study met recommended and their operationalization through streamlin-
cut-offs (Weston & Gore, 2006), CFI was at the ing the existing subscales and facilitating inte-
cutoff. Thus, evidence of acceptable model-data gration of new subscales as needed. This mod-
fit across samples would be useful. ularity would serve to save researchers from
We offer two additional issues to consider in developing redundant measures and also facili-
future work with the CMNI and CMNI-46. tate comparisons across studies. Masculine
First, CMNI total scores were computed and norms evolve across cultures and times, and the
examined in relation to other variables in the CMNI-46 can provide a useful template for
original measurement development study and in continued revision and refinement of the opera-
subsequent work (e.g., Burn & Ward, 2005; tionalization of such norms.
Good et al., 2006; Liu & Iwamoto, 2007;
Mahalik et al., 2003, Mahalik, Levi-Minzi, et
al., 2007; Smiler, 2006). As such, we reported References
Cronbach’s alphas for overall CMNI and
CMNI-46 items in the present study. But, the Bentler, P. M., & Chou, C.-P. (1987). Practical issues
in structural equation modeling. Sociological
CMNI and its underlying masculine norms are Methods and Research, 16, 78 –117.
conceptualized as multidimensional, and find- Berger, J. M., Levant, R., McMillan, K. K., Kelleher,
ings of the measurement development study and W., & Sellers, A. (2005). Impact of gender role
the present study support such multidimension- conflict, traditional masculine ideology, alexithy-
ality. Indeed, many intercorrelations of CMNI mia, and age on men’s attitudes toward psycholog-
subscales as well as those for the CMNI-46 ical help seeking. Psychology of Men & Masculin-
subscales and latent factors are nonsignificant or ity, 6, 73–78.
small (see Tables 2, 4, and 5). The multidimen- Brehm, S. S., Miller, R. S., Perlman, D., & Campbell,
sionality of masculine norms suggests that re- S. M. (2002). Intimate relationships (3rd ed.). New
searchers and practitioners should consider the York: McGraw-Hill.
Brown, L. S. (1990). Taking account of gender roles
dimensions of masculine norms that they are in the clinical assessment interview. Professional
interested in, and use corresponding CMNI or Psychology: Research and Practice, 21, 12–17.
CMNI-46 subscales to assess those specific con- Burn, S. M., & Ward, A. Z. (2005). Men’s confor-
structs (rather than use total scores). When the mity to traditional masculinity and relationship
full set of masculinity dimensions are of inter- satisfaction. Psychology of Men & Masculinity, 4,
est, the CMNI-46 may be appealing given its 254 –263.
brevity. Burns, S. M., & Mahalik, J. R. (2008). Treatment
A second consideration is the subscale label type and emotional control as predictors of men’s
“Disdain for Homosexuals.” We used this label self-assessed well-being following treatment for
to be consistent with prior research. But, we prostate cancer. Psychology of Men & Masculin-
ity, 9, 55– 66.
recommend renaming this subscale. First, the Committee on Lesbian and Gay Concerns, American
American Psychological Association recom- Psychological Association. (1991). Avoiding het-
mends against using “homosexuals” as a noun erosexual bias in language. American Psycholo-
to describe groups of people (Committee on gist, 46, 973–974.
Lesbian & Gay Concerns, 1991). Second, the Dodeen, H. M. (2003). Effectiveness of valid mean
original and abbreviated CMNI Disdain for Ho- substitution in treating missing data in attitude
mosexuals items assess the importance of pre- assessment. Assessment & Evaluation in Higher
senting oneself as heterosexual or not as gay Education, 28, 505–513.
(e.g., “I make sure people think I am heterosex- Eisler, R. M., & Skidmore, J. R. (1987). Masculine
ual,” “It would be awful if someone thought I gender role stress. Behavior Modification, 11,
123–136.
was gay”). In fact, none of the original items on Gini, G., & Pozzoli, T. (2006). The role of mascu-
this subscale directly assess affect toward les- linity in children’s bullying. Sex Roles, 54, 585–
bians or gay men, as “Disdain for Homosexu- 588.
als” implies. Thus, to more closely represent the Good, G. E., Schopp, L. H., Thomson, D., Hathaway,
content of this subscale, we suggest renaming S., Mazurek, M. O., & Sanford-Martens, T. C.
this subscale “Heterosexual Self-presentation.” (2008). Men with serious injuries: Relations
188 PARENT AND MORADI

among masculinity, age, and alcohol use. Rehabil- Norms Inventory. Psychology of Men & Masculin-
itation Psychology, 53, 39 – 45. ity, 4, 3–25.
Good, G. E., Schopp, L. H., Thomson, D., Hathaway, Mahalik, J. R., Perry, J. C., Coonerty-Femiano, A.,
S., Sanford-Martens, T., Mazurek, M. O., et al. Catraio, C., & Land, L. N. (2006). Examining
(2006). Masculine roles and rehabilitation out- Conformity to Masculinity Norms as a function of
comes among men recovering from serious inju- RAISEC vocational interests. Journal of Career
ries. Psychology of Men & Masculinity, 7, 165– Assessment, 14, 203–213.
176. Mahalik, J. R., & Rochlen, A. B. (2006). Men’s
Jakupak, M., Tull, M. T., & Roemer, L. (2005). likely responses to clinical depression: What are
Masculinity, shame, and fear of emotions as pre- they and do masculinity norms predict them? Sex
dictors of men’s expressions of anger and hostility. Roles, 55, 659 – 667.
Psychology of Men & Masculinity, 6, 275–284. Mahalik, J. R., Talmadge, W. T., Locke, B. D., &
Kenny, D. A., & McCoach, D. B. (2003). Effect of Scott, R. P. (2005). Using the Conformity to Mas-
the number of variables on measures of fit in culine Norms Inventory to work with men in a
structural equation modeling. Structural Equation clinical setting. Journal of Clinical Psychol-
Modeling, 10, 333–351. ogy, 61, 661– 674.
Kline, R. B. (2005). Principles and practice of struc- O’Neil, J. M., Helms, B., Gable, R., David, L., &
tural equation modeling (2nd ed.). New York: Wrightsman, L. (1986). Gender Role Conflict
Guilford Press. Scale: College men’s fear of femininity, Sex
Levant, R. F., Hirsh, L., Celantano, E., Cozza, T., Roles, 14, 335–350.
Hill, S., MacEcheron, M., et al. (1992). The male Penderson, E. L., & Vogel, D. L. (2007). Male gen-
role: An investigation of contemporary norms. der role conflict and willingness to seek counsel-
Journal of Mental Health Counseling, 14, 325– ing: Testing a mediation model on college-aged
337. men. Journal of Counseling Psychology, 54, 373–
Levi, A., Chan, K. K., & Pence, D. (2006). Real men 384.
Philpot, C. L., Brookers, G. R., Lusterman, D. D., &
do not read labels: The effects of masculinity and
Nutt, R. L. (1997). Bridging separate gender
involvement on college students’ food decisions.
worlds: Why men and women clash and how ther-
Journal of American College Health, 55, 91–98.
apists can bring them together. Washington, DC:
Liu, W. M., & Iwamoto, D. K. (2007). Conformity to
American Psychological Association.
masculine norms, Asian values, coping strategies,
Ponterotto, J. G., & Ruckdeschel, D. E. (2007). An
peer group influences, and substance use among
overview of coefficient alpha and a reliability ma-
Asian American men. Psychology of Men & Mas-
trix for estimating adequacy of internal consis-
culinity, 8, 25–39. tency coefficients with psychological research
Locke, B. D., & Mahalik, J. R. (2005). Examining measures. Perceptual and Motor Skills, 105, 997–
masculine norms, problem drinking, and athletic 1014.
involvement as predictors of sexual aggression in Quintana, S. M., & Maxwell, S. E. (1999). Implica-
college men. Journal of Counseling Psychol- tions of recent developments in structural equation
ogy, 52, 279 –283. modeling for counseling psychology research. The
MacCallum, R. C., Browne, M. W., & Sugawara, Counseling Psychologist, 27, 485–527.
H. M. (1996). Power analysis and determination of Rochlen, A. B., McKelley, R. A., Suizzo, M., &
sample size for covariance structure models. Psy- Scaringi, V. (2008). Predictors of relationship sat-
chological Methods, 1, 130 –149. isfaction, psychological well-being, and life satis-
Mahalik, J. R., Burns, S. M., & Syzdek, M. (2007). faction among stay-at-home fathers. Psychology of
Masculinity and perceived normative health be- Men & Masculinity, 9, 17–28.
haviors as predictors of men’s health behaviors. Schafer, J. L., & Graham, J. W. (2002). Missing data:
Social Science & Medicine, 64, 2201–2209. Our view of the state of the art. Psychological
Mahalik, J. R., Lagen, H. D., & Morrison, J. A. Methods, 7, 147–177.
(2006). Health behaviors and masculinity in Ken- Smiler, A. P. (2004). Thirty years after the discovery
yan and U.S. male college students. Psychology of of gender: Psychological concepts and measures of
Men & Masculinity, 7, 191–202. masculinity. Sex Roles, 50, 15–26.
Mahalik, J. R., Levi-Minzi, M., & Walker, G. (2007). Smiler, A. P. (2006). Conforming to masculine
Masculinity and health behaviors in Australian norms: Evidence for validity among adult men and
men. Psychology of Men & Masculinity, 8, 240 – women. Sex Roles, 54, 767–775.
249. Tabachnick, B. G., & L. S. Fidell, L. S. (1996). Using
Mahalik, J. R., Locke, B. D., Ludlow, L. H., Diemer, multivariate statistics. New York: Harper Collins.
M. A., Scott, R. P., Gottfried, M., et al. (2003). Tager, D., & Good, G. E. (2005). Italian and Amer-
Development of the Conformity to Masculine ican masculinities: A comparison of masculine
FACTOR STRUCTURE OF THE CMNI 189

gender role norms. Psychology of Men & Mascu- tudes. Psychology of Men & Masculinity, 5, 121–
linity, 6, 264 –274. 131.
Thompson, E. H., & Pleck, J. H. (1986). The struc- Worthington, R. L., & Whittaker, T. A. (2006). Scale
ture of male role norms. American Behavioral development research: A content analysis and rec-
Scientist, 29, 531–543. ommendations for Best Practices. The Counseling
Weston, R., & Gore, P. A., Jr. (2006). A brief guide Psychologist, 34, 806 – 838.
to structural equation modeling. The Counseling
Psychologist, 34, 719 –751. Received September 17, 2008
Wilkinson, W. W. (2004). Authoritarian hegemony, Revision received January 24, 2009
dimensions of masculinity, and male antigay atti- Accepted January 25, 2009 䡲

E-Mail Notification of Your Latest Issue Online!


Would you like to know when the next issue of your favorite APA journal will be available
online? This service is now available to you. Sign up at http://notify.apa.org/ and you will be
notified by e-mail when issues of interest to you become available!

View publication stats

Вам также может понравиться