english 135 paper gender discrimination week 2

maeae9

gender_discrimination_file_3.pdf

Home >English homework help >english 135 paper gender discrimination week 2

Differentiating Gender Discrimination and Sexist Behavior: An Examination

of Antecedents and Outcomes

Christopher D. Nye University of Illinois at Urbana–Champaign, Champaign, Illinois

Bradley J. Brummel The University of Tulsa, Tulsa, Oklahoma

Fritz Drasgow University of Illinois at Urbana–Champaign, Champaign, Illinois

The United States Defense Manpower Data Center (DMDC) currently assesses gen- der discrimination and sexist behavior separately in its surveys of active duty and re- serve service members. However, based on their definitional and theoretical similari- ties, it is possible that the two constructs may overlap to some extent. Using a military sample of 8,123 female respondents, this issue was examined using path analysis. The results of this study show that gender discrimination and sexist behavior are pre- dicted by several common antecedent variables. In addition, each construct is shown to predict both common and unique outcomes. Implications for the organizational as- sessment of sexist behavior and gender discrimination are discussed.

Gender discrimination and sexual harassment are both forms of treating women differently. As such, it is reasonable to assume that these two constructs are closely related. In fact, DiTomaso (1989) has suggested that sexual harassment is a form of gender discrimination. Legally, this is the case because sexual harassment is prose- cuted as sex discrimination in the United States under Title VII of the Civil Right Act of 1964. However, legal definitions do not always correspond directly to the

Correspondence should be addressed to Christopher D. Nye, 603 E. Daniel, Champaign, IL 61820. E-mail: [email protected]

psychological experience of a phenomenon. Therefore, psychological measures of sexual harassment and gender discrimination may reveal that these constructs func- tion somewhat differently in important and interesting ways. This article investigates the empirical overlap of gender discrimination and sexist behavior, the form of sex- ual harassment that is most theoretically similar to gender discrimination.

GENDER DISCRIMINATION AND SEXIST BEHAVIOR

Heckman and Siegelman (1993) described discrimination as occurring “whenever an individual in a matched pair is treated differently in the aggregate or on average” (p. 198). Gender discrimination occurs when a woman is treated differently com- pared to a man. Within organizations, the perception of differential treatment can be assessed in relation to three separate but related factors: career discrimination, assignment discrimination, and performance evaluation discrimination. Previous reports have found that this three-factor structure fits the psychological experience of discrimination well (Ormerod et al., 2003).

Similar to gender discrimination, sexual harassment is a gender-related form of workplace hostility. The Sexual Experiences Questionnaire (SEQ) measures these experiences by covering four separate dimensions of sexual harassment: sexist be- havior, sexual hostility, unwanted sexual attention, and sexual coercion (Fitzger- ald, Magley, Drasgow, & Waldo, 1999). Fitzgerald, Magley, et al. (1999) defined sexist behaviors as “experiences that can be considered primarily discriminatory based on one’s sex” and sexual hostility as “more explicitly sexual in nature” (p. 249). In contrast, unwanted sexual attention includes behavior that is offensive, un- wanted, and unreciprocated, whereas sexual coercion involves providing job-re- lated benefits in exchange for sexual cooperation.

Although it is clear that the broad definitions of gender discrimination and sex- ual harassment have differences, a similarity can be seen in the definitions given for gender discrimination and the sexist behavior dimension of sexual harassment. Due to this definitional similarity, some redundancy in the outcomes and anteced- ents of these two constructs could be expected. Nonetheless, DMDC assesses these constructs separately in its Workplace and Gender Relations (WGR) surveys. Al- though previous research has shown that sexist behavior is a facet of a broader sex- ual harassment construct (Fitzgerald, Magley, et al., 1999), Mazzeo, Bergman, Bu- chanan, Drasgow, and Fitzgerald (2001) demonstrated that this factor has a negative correlation with other forms of sexual harassment when a single harass- ment situation is assessed. Therefore, it reasonable to expect gender discrimination to covary with sexist behavior but not with sexual harassment more broadly. For this reason, the present study focuses on sexist behavior alone rather than the broader sexual harassment construct.

300 NYE, BRUMMEL, AND DRASGOW

Gender discrimination and sexist behavior are also theoretically similar. Previ- ous research has differentiated between two types of sexism, hostile and benevo- lent (Glick & Fiske, 1996, 2001). Within this framework, both gender discrimi- nation and sexist behavior are considered forms of hostile sexism which is characterized by antipathy towards women (Glick & Fiske, 1996). In this case, the differential treatment of women is justified by perceptions of their inferiority and lower status as a group. Because these behaviors may be two aspects of the same underlying construct, one would expect them to be highly related.

Despite their theoretical and definitional links, gender discrimination and sexist behavior can be differentiated along two conceptual dimensions. Possibly because of these differences, sexist behavior and gender discrimination have traditionally been and continue to be measured and studied separately (e.g., Ormerod et al., 2003). First, these two types of behavior differ on the extent to which the motive behind the behavior is obvious. For sexist behavior, the acts are clearly gender related (e.g., put people of your gender down or make offensive sexist remarks). In contrast, gender discrimination requires an inference about the motivation behind the behaviors. Stated differently, if an employee does not receive a promotion, he or she must attribute the event to a gender-related moti- vation for gender discrimination to occur. In this case, the experience is not nec- essarily related to the individual’s gender in the same manner as a joke that de- grades women might be. Therefore, gender discrimination is inherently less explicitly gender related.

Second, there is a difference in the most likely sources of these behaviors. Though the same underlying beliefs may lead to both, decision-making power within the organization is required for gender discrimination. In DMDC’s WGR surveys, sexist behavior and gender discrimination are operationalized in a manner that separates attitudes expressed in verbal and nonverbal behaviors from institu- tional or supervisory actions affecting a person’s career. Gender discrimination re- flects the degree to which important outcomes or opportunities (e.g., being passed over for a promotion or another important assignment), often controlled by the su- pervisor, have been withheld because of an individual’s gender. Thus, by defini- tion, discrimination can only be perpetrated by individuals with control of valued resources. In contrast, sexist behavior indicates whether an individual has been treated differently or offended because of her gender and is operationalized as a so- cial phenomenon that captures the actions of coworkers regardless of supervisory relationships (e.g., made offensive sexist remarks or insulted people of your gen- der). Although the question stems for these scales do not differentiate between the perpetrator and his or her status, we believe that this difference is implicit in the be- haviors because coworkers are unlikely to have the opportunity or authority to en- act discriminatory behaviors. These differences suggest that sexist behavior and gender discrimination may have different antecedents and affect different out- comes.

GENDER DISCRIMINATION AND SEXIST BEHAVIOR 301

If gender discrimination and sexist behavior are indeed distinct constructs, there are important implications for organizations. Perhaps foremost is that organi- zational interventions designed to combat these two types of aversive behavior would be different. For gender discrimination, individuals in supervisory and lead- ership roles must be educated about how their managerial actions affect subordi- nates and they need to be held accountable for the career success of the women they supervise. In contrast, interventions designed to combat sexist behavior should be targeted at coworkers. These individuals should learn how their behav- iors affect others and should be held accountable for their nonprofessional acts.

RESEARCH PURPOSE

Given the complicated relationship between gender discrimination and sexist be- havior, the purpose of the current research project is to determine the value of sepa- rately measuring these constructs via the Gender Discrimination Scale (2002 Sta- tus of the Armed Forces Survey [2002 WGR]) and the Sexist Behavior Scale (Fitzgerald, Magley, et al., 1999). The criteria for this evaluation include the ability to empirically distinguish between scales and the discriminant validity of the scales with regard to both their antecedents and outcomes. Again, given the nega- tive relationship between sexist behavior and other forms of sexual harassment in a single situation (Mazzeo et al., 2001), the other harassment factors were excluded from these analyses.

The results of this research have a number of implications. Pragmatically, it will test the utility of measuring gender discrimination separately from sexist behavior. This is particularly important for large-scale employee surveys, such as those done by the DMDC, because each measure increases the length of the survey and limits the number of additional scales that can be included. Theoretically, this research will investigate whether the experiences of gender discrimination and sexist be- havior covary to the degree of being psychologically redundant given their com- mon tie to hostile sexism.

METHODS

Data and Procedures

This study utilized data from the Status of the Armed Forces Survey: Workplace and Gender Relations 2002 (2002 WGR) collected by DMDC. This survey used a nonproportional stratified, single-stage random sampling procedure to ensure suf- ficient sample sizes for the populations of interest and included members of all De- partment of Defense services and the Coast Guard. The stratification variables in-

302 NYE, BRUMMEL, AND DRASGOW

cluded gender, race/ethnicity, service, pay grade group, and duty location. The survey administration is documented in Willis, Mohamed, and Lipari (2002). Sur- veys were sent to 56,521 eligible members of the population of interest. Of this sample, 19,960 individuals (9,725 women and 10,235 men) returned the survey, for a response rate of 35%.

Based on previous findings that men are less likely to endorse sexist behavior items than women (Donovan & Drasgow, 1999) and that the base rate for male sex- ual harassment is significantly lower than for females (Fitzgerald, Magley, et al., 1999), the male responses were excluded from the analyses. Missing data were also excluded by list-wise deletion. Therefore, 8,123 cases were examined in this study. Of this sample, 12.8% experienced some form of gender discrimination, 41.7% ex- perienced sexual harassment, and 32.6% experienced sexist behaviors at least once.

Scales

The development and psychometric properties of each of the scales used in this study are described in further detail in the Scales and Measures Report for the 2002 WGR (Ormerod et al., 2003).

Sexual Harassment

The Sexist Behavior Scale consists of four items taken from the Sexual Experi- ences Questionnaire (Fitzgerald, Magley, et al., 1999) that ask respondents to indi- cate whether they were targeted with unprofessional, gender-related behavior in the past 12 months. The response options range from 0 (never) to 4 (very often) and pre- vious research has found the alpha reliability of the Sexist Behavior Scale to be .87 (Ormerod et al., 2003). Items from the Sexist Behavior Scale are included in Table 1.

Gender Discrimination

The Gender Discrimination Scale (2002 WGR) is a 12-item scale that asks sur- vey participants to report whether they had experienced adverse behaviors related to performance evaluations, assignments, and careers in the past 12 months. The three subscales are made up of four, three, and five items for the Evaluation, As- signment, and Career Discrimination scales, respectively. The response options are 1 (no, or does not apply), 2 (yes, but your gender was NOT a factor), and 3 (yes, and your gender was a factor). Previous research has found the alpha reliability for the combined scale to be .81 (Ormerod et al., 2003). Items from the Gender Dis- crimination Scale are included in Table 1.

Workplace Hostility

Workplace hostility was assessed by asking individuals how many hostile be- haviors they had experienced at work during the previous 12 months (e.g., “Using

GENDER DISCRIMINATION AND SEXIST BEHAVIOR 303

an angry tone of voice” or “Making you look bad”). Items for this scale were adopted from the Aggressive Experiences Scale (Glomb & Liao, 2003) and re- sponse options ranged from 1 (never) to 5 (very often).

Climate

The climate for sexual harassment was assessed by presenting three hypotheti- cal harassment scenarios and asking respondents to indicate how they believed the organization and their coworkers would respond to such behaviors (e.g., “Others in the unit would not care” and “The complaint would be taken seriously”). The re- sponse scale ranged from 1 (strongly disagree) to 5 (strongly agree).

304 NYE, BRUMMEL, AND DRASGOW

TABLE 1 Factor Loadings for the Sexist Behavior and Gender Discrimination Items

Loading on the Factor Second-Order

Factor Items Loadings

Sexist behavior .38 #55b-Referred to people of your gender in insulting or offensive terms

.69

#55d-Treated you differently due to gender

.80

#55g-Made offensive sexist remarks .80 #55i-Put you down because of gender .83

Gender discrimination- Evaluation discrimination

.72 #54a-Rated lower than deserved .67 #54b-Last evaluation contained

unjustified negative comments .59

#54c-Held to a higher performance standard than others

.39

#54d-Did not receive the same award as others in similar circumstances

.49

Gender discrimination- Assignment discrimination

.77 #54e-Your current assignment does not use job skills

.58

#54f-Your current assignment is not good for your career

.60

#54g-Did not receive tasks that would have prepared you for advancement

.62

Gender discrimination- Career discrimination

.95 #54h-Did not have someone who advised you on career development

.49

#54i-Did not learn of opportunities that would have helped your career

.63

#54J-Unable to get straight answers about your promotion possibilities

.62

#54k-Excluded from events important for career development and being informed

.59

#54L-Didn’t get a job you wanted for which you were qualified

.47

Leadership Effort to Stop Sexual Harassment

The leadership’s effort to stop sexual harassment was measured by asking par- ticipants whether they thought their organization’s leaders (e.g., immediate super- visor, service leadership, and installation leadership) made a reasonable effort to stop sexual harassment. The response options were 1 (no), 2 (don’t know), or 3 (yes).

Job-Gender Context

The job-gender context variable was created by combining four items that as- sess the gender of the immediate supervisor; the gender mix of the current group; whether the current position is typically held by people of his or her gender; and whether the immediate work environment primarily consists of other individuals of the same sex. Because these items were each measured on different scales, the items were standardized and then averaged to create a single context variable.

Exit Actions

Exit actions were assessed using a 10-item scale that asks whether the individ- ual has taken actions to facilitate leaving the Armed Services (e.g., “Prepared a re- sume”). Responses were scored on a 0 (no) or 1 (yes) scale and higher scores indi- cate participation in a higher number of activities to leave the military.

Satisfaction

The Supervisor, Coworker, and Work Satisfaction Scales asked respondents whether they agreed or disagreed with statements regarding the target of the satis- faction. All three scales assessed satisfaction using a 5-point response scale rang- ing from 1 (strongly disagree) to 5 (strongly agree). Example items include “You are impressed with the quality of leadership in your work group” for Supervisor Satisfaction, “You like your coworkers” for Coworker Satisfaction, and “You like the kind of work you do” for the Work Satisfaction scale.

Physical Health

General health was measured by asking individuals to assess the veracity of several statements about their health in general (e.g., “My health is as good as any- body I know”). Response options for this scale ranged from 1 (definitely false) to 4 (definitely true).

Psychological Distress

Respondents rated their psychological distress by indicating the frequency with which they had experienced several symptoms of distress over the past several

GENDER DISCRIMINATION AND SEXIST BEHAVIOR 305

weeks (e.g., “Felt downhearted and blue” or “Felt so down in the dumps that noth- ing could cheer you up”). These responses were scaled from 1 (little or none of the time) to 4 (most or all of the time).

Role Limitations

Role limitations due to physical health were assessed by asking respondents to indicate how often their work was affected by their physical health over the past four weeks (e.g., “Had difficulty performing the work or other activities you do”). Response options ranged from 1 (little or none of the time) to 4 (all or most of the time).

Analyses

The factor structure described in the Scales and Measures Report for the 2002 WGR (Ormerod et al., 2003) was tested using a confirmatory factor analysis in LISREL VIII (Jöreskog & Sörbom, 1993). Given the hypothesized three-factor structure of gender discrimination, the four items from the Sexist Behavior Scale and the 12 items from the Gender Discrimination Scale were examined using one-factor, two-factor, three-factor, and four-factor models. These models were tested to ensure that the one-factor Sexist Behavior Scale and the three factors of the Gender Discrimination scale were statistically separable. In the two-factor model, the Sexist Behavior items loaded on one factor and the Gender Discrimina- tion items loaded on a second factor. In the three-factor model, the gender discrim- ination factor was split into evaluation and assignment discrimination factors, and in the four-factor model all three of the gender discrimination factors and the sexist behavior factor were examined. Finally, we also tested a model where the three gender discrimination subfacets and the sexist behavior factor loaded on a sec- ond-order gender discrimination factor.

Next, we examined the extent of discriminant validity using a path model esti- mated by LISREL VIII (Jöreskog & Sörbom, 1993). Several variables from this survey were specified as antecedents of sexist behavior and gender discrimination based on previous research and theory. These variables include workplace hostil- ity, climate for harassment, leadership effort to stop sexual harassment, and the job-gender context. Previous research has empirically linked leadership efforts, job-gender context, and climate to sexual harassment (Fitzgerald, Drasgow, & Magley, 1999). Furthermore, Glick and Fiske’s (1996) conceptualization of hostile sexism provides a theoretical link between discrimination and workplace hostility. The conceptualized outcomes of gender discrimination and sexist behavior in- cluded supervisor satisfaction, coworker satisfaction, work satisfaction, role limi- tations due to physical constraints, exit actions, general health, and psychological

306 NYE, BRUMMEL, AND DRASGOW

distress. Again, these outcomes have been found to be related to sexual harassment in previous research (Fitzgerald, Drasgow, et al., 1999).

Paths were estimated from each of the antecedent variables to the Sexist Behav- ior and Gender Discrimination Scales, as well as from Sexist Behavior and Gender Discrimination to all of the outcome variables. Thus, this model assumes Sexist Behavior and Gender Discrimination fully mediate the relation between the exoge- nous variables and the other endogenous variables. All paths were specified as free and were estimated using maximum likelihood estimation.

RESULTS

Confirmatory Factor Analyses

The correlation matrix for all variables appears in Table 2 and fit indices for the models tested are shown in Table 3. Due to the sensitivity of the χ2 statistic to large sample sizes (N = 8,123), this statistic was adjusted to a sample size of 400 using an estimate of the chi-square noncentrality parameter (F. Drasgow, personal com- munication, September 10, 2005). In 1999, Hu and Bentler suggested fit statistics be evaluated relative to the following criteria: .06 for the RMSEA, .08 for the SRMR, and .95 for the NNFI and the CFI. In addition, Byrne (1998) has suggested that values near 1.00 for the GFI and AGFI indicate good fit.

In the initial confirmatory factor analysis, the fit of the one-factor model was clearly unsatisfactory (e.g., the RMSEA was .15 and the AGFI was .71). In con- trast, the two-factor solution was adequate (χ2diff(1) = 8,542.56), but the four-factor model fit noticeably better (χ2diff(5) = 1,887.19). As shown in Table 3, the sec- ond-order factor model also fit the data well. Although the χ2 difference test com- paring the four-factor model and the second-order model was significant (χ2diff(2) = 153.05), the similarity of the fit indices suggest that this difference may be a prod- uct of the well-known sensitivity of χ2 tests to sample size. Therefore, both of these models appear to fit equally well. Table 1 lists the items and the factors they loaded on, as well as the loadings for the second-order model.

As shown in Table 1, the sexist behavior factor had a factor loading of .38 on the second-order gender discrimination factor. This indicates that both gender discrimination and sexist behavior are moderately related. Therefore, an analysis of the discriminant validity of these constructs was conducted to further explore their similarity. Because the CFA results indicate a strong relationship between the gender discrimination subscales, only the aggregate gender discrimination construct was examined in the path analysis. As the correlation matrix in Table 2 shows, this practice is appropriate given the similarities of the correlations be- tween the gender discrimination subscales and the other constructs examined in the model. Only the correlations with work satisfaction were significantly differ-

GENDER DISCRIMINATION AND SEXIST BEHAVIOR 307

308

TA B

L E

2 C

o rr

e la

tio n s

A m

o n g

th e

V a ri

a b le

s E

xa m

in e d

(G ro

u p e d

b y

P o te

n tia

lM e d ia

to rs

, A

n te

ce d e n ts

, a n d

O u tc

o m

e s)

1 2

3 4

5 6

7 8

9 1 0

1 1

1 2

1 3

1 4

1 5

1 6

1. S

ex is

t be

ha vi

or .8

7 2.

E va

lu at

io n

di sc

ri m

in at

io n

.4 4

.5 6

3. A

ss ig

nm en

t di

sc ri

m in

at io

n .3

6 .3

5 .6

4 4.

C ar

ee r

di sc

ri m

in at

io n

.4 6

.5 0

.6 7

5. G

en de

r di

sc ri

m in

at io

n .5

4 .7

7 .7

5 .8

4 .8

2 6.

W or

kp la

ce ho

st il

it y

.5 2

.4 0

.3 5

.4 1

.4 9

.9 3

7. C

li m

at e

fo r

ha ra

ss m

en t

.4 3

.2 9

.3 4

.3 9

.4 1

.9 1

8. L

ea de

r ef

fo rt

to st

op –.

43 –.

28 –.

26 –.

31 –.

36 –.

47 .8

0 9.

Jo b-

ge nd

er co

nt ex

t –.

27 –.

12 –.

10 –.

11 –.

14 –.

10 –.

07 .0

9 N

/A a

10 .S

up er

vi so

r sa

ti sf

ac ti

on –.

23 –.

15 –.

22 –.

25 –.

31 –.

28 .2

4 .0

3 .9

0 11

.C ow

or ke

r sa

ti sf

ac ti

on –.

32 –.

21 –.

25 –.

26 –.

30 –.

47 –.

36 .2

7 .0

2 .4

0 .9

2 12

.W or

k sa

ti sf

ac ti

on –.

25 –.

16 –.

49 –.

26 –.

37 –.

27 –.

30 .2

1 .0

5 .3

4 .3

6 .9

1 13

.R ol

e li

m it

at io

ns -P

hy si

ca l

.1 4

.1 0

.1 4

.1 6

.1 7

.1 1

–. 09

–. 03

–. 05

–. 10

–. 15

.8 9

14 .E

xi t

ac ti

on s

.1 8

.1 6

.1 9

.2 0

.2 4

.2 1

.1 8

–. 16

–. 04

–. 14

–. 15

–. 24

.1 5

.7 9

15 .G

en er

al he

al th

–. 11

–. 10

–. 13

–. 15

–. 16

–. 15

.0 9

.0 2

.0 6

.1 0

.1 5

–. 41

–. 14

.7 7

16 .P

sy ch

ol og

ic al

di st

re ss

–. 29

.1 9

.2 6

.2 5

.3 0

.3 6

.2 8

–. 19

–. 10

–. 22

–. 31

–. 36

.2 4

.2 0

–. 32

.8 4

M ea

n .5

4 3.

48 3.

46 .5

4 1.

21 2.

01 2.

17 2.

60 5.

88 3.

23 3.

48 3.

46 1.

33 .4

1 3.

31 1.

86 S

ta nd

ar d

de vi

at io

n 0.

81 0.

87 1.

01 0.

81 0.

25 0.

94 0.

70 0.

53 1.

75 0.

89 0.

87 1.

01 0.

60 .2

5 0.

58 .6

N o te

. T

he di

ag on

al of

th is

m at

ri x

co ns

is ts

of th

e al

ph a

re li

ab il

it ie

s of

ea ch

of th

e sc

al es

. a T

he re

li ab

il it

y of

th e

Jo b-

G en

de r

C on

te xt

sc al

e w

as no

tc al

cu la

te d

be ca

us e

th e

sc al

e is

a co

m po

si te

of se

pa ra

te it

em s

ra th

er th

an a

sc al

e as

se ss

in g

a ps

yc ho

- lo

gi ca

l co

ns tr

uc t.

ent between the facets of gender discrimination (z = 12.16 between the two high- est correlations). These similarities suggest that there is little utility in analyzing the subscales separately.

Path Analyses

Analyses of the fit statistics for the initial path model revealed that the model fit the data poorly (RMSEA 0.15; CFI 0.83; GFI 0.80; AGFI .66). An examination of the standardized residuals showed that there were several relationships that were not accounted for in the model. As a result, additional paths were added based on em- pirically validated links between variables. Specifically, previous research has pro- vided conceptual and empirical links between work, coworker, and supervisor sat- isfaction (rs between .42 and .55; Fitzgerald, Drasgow, et al., 1999); exit actions and work satisfaction (r of –.16; Griffeth, Hom, & Gaertner, 2000); psychological well-being and coworker and work satisfaction (rs of .16 and .21, respectively; Bergman, Day Langhout, Palmieri, Cortina, & Fitzgerald, 2002); general health and role limitations and psychological health (rs of .18 and .45, respectively; Lub- bers, Loughlin, & Zweig, 2005); workplace hostility and supervisor satisfaction, coworker satisfaction, and psychological distress (rs of –.43, –.34, and .40, respec- tively; Cortina, Magley, Hunter Williams, & Day Langhout, 2001); and the climate for sexual harassment and coworker satisfaction (r of –.28; Fitzgerald, Drasgow, et al., 1999).

To test the partially mediated model, we randomly split the sample into calibra- tion and validation samples of approximately equal size (N = 4,025 and 4,098, re- spectively). The partially mediated model was then estimated in the first sample by adding paths as described above. This model was then cross-validated in the sec- ond sample and compared against the fully mediated model using a χ2 difference test. The resulting model fit well in the cross-validation sample (RMSEA 0.050; CFI 0.98; GFI 0.98; AGFI .96) and the difference in χ2 was significant (χ2diff(13) = 3,682.55).

GENDER DISCRIMINATION AND SEXIST BEHAVIOR 309

TABLE 3 Fit Statistics for the Confirmatory Factor Analyses of Female Responses to the Gender Discrimination Subscales and the Sexist Behavior Scale

2 df 2/dfa RMSEA NNFI GFI CFI AGFI SRMR

One-factor CFA 12,705.39 104 6.97 .150 .81 .78 .83 .71 .096 Two-factor CFA 4,162.83 103 2.94 .069 .94 .94 .95 .92 .048 Three-factor CFA 2,876.20 101 2.35 .057 .96 .96 .96 .95 .039 Four-factor CFA 2,275.64 98 2.09 .051 .96 .97 .97 .96 .034 Second-order CFA model 2,428.69 100 2.14 .052 .96 .97 .97 .95 .039

aChi-square was adjusted to a sample size of 400.

Figure 1 shows the significant paths involving Sexist Behavior and Gender Discrimination (for the sake of clarity in this figure, other paths were omitted). There were several interesting and statistically significant patterns in the path coefficients of the model. The path coefficients from workplace hostility to gender discrimination and sexist behavior were .36 and .24, respectively. These coefficients were the largest of any of the exogenous paths. The Climate for Sexual Harassment also significantly predicted gender discrimination and sex- ist behavior (.18 and .13, respectively). The path from job-gender context to sexist behavior was estimated at –.20 compared to –.08 for gender discrimina- tion. Each of these pairs of paths was significantly different at the .05 level. The paths from leader effort to stop sexual harassment were estimated at –.13 and –.15 for gender discrimination and sexist behavior, respectively. Although these paths showed a comparable magnitude between the sexist behavior and gender discrimination measures, the difference was significant (z = 9.8, p < .001).

There were several statistically significant paths from gender discrimination to the other endogenous variables that differed significantly from the corresponding sexist behavior paths. In the completely standardized solution, the path coefficient from gender discrimination to exit actions was estimated at .18 compared to .05 for sexist behavior. General health also had a significantly higher path from gender discrimination (–.13) than from sexist behavior (–.06).

310 NYE, BRUMMEL, AND DRASGOW

FIGURE 1 Completely standardized path coefficients for the partially mediated model with non- significant paths excluded.

Workplace hostility

Job-Gender Context

Leader effort to stop

Supervisor Satisfaction

Climate for harassment

Coworker Satisfaction

Work Satisfaction

Exit Actions

Role Lim.- Physical

Psychological Distress

General Health

Sexist Behavior

Gender Discrimination

.24 .36

–.18

–.08

–.23

–.15

–.13

.13

.18

.29

–.04

.05

.06

–.08

–.06

.18 .07

.08

–.13

In addition, both gender discrimination and sexist behavior were related to unique outcome variables. The path coefficient from sexist behavior to coworker satisfaction was significant, whereas the corresponding path from gender discrimi- nation was not. Similarly, the paths from gender discrimination to work and super- visor satisfaction were significant, whereas these paths from sexist behavior were not. These findings indicate that although several outcomes are predicted by both sexist behavior and gender discrimination, each of these constructs uniquely pre- dicts other outcomes.

DISCUSSION

Definitional and theoretical similarities between gender discrimination and sexist behavior suggest that there may be some redundancy in their antecedents and out- comes. As such, the purpose of the current study was to determine the utility of measuring these constructs separately. In this study, the confirmatory factor ana- lytic results replicate findings from the 2002 WGR (Ormerod et al., 2003) and indi- cate that these constructs can be considered two separate, yet moderately related factors. This moderate association is consistent with previous theory suggesting that sexist behavior and gender discrimination may both stem from hostile sexism (Glick & Fiske, 1996). However, the magnitude of this relationship also necessi- tates an examination of the discriminant validity of these constructs to determine whether measuring them separately adequately represents the psychological expe- rience of these behaviors.

The results of the path analysis indicated that these constructs function differ- ently in several ways. Workplace hostility and the climate for harassment were sig- nificant antecedents of both constructs. However, these antecedents were shown to be stronger predictors of gender discrimination relative to sexist behavior. In con- trast, leadership effort to stop sexual harassment and job-gender context were sig- nificantly stronger predictors of sexist behavior. Previous research has found simi- lar results linking organizational tolerance and job-gender context to the broader sexual harassment construct (Fitzgerald, Drasgow, et al., 1999; Fitzgerald, Dras- gow, Hulin, Gelfand, & Magley, 1997). Theoretically, these results suggest that gender discrimination occurs in an environment where there is a climate for disres- pecting other individuals in the workplace. In contrast, sexist behaviors appear to result from a combination of women operating outside of stereotypical female gender roles (as assessed by job-gender context) and the permissiveness of organi- zational leadership. In this case, gender discrimination may act to prevent women from occupying certain traditionally male positions and sexist behavior may be a form of retribution, either subconscious or otherwise, for obtaining these posi- tions. Indeed, recent research suggests that sexual harassment may in fact operate in this manner (Berdahl, 2007).

GENDER DISCRIMINATION AND SEXIST BEHAVIOR 311

The results of this study indicate that gender discrimination is a stronger predic- tor of both exit actions and general health than sexist behavior. This difference is most likely due to the substantial influence of these behaviors on important ca- reer-related outcomes. It is likely that women view discrimination as more severe than sexist behavior because of its effect on outcomes and resources that influence an individual’s life. As a result of these behaviors, women may make less money and receive fewer promotions and opportunities in their career. When faced with these disadvantages, it is likely that individual health, work satisfaction, and turn- over behavior will be substantially influenced. In fact, the current study shows that the relationships between gender discrimination and these outcome variables are among the largest endogenous paths observed in the model.

Although it is interesting that gender discrimination was a significantly stron- ger predictor of these outcomes, the unique outcomes predicted by sexist behavior and gender discrimination have broader implications for the measurement and study of these behaviors. Specifically, there was a significant link between gender discrimination and work and supervisor satisfaction. As described above, commit- ting discriminatory acts requires decision-making power and control over re- sources. Therefore, it is unlikely that colleagues in a work environment are able to perpetrate these types of behaviors. Furthermore, because discrimination naturally leads to the attainment of fewer desired outcomes, victims of these acts are likely to be less satisfied with their work. Previous meta-analytic research has shown that work satisfaction is the strongest predictor of turnover intentions, possibly ex- plaining the substantial relationship between gender discrimination and exit ac- tions (Griffeth et al., 2000). In contrast, sexist behavior significantly predicted coworker satisfaction. Again, these behaviors are operationalized as social phe- nomena that often include routine interactions with coworkers in the work environ- ment. As such, these behaviors significantly influence coworker satisfaction.

It is important to distinguish between practically and statistically significant re- sults. Although the differences between several of these paths were significantly different statistically, the low magnitude of the difference may indicate that there is little practical worth in differentiating between these relationships. For example, a difference of .02 for the standardized paths from leadership effort to stop harass- ment to sexist behavior and gender discrimination may have little practical impor- tance. Still, other larger differences (i.e., those from job-gender context and work- place hostility) are clearly important.

One limitation of this study is that it was conducted using a single military sam- ple. Although previous research has found similar results in civilian organizations (Fitzgerald et al., 1997), the construct of gender discrimination was not addressed. Future research should attempt to replicate these findings in non-military organi- zations.

A second limitation lies in the method of data collection. As always, method- ological artifacts may have affected survey responses. For example, although the

312 NYE, BRUMMEL, AND DRASGOW

Gender Discrimination Scale does not specify the perpetrators of these behaviors, the stem for the Sexist Behavior Scale asks about behaviors involving on- or off-duty military and civilian personnel in the respondent’s workplace or on their ship/installation. It is possible that this difference could affect the measurement of these behaviors.

Finally, the current study was limited by its focus on sexist behavior rather than the broader construct of sexual harassment. Possibly due to their common anteced- ents, the facets of sexual harassment are typically highly correlated. The magni- tude of this correlation reduces the likelihood of observing differences between the antecedents and outcomes of each of the facets. As a result, previous research has demonstrated the benefit of examining the specific incidences of harassment sepa- rately (Mazzeo et al., 2001). For this purpose, we examined the harassment factor that has the most compelling theoretical link with gender discrimination. However, future research should examine the behaviors that are theoretically linked to the other dimensions of sexual harassment as well.

Future research should also focus on differentiating the role of these two con- structs in the mistreatment of women. The results presented here provide anecdotal evidence that discrimination acts as a barrier to women obtaining traditionally male-dominated jobs and sexual harassment functions to punish those who do. Al- though these relationships have received partial support elsewhere (Berdahl, 2007), future research should examine these links more explicitly.

CONCLUSION

The results of this study indicate that sexist behavior and gender discrimination, al- though conceptually similar, can be differentiated in meaningful and important ways. As such, these constructs should be measured separately for both theoretical and practical reasons. Theoretically, although both classes of behaviors may be part of a broader psychological construct (i.e., hostile sexism), they may be caused by different constituents in the workplace. Therefore, measuring these constructs sepa- rately will allow researchers to clarify the psychological meaning of these forms of hostility and better target organizational intervention. Clearly, the combined mea- surement of these constructs would obscure the antecedents of these costly organiza- tional outcomes and may limit the organization’s ability to prevent them.

REFERENCES

2002 Status of the Armed Forces Survey—Workplace and gender relations [Data file and documenta- tion on CD-ROM]. (2003). Arlington, VA: United States Defense Manpower Data Center.

Berdahl, J. L. (2007). The sexual harassment of uppity women. Journal of Applied Psychology, 92, 425–437.

GENDER DISCRIMINATION AND SEXIST BEHAVIOR 313

Bergman, M. E., Day Langhout, R., Palmieri, P. A., Cortina, L. M., & Fitzgerald, L. F. (2002). The (un)reasonableness of reporting: Antecedents and consequences of reporting sexual harassment. Journal of Applied Psychology, 87, 230–242.

Byrne, B. M. (1998). Structural equation modeling in LISREL, PRELIS, and SIMPLIS: Basic concepts, applications, and programming. Mahwah, NJ: Lawrence Erlbaum.

Cortina, L. M., Magley, V. J., Hunter Williams, J., & Day Langhout, R. D. (2001). Incivility in the workplace: Incidence and impact. Journal of Occupational Health Psychology, 6, 64–80.

DiTomaso, N. (1989). Sexuality in the workplace: Discrimination and harassment. In J. Hearn, D. L. Sheppard, P. T. Sheriff, & G. Burrell (Eds.), The sexuality of the organization (pp. 71–90). Newbury Park, CA: Sage.

Donovan, M., & Drasgow, F. (1999). Do men’s and women’s experiences of sexual harassment differ? An examination of the differential test functioning of the sexual experiences questionnaire. Military Psychology, 11, 265–282.

Fitzgerald, L. F., Drasgow, F., Hulin, C. L., Gelfand, M. J., & Magley, V. J. (1997). Antecedents and consequences of sexual harassment in organizations: A test of an integrated model. Journal of Ap- plied Psychology, 82, 578–589.

Fitzgerald, L. F., Drasgow, F., & Magley, V. J. (1999). Sexual harassment in the armed forces: A test of an integrated model. Military Psychology, 11, 329–343.

Fitzgerald, L. F., Magley, V. J., Drasgow, F., & Waldo, C. R. (1999). Measuring sexual harassment in the military: The sexual experiences questionnaire (SEQ-DoD). Military Psychology, 11, 329–343.

Glick, P., & Fiske, S. (1996). The ambivalent sexism inventory: Differentiating hostile and benevolent sexism. Journal of Personality and Social Psychology, 70, 491–512.

Glick, P., & Fiske, S. (2001). An ambivalent alliance: Hostile and benevolent sexism as complementary justifications for gender inequality. American Psychologist, 56(2), 109–118.

Glomb, T. M., & Liao, H. (2003). Interpersonal aggression in work groups: Social influence, reciprocal, and individual effects. Academy of Management Journal, 46, 486–496.

Griffeth, R. W., Hom, P. W., & Gaertner, S. (2000). A meta-analysis of antecedents and correlates of employee turnover: Update, moderator tests, and research implications for the next millennium. Journal of Management, 26, 463–488.

Heckman, J., & Siegelman, P. (1993). The urban institute audit studies: Their methods and findings. In M. Fix & R. Stuyk (Eds.), Clear and convincing evidence: Measurement of discrimination in Amer- ica (pp. 187–258). Washington, DC: The Urban Institute Press.

Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conven- tional criteria versus new alternatives. Structural Equations Modeling, 6, 1–55.

Jöreskog, K., & Sörbom, D. (1993). New features in LISREL 8. Chicago: Scientific Sofware Interna- tional.

Lubbers, R., Loughlin, C., & Zweig, D. (2005). Young workers’ job self-efficacy and affect: Pathways to health and performance. Journal of Vocational Behavior, 67, 199–214.

Mazzeo, S. E., Bergman, M. E., Buchanan, N. T., Drasgow, F., & Fitzgerald, L. F. (2001). Situation-spe- cific assessment of sexual harassment. Journal of Vocational Behavior, 59, 120–131.

Ormerod, A. J., Lawson, A. K., Sims, C. S., Lytell, M. C., Wadlington, P. L., Yaeger, D. W., et al. (2003). 2002 Status of the Armed Forces Survey—Workplace and gender relations: Report on scales and measures (DMDC Rep. No. 2002-031). Champaign, IL: University of Illinois, Work Experi- ences Lab.

Willis, E. J., Mohamed, S. G., & Lipari, R. N. (2002). 2002 Status of the Armed Forces Survey—Work- place and gender relations: Administration, datasets, and codebook (Report No. 2002-010). Arling- ton, VA: United States Defense Manpower Data Center.

314 NYE, BRUMMEL, AND DRASGOW