Vous êtes sur la page 1sur 26

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/227604263

Reported incidence rates of work-related sexual


harassment in the United States: Using meta-analysis
to explain reported rate dispari....

Article  in  Personnel Psychology · December 2006


DOI: 10.1111/j.1744-6570.2003.tb00752.x

CITATIONS READS
160 3,499

4 authors, including:

Remus Ilies Nancy Hauserman


National University of Singapore University of Iowa
104 PUBLICATIONS   9,490 CITATIONS    15 PUBLICATIONS   177 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Employee Selection View project

All content following this page was uploaded by Remus Ilies on 06 October 2017.

The user has requested enhancement of the downloaded file.


PERSONNEL PSYCHOLOGY
2003,56,607-631

REPORTED INCIDENCE RATES OF WORK-RELATED


SEXUAL HARASSMENT IN THE UNITED STATES:
USING META-ANALYSISTO EXPLAIN REPORTED RATE
DISPARITIES

REMUS ILIES
Department of Management
Warrington College of Business Administration, University of Florida
NANCY HAUSERMAN
Department of Management and Organizations
Henry B. Tippie College of Business, University of Iowa
SUSAN SCHWOCHAU
Dickinson Wright PLLC
JOHN STIBAL
Equant N.V.

This study presents a meta-analytic review of the incidence of sexual


harassment in the U.S.The impact of 3 main moderator variables (type
of survey used, sampling technique, and the type of work environment
in which the study was conducted) on the reported incidence rate was
estimated by cumulating incidence rates reported in the literature. Re-
sults show that directly querying the respondents about whether or
not they experienced sexual harassment (vs. using questionnaires that
listed behaviors believed to constitute sexual harassment), and employ-
ing probability-sampling techniques (vs. convenience sampling), led to
substantially lower estimates of sexual harassment incidence. In ad-
dition, the results suggest that sexual harassment is more prevalent in
organizations characterized by relatively large power differentials be-
tween organizational levels. Based on more than 86,000 respondents
from 55 probability samples, on average, 58% of women report having
experienced potentially harassing behaviors and 24% report having ex-
perienced sexual harassment at work.

During the past 2 decades, sexual harassment has emerged as an im-


portant social issue with critical implications for the society at large and
for organizations in particular. Consequently, scientific research has in-
creasingly focused on the study of the sexual harassment phenomenon,

We thank Tim Judge, Huy Le, Sara Rynes, and Frank Schmidt for their helpful com-
ments on earlier versions of this paper
Correspondence and requests for reprints should be addressed to Remus Ilies, who
is now at the Department of Management, Eli Broad Graduate School of Management,
N475 North Business Complex, East Lansing, MI, 48824; admin@studies-online.org.

COPYRIGHT 0 2003 PERSONNELPSYCHOLOGY,INC.


607
608 PERSONNEL PSYCHOLOGY

its extent in the workplace, its psychological underpinnings, and its an-
tecedents and consequences. Many of the studies on sexual harassment
purport to offer quantitative estimates of the incidence rate of sexual ha-
rassment in certain organizations or occupations. Information about the
level of sexual harassment is valuable; it can be used to assess whether
there is an increase or a decrease in the amount of sexual harassment
generally or in specific workplaces or occupations. Incidence data may
also be used to measure the success or failure of various training pro-
grams or the effect of specific, highly publicized cases or claims based
on sexual harassment (e.g., the Clarence Thomas/Anita Hill hearings in
1991). Furthermore, to the extent that legislation is expected to elimi-
nate or at least minimize sexual harassment, estimates of incidence rate
may say something about the effect of law (Gruber, 1990).
Because the incidence rate of sexual harassment may be used to draw
a variety of conclusions, it is critical that surveys attempting to assess
sexual harassment incidence use sound sampling design principles and
valid and reliable measures. But in fact, if one considers the studies
that attempted to measure the incidence rate of sexual harassment, it is
difficult to draw any firm conclusion about the actual incidence of such
behaviors (Lengnick-Hall, 1995). Empirical data show that reported
incidence rates vary widely across studies. Studies included in Gruber’s
(1990) review, for example, report incidence rates between .28 and .75.
Inconsistent results do not offer clear policy implications. Further-
more, gross variance in reports of the incidence of sexual harassment
may provide fodder for those who would argue that the variability in
empirical findingsshows the difficultyof identifyingwhat constitutes sex-
ual harassment and that reports of frequent occurrence are a reflection
of researcher bias or faulty study design. Luthar (1999, for example,
criticizes the sexual harassment research by saying that “policy makers
should look at the results from sexual harassment research with some
skepticism, as it is dominated by a feminist ideology, tainted by implicit
biases against males, and functions from a rather narrow paradigm”
(p. 271). Whereas the variability in incidence rates may be due to avari-
ety of rational factors and methodological problems or inconsistencies,
it is critical that we be able to identify those factors and problems. The
present study integrates empirical findings relevant to the incidence of
sexual harassment in the workplace and identifies major factors that in-
fluence the reported incidence rate.

Moderating Effects

Various critics of sexual harassment studies suggest several problems


with the existing research. Criticisms fall into two broad categories: (a)
REMUS ILIES ET AL. 609

lack of a consistent definition of sexual harassment (construct confusion;


e.g., Lengnick-Hall, 1995), which, operationally, leads to the use of dif-
ferent types of surveys, and (b) faulty study design (use of nonprobabil-
ity [convenience]samples; e.g., Gutek, 1995;Lengnick-Hall, 1995). This
study’s first main objective is to provide cumulative estimates of the in-
cidence of sexual harassment in the workplace as assessed with different
types of surveys and different sampling techniques, and to estimate the
magnitude of the impact that methodological moderators have on the re-
ported incidence rates. Second, after accounting for the effect of these
methodological factors on the reported incidence rate, we investigate
whether the type of work environment in which studies were conducted
influenced the reported incidence rates.

Construct Definition

Much criticism of the sexual harassment literature focuses either on


various authors’ apparent failure to offer explicit and clear construct
definitions, or the inconsistency of implicit definitions across studies.
Several authors point to the related problem of who defines the event
or behavior as sexual harassment ( h e y & Cavanaugh, 1995; Braine,
Bless, & Fox, 1995; Foulis & McCabe, 1997; Gardner & Allen, 1996;
Lengnick-Hall, 1995). When authors rely on the term “sexual harass-
ment,” it is often not clear whether the term is advanced from a legal or
a psychological perspective. Because the current legal definition is gen-
erally considerably narrower than definitions from other fields, critics
like Lengnick-Hall (1995) argue that authors must make such distinc-
tions clear. Lengnick-Hall points out that if authors do not differentiate
between legal and psychological or other definitions they may be over-
estimating the problem of sexual harassment. If, for instance, authors
use a definition that includes more behaviors than court cases have rec-
ognized as constituting sexual harassment, the incidence rate will reflect
the authors’ perception but not necessarily the legal system’s construc-
tion of sexual harassment.
In the assessment of the prevalence of sexual harassment, the per-
spective from which sexual harassment is defined leads to distinct mea-
surement methodologies. That is, two distinct approaches to the mea-
surement of sexual harassment incidence can be found in the research
literature: the direct query survey, which allows the respondent to de-
fine sexual harassment, and the behavioral experiences survey, which in-
volves providing the respondent with a list of experiences defined by the
researcher to constitute sexual harassment (Lengnick-Hall, 1995). The
distinction between these two approaches has important theoretical and
methodological implications.
610 PERSONNEL PSYCHOLOGY

Directly querying respondents about the experience of sexual harass-


ment measures a perceptual construct; thus, reported incidence rates
will be influenced by the individual characteristics of the respondent
and, possibly, by those of the psychological climate in the organization.
As people’s perceptions about what constitutes sexual harassment may
change, the direct query method does not provide data that can be used
to investigate temporal trends in the prevalence of sexual harassment in
the workplace (Gutek, 1995), or to evaluate the effectiveness of orga-
nizational programs or legal provisions. Because perceptions of sexual
harassment may be different for men and women, subjective measures
make it difficult to assess the extent to which the difference in reported
incidence rates between men and women is real or perceptual (although
recent meta-analytic evidence suggests that the gender difference on
perceptions of what constitutes sexual harassment is smaller than was
previously believed; Blumenthal, 1998; Rotundo, Nguyen, & Sackett,
2001). This is not to say that perceptions of sexual harassment are less
important than more objective measures. For example, although objec-
tive measures may have stronger relationships with legal outcomes, per-
ceptual measures may be more important in predicting victim responses
and organizational outcomes (Lengnick-Hall, 1995).
The behavioral experiences approach, on the other hand, has two
major advantages: It minimizes respondent perceptual bias and, if used
consistently, allows for comparing incidence rates across studies and
time.l In order to more aggressively address the problem of sexual ha-
rassment in certain companies or industries, it is important to be able
to compare empirical findings about the incidence of sexual harassment
and to identify industry or organizational moderators of this behavior.
Comparing empirical findings with regard to sexual harassment inci-
dence rate is important for research scholars, human resource managers,
and policy makers. Research scholars can employ longitudinal designs in
an attempt to place the sexual harassment construct in a causal nomolog-
ical model, managers can assess the effectiveness of the human resource
policies with regard to sexual harassment, and policy makers can study
the effect of their legislative interventions.
A substantive question of interest with regard to the use of a particu-
lar type of survey is whether direct query and behavioral experiences sur-
veys lead to different reported incidence rates of sexual harassment, and,
if these two types of surveys indeed lead to different results, what is the
magnitude of the difference. Previous studies have consistentlyreported

‘Though it minimizes perceptual bias, the behavioral experiences method does not
eliminate such bias because behavioral items such as “Suggestive stories and offensive
jokes” (Fitzgerald et al., 1988) implicitly require perceptual interpretation (i.e., what one
perceives to be “suggestive”or “offensive”will vary across respondents).
REMUS ILIES ET AL. 611

that a relatively small percentage of respondents who report experienc-


ing behaviors that researchers believed to constitute some form of sex-
ual harassment actually themselves perceive those experiencesas sexual
harassment (e.g., Fitzgerald et al., 1988; Magley, H u h , Fitzgerald, &
DeNardo, 1999; Stockdale & Vaw, 1993; Stockdale, Vaux, & Cashin,
1995).
The difference in estimates of sexual harassment prevalence ob-
tained with direct query and behavioral experiences surveys reflects, in
part, women’s reluctance to label offensive experiences as sexual ha-
rassment (e.g., Magley et al., 1999). The literature on sexual harass-
ment labeling (e.g., Brooks & Perot, 1991; Magley et al., 1999;Munson,
Miner, & Hulin, 2001; Stockdale et al., 1995) seeks to understand the
reasons behind women’s hesitance in labeling such experiences as ha-
rassment and to examine the psychological and practical consequences
of labeling. Most recent research on the topic suggests that whether or
not respondents label unwanted experiences as sexual harassment has
little influence on the negative consequences of the unwanted experi-
ences (Magely et al., 1999; Munson et al., 2001). Magley et al. (1999,
p. 399) conclude: “These data from three organizations demonstrate
that whether or not a woman considers her experience to constitute sex-
ual harassment, she experiences similar negative psychological, work,
and health consequences.” These authors make a strong case against
the use of direct query measures of sexual harassment because of the
implicit labeling process associated with such measures.
A position that is in sharp contrast with the conclusions from the re-
cent labeling research reviewed above reflects what Magley et al. (1999)
call the “whiner hypothesis.” Writers who take this position, typically in
the popular business or cultural press (e.g., Alger & Flanagan, 1996), ar-
gue that the prevalence of sexual harassment is overestimated and data
on sexual harassment incidence reflect women’s “whining” about minor
events. The research mentioned above (Magley et al., 1999; Munson et
al., 2001) presents empirical tests that refute the whiner hypothesis; our
study can test this hypothesiswith meta-analytic data. That is, if the pop-
ulation incidence rate derived with direct query surveys exceeds the rate
derived with behavioral experiencessurveys, then the women are indeed
whining about events that were considered too minor to be included in
behavioral experiences questionnairesby researchers; if the direct query
estimate is lower than the behavioral experiencesestimate, then women
refuse to label all the unwelcome behaviors that they experience as sex-
ual harassment.
In sum, the literature on sexual harassment lacks national-level esti-
mates of the actual prevalence of sexual harassment measured with ei-
ther type of survey (direct query or behavioral experiences). Further-
612 PERSONNEL PSYCHOLOGY

more, as explained above, comparing population estimates obtained


with direct query and behavioral experiences surveys allows testing of
the “whiner hypothesis.” This study’s first goal is to fill these gaps in the
literature by offering meta-analytical estimates for (a) the extent of sex-
ual harassment, as measured by behavioral experiences and direct query
surveys, and (b) the magnitude of the difference between the estimates
obtained by the two different surveying methods.

Design Issues
Several articles have commented on the problems inherent in the
sample selection bias associated with the reliance on convenience, non-
probability samples (Gutek, 1995; Lengnick-Hall, 1995; Pryor &
McKinney, 1995). Probability sampling designs involve a sampling plan
that gives each respondent from the sampling frame a nonzero known
probability of being surveyed. Results from probability samples can be
generalized to the population of interest (samplingframe). Convenience
samples do not allow generalization of findings (Lengnick-Hall, 1995)
and may provide biased estimates of the prevalence of sexual harass-
ment in the workplace through a selection-bias mechanism.
Whereas respondent self-selection processes are likely to be present
in both convenience and probability samples, convenience samples are
subject to another selection bias process: sample-selectionbias. Sample-
selection processes can bias the reported incidence rates in either di-
rection. First, it is possible that the selection of convenience samples
for sexual harassment research may be biased towards samples that are
likely to report higher incidence rates through an availabilitymechanism.
Samples of women who have filed grievances or sought help are simply
more available to researchers and they are also likely to report higher
incidence rates of sexual harassment. At the extreme, sampling from a
population of people who filed sexual harassment complaints will give
an estimate of sexual harassment incidence close to 1.00. An illustra-
tive example of the sample-selection bias is given by Loy and Stewart
(1984, p. 32):
Some estimates may be inflated because of the composition of the sample.
For example, women who attend a speak-out on sexual harassment (used
as the sample in the research by Silverman, 1976) may attend because they
have experienced sexual harassment themselves. The sexual harassment
frequency estimate of 90 percent in Silverman’s study may be inflated
because it comes from a biased sample.

Second, convenience samples can bias the incidence rate of sex-


ual harassment downwardly, through a different sample-selection ef-
fect. Progressive organizations, where there is an increased awareness
REMUS rrms ET AL. 613

about sexual harassment and, thus, the prevalence of sexual harassment


is likely to be lower, are more likely to monitor sexual harassment and
cooperate with researchers from the academia in studying sexual harass-
ment. Therefore, samples from such organizations may be more readily
availablefor sexual harassment researchers, and these samples are likely
to report lower levels of sexual harassment.
In sum, convenience samples are subject to sample-selection bias,
and the two sample-selection mechanisms described above have oppo-
site biasing effects on the reported prevalence of sexual harassment.
Thus, it is important to investigatethe impact of the type of sample on the
reported incidence rate empirically. Solid, cumulative estimates of the
extent of sexual harassment based on (a) convenience and (b) probabil-
ity samples are needed in order to investigate this impact. Documenting
this effect constitutes the second goal of the study.

Organizational Context
An organizational context dimension relevant to sexual harassment
is the relative power difference between organizational levels. As most
definitions of sexual harassment include inappropriate use of power
(e.g., Loy & Stewart, 1984; McKinney, 1990; Niebuhr, & Boyles, 1991;
Rosell, Miller, & Barber, 1995), and power relationships among organi-
zational members are largely determined by organizational structure, it
follows logically that “highly structured and stratified organizations are
more likely settings for harassment” (Niebuhr & Boyles, 1991, p. 448).
Other things being equal (policies, training, etc.), we would predict that
the extent of sexual harassment would be higher in highly structured, bu-
reaucratic organizations than in organizationswith a more organic struc-
ture (Bums & Stalker, 1961). Testing this prediction was the third goal
of this study.
To investigate the association between power differentials (deter-
mined by organizational structure) and sexual harassment with the
present data, we can examine whether the type of work environment
where sexual harassment incidence data were reported influences the
reported rates. We grouped studies according to the work environment
of the respondents into academic,private sector; government, and militaly
studies, assuming that organizational characteristics such as structure
and the nature of power relationships are more similar within groups
than across these groups.
As the military is highly structured with large power differentials
among organizational levels, we expect sexual harassment incidence
rates to be the highest in the military. Because academic organizations

2We thank an anonymous reviewer for pointing out this possibility.


614 PERSONNEL PSYCHOLOGY

can be viewed as professional organizations with small power differen-


tials (Scott, 1998),we expect sexual harassment incidence rates to be the
lowest in academia. We do not make predictions with respect to differ-
ences in sexual harassment incidence rates reported by individuals em-
ployed by the government and those employed in the private sector be-
cause we are not in a position to compare these environments on power
differentials; but, we investigate whether these rates are different on an
exploratory basis.

Time Trenak
The year when a specific study has been conducted may influence the
reported incidence rate of sexual harassment. On the one hand, because
public awareness about sexual harassment has been on the increase over
the past decades, it is possible the sexual harassment incidence rates re-
ported by women have also been increasing because, over time, women
progressively changed their definition of sexual harassment to include
more behaviors. On the other hand, because normative measures aimed
at decreasing sexual harassment in organizations have been increasingly
implemented over the same period of time, it is possible that reported
incidence rates have been decreasing because of a substantive reduction
in the extent of sexual harassment. Because it is not clear how time in-
fluences the reported incidence rate, we will examine whether there are
time trends in the incidence data on an exploratory basis.

Method

Hunter and Schmidt’s (1990) methods for meta-analysis were used


to integrate results from the studies in our database and investigate the
presence of moderators. Meta-analysis was chosen over other possible
methods (e.g., least squares or log-odds regression) because of several
advantages offered by this method: (a) it gives estimates for the popula-
tion incidence rate within each category of studies (e.g., direct query vs.
behavioral experience surveys, or academic vs. private sector vs. govern-
ment vs. military samples), (b) it accounts for differences in sample size
and the amount of sampling error present in individual estimates, and
(c) it gives interpretable information with regard to the variability of the
incidence rate in the study population and with regard to the confidence
in the meta-analytic estimates.

Meta-Analysis Procedures
Using the Schmidt-Hunter psychometric meta-analysis method
(Hunter & Schmidt, 1990),we conducted a meta-analyticreview that cu-
REMUS ILIES ET AL. 615

mulates across studies the reported incidence of sexual harassment expe-


riences. Although the Schmidt-Hunter meta-analytic procedures have
been applied mostly to correlations or effect sizes, those procedures can
be extended to cumulate proportions (Viswesvaran & Schmidt, 1992).
We conducted an overall meta-analysis to cumulate the results from all
independent samples, and multiple meta-analyses to investigate moder-
ator effects. The observed variance was corrected before testing for the
presence of any moderating influence.
To estimate the incidence rate of sexual harassment in the population
sampled by the studies included in this review, we followed the proce-
dure for applying the Schmidt-Hunter method to meta-analyses of pro-
portions detailed byviswesvaran and Schmidt (1992).3 In addition to re-
porting point estimates of the true incidence rate, it is also important to
describe variability in these point estimates. Accordingly, we report 95%
confidence intervals and 90% credibility intervals around the estimated
population incidence rates. Confidence intervals provide an estimate of
the variability around the estimated mean incidence rate that is due to
sampling error; a 95% confidence interval means that if the estimation
procedures were repeated a large number of times, the estimate would
fall within the confidence interval range in 95% of the cases. Credibility
intervals provide a measure of the variability of incidence rates in the
study population; a 90% credibility interval indicates the range within
which 90% of individual incidence rates fall. Thus,confidence intervals
show the confidence in the estimated mean incidence rate and credibility
intervals estimate the true variability in the individual incidence rates in
the study population.
We divided the studies into categories according to the expected
moderator variables. We conducted separate meta-analyses for each
of the categories in order to estimate the incidence rate of sexual ha-
rassment in the categories delimited by moderator variables. Meta-
analytical evidence for the presence of moderators requires that (a) true
estimates are different in the categories formed by the potential moder-
ator variable, and (b) the mean corrected standard deviation within cat-
egories is smaller than the corrected standard deviation computed for
combined categories.

3Samplesizes (n;)and the proportionof people who reported having experienced sexual
harassment (Pi)were available for each study. For each analysis the mean incidence
rate (proportion; P) was computed with the formula: F = C ( n i P , ) / h , . The observed
variance in the incidence rates was computed as: n i ( P ; - n 2 / C n i . The sample-weighted
mean sampling-errorvariance of incidence rates was obtained as: En; PiQ;)/Cni)/Cni=
(E[PiQi])/Eni. The true variance estimate was then computed as the differencebetween
the observed variance and the sample-weighted mean sampling error variance (this is an
upper-boundvalue because it includes variance due to artifacts not accounted for by the
meta-analytic procedure).
616 PERSONNEL PSYCHOLOGY

Literature Search

In order to identify all possible studies that present estimates of the


incidence of sexual harassment in the workplace, we searched the Psych-
INFO database, 1967-2000, the ERIC 1966-2000 database, and the Aca-
demic Search Elite, 1990-2000 full-text database (provided by EBSCO
Publishing) for studies (articles, book chapters, dissertations, and un-
published reports) that referenced sexual harassment as a keyword or
were included in the “sexual harassment” subject-heading category. This
effort resulted in the identification of 1,203 abstracts from ERIC, 1,081
abstracts from PsychINFO, and 323 full text articles from Academic
Search Elite.

Rules for Inclusion in the Meta-Analysis

We first constructed a database that included the abstracts selected


from PsychINFO and ERIC. In all, the database contained 2,231 entries
(after eliminating 53 duplicate entries). We reviewed these abstracts to
determine whether they contain the required information to be included
in the meta-analysis. We applied the initial screening to the PsychINFO
and ERIC abstracts in order to reduce the number of articles that would
have to be examined in the library or ordered through the interlibrary
loan service. We did not have to perform this initial screening for the
articles selected from Academic Search Elite because those articles were
available online in full text and their number was manageable (323).
In reviewing the abstracts from the PsychINFO-ERIC database, we
eliminated abstracts of articles that did not appear to include any em-
pirically derived incidence data. We also eliminated studies that fo-
cused solely on assessing perceptions of what constitutes sexual harass-
ment (studies that used vignettes or hypothetical scenarios) or opinions
about the extent of sexual harassment not based on personal experi-
ences and studies that involved experiments. Studies that examined sex-
ual harassment outside work settings (e.g., sexual harassment by domes-
tic partners) or involved the sexual harassment of students, and stud-
ies that were conducted outside the U.S were excluded. We excluded
nonwork studies because the focus of the present project is on sexual
harassment of employees in the workplace; the non-U.S. studies were
excluded because of the potential confounding effects between the cul-
tural or legal environment where a study was conducted and the mod-
erator variables examined in the present study. Studies reporting sexual
harassment incidence in the military, such as the Navy Equal Oppor-
tunity/Sexual Harassment Survey series (e.g., Culbertson & Rosenfeld,
1993; Newell, Rosenfeld, & Culbertson, 1995) or the Department of De-
REMUS ILIES ET AL. 617

fense surveys (e.g., Hay & Elig, 1999; Martindale, 1990) were included
in the d a t a b a ~ e . ~
Following the suggestion of an anonymous reviewer, we examined
the reasons for which the PsychINFO-ERIC abstracts were excluded.
Based on a random 10% sample of the 2,231 abstracts that were selected
from PsychINFO and ERIC, the main reasons for excluding abstracts
were as follows: studies that did not contain empirically derived data,
81%; studies that focused on perceptions or involved experiments, 11%;
studies conducted outside the work environment (including those that
examined the harassment of students) or outside the U.S., 13% (these
percentages do not add up to 100%because some studies were excluded
for more than one reason).
For the 567 studies initially included in our study database (244 have
resulted from the PsycINFO and ERIC selections of abstracts, and 323
from Academic Search Elite), we examined each study to determine
whether it met the criteria detailed above and contained an estimate
of sexual harassment incidence rate based on an independent sample
of females (several studies reported data from the same sample, such
as the USMSPB samples; in such cases we included only the primary
study). We also examined the articles referenced in all studies selected
up to this point in an attempt to identify additional studies. Fourteen
referenced reports that met the criteria detailed above were included in
the database.
In total, 71 studies met our criteria. Nine studies reported data col-
lected from multiple independent samples, and several others reported
two incidence rates based on the same sample but on different types of
surveys (direct query and behavioral experiences). Thus, in all, 96 esti-
mates of sexual harassment incidence from 84 independent samples re-
ported in 71 studies were included in the analyses. All studies that con-
tained estimates included in the meta-analysis are listed as references.

Data Classification and Analyses

We categorized the sexual harassment incidence rates included in our


database according to the following moderator variables: (a) the type of
survey used to estimate average incidence (direct query vs. behavioral
experiences); (b) sampling technique (probability vs. nonprobability);
and (c) the type of work environment in which the study was conducted
(academic, private sector, government, and military studies). Categoriz-
ing the estimates with regard to the type of survey used and with respect

‘From the military studies that were identified by the literature search, we could not
retrieve three unpublished reports.
618 PERSONNEL PSYCHOLOGY

Private Sector
Sampling

LL&
Behavioral

Convenience

'1 Military 1
I Overall
(___________________-----------------------

Academia I
I Private Sector I
I Probability
Sampling 1
tDirect Query ........................................

Convenience

Figure 1: Hierarchical structure for the moderator analyses.'

*Note: The results for categories included in the areas delimited by dashed perimeters
are not reported
REMUS ILIES ET AL. 619

to the work environment was relatively ~traightforward.~ To divide the


reported estimates into estimates based on probability samples and es-
timates based on convenience samples, we developed a coding scheme
as follows. To be included in theprobability category, an estimate had to
be based on (a) a national probability sample (random or stratified ran-
dom) or (b) a probability sample across multiple organizations or in a
multiple-site organization (e.g., government or state employees), or (c)
a sample that resulted from the sampling of the entire sampling frame
(as defined by the study) in a single-site organization. We conducted a
fully hierarchical moderator analysis (Hunter & Schmidt, 1990) by esti-
mating the sexual harassment incidence rates in each of the categories
delimited by the moderator variables. Figure 1 shows the hierarchical
structure used to conduct the moderator analyses. To investigate the
existence of time trends in the incidence data, we recorded the year in
which the data were collected, and we examined the correlation between
the data collection year and the reported incidence rate in the four cat-
egories determined by the type of sampling procedure and the type of
survey used to measure sexual harassment.6

Results

Table 1 presents the results of the seven meta-analyses conducted to


estimate the impact of methodological moderators on the reported in-
cidence rate. In general, the data indicated that moderator variables
did influence the reported rates; the incidence rates in categories de-
termined by the moderators differed, and the corrected standard de-
viation decreased in categories determined by moderator variables as
compared to the broader categories. For example, dividing the overall
sample into studies that used direct query and behavioral experiences
surveys resulted in incidence rates of .35 and .62, and the standard devi-
ation decreased by approximately 16% (we did not expect a substantial
decrease of the variability as other moderating effectsare present in each
of the two categories).’
Respondents from direct query-probability sampling studies re-
ported an average incidence rate of .24, whereas those from direct

51n all studies that used behavioral experiences surveys respondents who endorsed at
least one item were considered to have been sexually harassed.
‘For studies that did not report the year when the data were collected we entered the
year preceding their publication year.
7The overall weighted average contains both direct query and behavioral experiences
estimates. Thus,this estimate is not meaningful by itself. Following Hunter and Schmidt’s
(1990) recommendations for interpreting in hierarchical moderator analyses, we do not
interpret the overall estimate.
TABLE 1
Meta-Analyses of the Reported S a w 1 Harassment Incidence Rate in the Workplace: Methodological Moderators
Analysis K N Mean P sop SE 90% cv 95% CI
Overall 84 111,481 54 .22 .02 .18-.90 .49-.59
Studies that used direct query surveys 44 33,153 .35 .19 .03 .03-.67 .29-.41
Probability sampling 26 19,639 .24 .12 .02 .04-.43 .19-.28
Nonprobability sampling 18 13,514 .51 .16 .04 .24-.78 .44-.59
cd
Behavioral experiences surveys 52 81,268 .62 .18 .03 .32-.92 .57-.67 v1
Probability sampling 35 68,765 .58 .17 .03 .31-.84 .52-.63 4
A

Non-probability sampling 17 12,507 .84 .10 .02 .68-1.0 .79-.89 x


Notes: K = number of correlations. N = combined sample size. Mean P = mean incidence rate. SDp = standard deviation of incidence 9
rate after correction for sampling error. SE= standard error of the mean incidence rate. Whitener’s (1990) procedure for computing the b
standard error was used to estimate the standard error of the mean incidence rate for each meta-analysis. CV = Credibility interval. 2
CI = Confidence interval.
REMUS ILIES ET AL. 621

query-convenience sampling studies reported an average incidence rate


of .5 1; in the behavioral experiences category, respondents selected
through a probability sampling design reported an average incidence
rate of .58, whereas those who were selected with nonprobability de-
signs (convenience samples) reported an average rate of 34. The aver-
age standard deviation across the four categories determined by the type
of survey used and sampling design was 26% smaller than the average
standard deviation across the two broader categories determined by the
type of survey used.8
With respect to the type of work environment in which studies were
conducted, we expected the comparison of incidence rates reported by
respondents employed in academia, the private sector, the government,
and the military to reveal that the lowest incidence rates are reported by
studies conducted in academic organizations and the highest incidence
rates are reported by military studies.
The results of the moderator analyses for the type of work environ-
ment are shown in Thble 2. We only report analyses for incidence rates
reported by probability samples because the results reported in Table
1 show that a sample-selection bias is present in the estimates based on
convenience samples. In addition to the sample-selectionbias, five of the
eight average estimates (4 work environment categories x 2 types of sur-
veys) derived by using convenience samples would be based on only one
or two studies, which does not allow meaningful comparisons of these ef-
fects because of second-order sampling error (Hunter & Schmidt, 1990).
Dividing estimates according to the type of work environment of the
respondents, in general, did influence the average reported rates of sex-
ual harassment as expected, and the average corrected standard devi-
ation decreased by 20%. Within both the direct query and the behav-
ioral experiences categories, as expected, the average rate reported by
military personnel was the highest (see Table 2). The average rate re-
ported by respondents employed in academic settings was, as expected,
the lowest among incidence rates estimated with direct query surveys
(.16 vs. .31 for industry samples, and .36 for military samples) but, unex-
pectedly, it was higher than the estimates obtained from private sector
and government employees within the behavioral experiences category.
Finally, the results of the time trend analyses revealed that the in-
cidence rate reported by participants selected with probability designs

'Within the behavioral experiences category, it is possible that the number of behaviors
listed on the survey moderates the reported incidence rate (e.g., surveysthat list a relatively
large number of behaviors may include less severe behaviors so more respondents are
likely to endorse at least on item from such surveys). To investigate this possibility, we
correlated the number of behaviors listed on the surveys with the reported incidence rate.
The correlation wasweak and nonsignificant ( r = .08, ns),which indicates that the number
of behaviors on the survey has little effect on the reported rate of sexual harassment.
TABLE 2
Meta-Analysesof the Reported Senral Harassment Incidence Rate in the Workplace: Organizational Context Moderator

Analysis K N Mean P SOP SE 90% cv 95% CI


Direct query-probability sampling 26 19,639 .24 .12 .02 .04-.43 .19-.28
Academic samples 12 12,005 .16 .06 .02 .07-.26 .13-.20
Private sector samples 1 304 .23 - .05 - .13-.34
Government samples 3 679 .31 .19 .12 -.01-.62 .08-.53
Military samples 6 5,980 .36 .08 .03 .23-.48 .29-.42
Behavioral experiences-probability sampling 35 68,765 .58 .17 .03 .31-.84 .52-.63
Academic samples 5 1,054 .58 .18 .08 .29-.87 .42-.74
Private sector samples 14 5,023 .46 .13 .04 .25-.68 .39-.53
Government samples 7 21,205 .43 .06 .02 .34-.52 .39-.47
Military samples 4 37,681 .69 .ll .06 -51-.87 .58-.80
Nofe: K = number of correlations. N = combined sample size. Mean P = mean incidence rate. SDp = standard deviation of incidence rate after
correction for sampling error. SE= standard error of the mean incidence rate. Whitener’s (1990) procedure for computing the standard error was
used to estimate the standard error of the mean incidence rate for each meta-analysis. cv = Credibility interval. CI = Confidence interval.
REMUS ILIES ET AL. 623

slightly increased over time, whereas the incidence rate from conve-
nience samples actually decreased. That is, for probability samples,
within the direct query category the correlation between the year in
which the data were collected and the reported incidence rate was T = .ll
( N = 26), and within the behavioral experiences category the correla-
tion was T = .26 ( N = 35). In contrast, for convenience samples, the
correlations between the year in which the data were collected and the
reported incidence were T = -.33 ( N = 18) and T = -.36 ( N = 17) in
the direct query and behavioral experiences category, respectively.
We need to point out that in all meta-analyses conducted in this
study statistical artifacts explained modest proportions of the variance
observed in reported incidence rate. Given the nature of the data, it is
a result that should be expected. That is, sampling error did not explain
a great proportion of the observed variance because of the very large
samples reported in some studies (estimates from those samples contain
practically no sampling error).

Discussion

The major objective of this study was to provide cumulative esti-


mates of the incidence of sexual harassment in the workplace and iden-
tify methodological and organizational factors that impact reported in-
cidence rates. Results show that the type of surveys used to assess the ex-
tent of sexual harassment and the type of sampling procedure used to se-
lect respondents have a substantial impact on the reported incidence rate.
The meta-analytic review gives two independent estimates for the
prevalence of sexual harassment at work. Estimates based on directly
querying the survey participants with regard to their experience of sex-
ual harassment and those based on reports of experiencing incidents be-
lieved by researchers to constitute sexual harassment tell us different
stories. Whereas the direct query method gives an estimate of the inci-
dence of sexual harassment based on respondents’ perceptions, the be-
havioral experiences method estimates the extent to which incidents that
are potentially harassing happen in organizations.
Across a variety of work environments and based on 86,578 respon-
dents from 55 independent probability samples, 58% of women report
having experienced potentially harassing behaviors and 24% report hav-
ing experienced sexual harassment at work. The difference between the
two estimates reflects, at least in part, differences between researchers’
and respondents’ definitions of sexual harassment, as well as, perhaps,
the respondents’ willingness to actually call the behaviors sexual harass-
ment (Blumenthal, 1998; Gutek, 1995; Magley et al., 1999; Pryor, 1995).
The fact that less than half as many women report to have experienced
624 PERSONNEL PSYCHOLOGY

sexual harassment, compared to those reporting to have experienced


at least one behavior from a behavioral experiences survey, shows that
women are reluctant to label offensive experiences as sexual harassment
and, thus, provides strong evidence against the “whiner hypothesis.”
Comparing incidence rates across academic, private-sector, govern-
ment, and military respondents, we found that sexual harassment is most
prevalent in the military (regardless of the surveying method) and that
a lower percentage of women employed in academia consider them-
selves to have experienced sexual harassment, as compared with the
other work environments. However, in the behavioral experiences cat-
egory, contrary to our expectations and in contrast to the comparison
of incidence rates derived with direct query surveys, women employed
in the academia actually reported a higher average incidence rate than
women employed in the private sector and the government. This in-
triguing effect suggests that even though women employed in academia
experience more potentially harassing behaviors than women employed
in the private sector and the government, when they are asked whether
they have been sexually harassed, they report less incidents of sexual
harassment (according to their own definition). This explanation is con-
sistent with the results of Schneider, Swan, and Fitzgerald (1997) who
administered the SEQ to 447 women employed in the private sector and
to 300 female faculty and staff employed at a university. In that study,
even though a similar percentage of women from the two samples en-
dorsed at least one SEQ item (68% and 63% for the private-sector and
academic samples, respectively), almost twice as many industry-sector
respondents who have endorsed at least one harassing behavior labeled
their experience as sexual harassment, as compared with women from
the academic sample (28%vs. 15% for the private-sector and academic
samples, respectively). It may be the case that even though potentially
harassing behaviors happen with a similar frequency in academia and the
private sector, due to different employment security and power struc-
tures, women employed in universities feel less threatened by these be-
haviors. We certainly do not have data to support this speculation or to
refute alternative explanations. Future research should further investi-
gate this effect.
Because the overall estimates for the sexual harassment assessed with
direct query and behavioral experiences surveys presented above (24%
and 58%) represent data reported by employees from various types of
work environments, it is possible that a confounding effect of the type of
work environment is responsible for the difference between the two es-
timates. Our fully hierarchical moderator analysis presented in Table 2
allows us to investigate this possibility. That is, within academia, the
average incidence of direct query sexual harassment was 16%, whereas
REMUS ILIES ET AL. 625

the average incidence of behavioral experiences harassment was 58%.


Similarly, for military samples, the average incidence for direct query
reports was 36%, and the average incidence of behavioral experiences
reports was 69%. These results suggest that, even though the difference
between the two types of estimate is somewhat lower in the military, the
type of survey used to assess sexual harassment does have a substan-
tial impact on the reported incidence rate in specific work environment
categories, as it does in the overall sample. Thus, the type of the work
environment does not seem to be a principal cause for the difference
between the overall estimates obtained with direct query and behavioral
experiences surveys.
The time trend analyses revealed an upward trend for the incidence
rates reported by probability samples and a downward trend for the rates
reported by convenience samples. The upward trend in the probability-
direct query data were weak (r = .11) and, thus, we do not interpret
it. The upward trend for probability-behavioral experiences data were
somewhat stronger (T = .26), suggesting that, over the 25years examined
in this study, women report increasingly more behavioral experiences
that represent sexual harassment. However, it is also possible that over
this time period, researchers started to include milder behaviors on their
survey, which lead to increased incidence reports. Our data cannot
investigate whether these speculations are true; only time trend analyses
of data obtained with the same instruments can answer such questions.
The incidence rates reported by women selected with nonprobability
designs showed downward trends in both the direct query and the behav-
ioral experiences categories (T = -.33 and T = -.36). As noted, these
data are subject to sample-selection biases. Most likely, the downward
trend in the incidence reflects a decrease in the overreporting bias as-
sociated with early convenience samples (e.g., Silverman [1976] used a
sample of women who attended a speakout on sexual harassment).

Contributions

The first major contribution of the present study resides in providing


solid quantitative estimates of the extent of sexual harassment in organi-
zations as assessed by specific types of surveys and different sampling
techniques. Second, our results refute the “whiner hypothesis” and,
given the large cumulative sample sizes on which these results are based,
our evidence against this hypothesis makes a much stronger case than
any individual study could make. The meta-analytical evidence in sup-
port of the relationship between the type of work environment in which
the study was conducted and sexual harassment incidence constitutes the
third major contribution of this study.
626 PERSONNEL PSYCHOLOGY

Implications

In order to decrease the prevalence of sexual harassment at work,


organizations first need to know how to measure sexual harassment, that
is, they need accurate means to monitor both employees’perceptions of
sexual harassment and the occurrence of behaviors that are potentially
harassing. Our results inform organizational researchers in that they
show that the method of assessing sexual harassment has a substantial
impact on the reported incidence rate. With respect to the impact of the
type of work environment on the reported incidence rate, the results of
this study suggest that highly structured organizations with large power
differentials among organizationallevels should pay increased attention
to sexual harassment and implement specific training and prevention
programs, as they are probably more sexual harassment prone than other
organizations.
From a policy perspective, our research reaffirms the assertion that
women often do not report unwanted behaviors as sexual harassment. To
the extent that women either do not consider some behaviors as consti-
tuting sexual harassment or do not want the stigma of identifying them-
selves as victims (e.g., Magley et al., 1999), they may not initially report
behaviors or may report but not use the label “sexual harassment.” (e.g.,
Bremer, Moore, & Bildersee, 1991; Brooks & Perot, 1991). If a formal
complaint is filed at a later date, the failure to report initially may be
used to discount the woman’s story. It would be unfortunate if policy-
makers or the judiciary insisted on women’s initial labeling of behaviors
as sexual harassment in order to pursue a claim.

Limitations

Though the findings reported here contribute to the literature on sex-


ual harassment, this study has certain limitations that merit discussion.
First, like most meta-analyses,this study cannot test causal relationships.
Even though the results supported our prediction with respect to the ef-
fect of organizational context, these results should be interpreted with
caution. Second, it should be noted that moderators not included in
this meta-analysis are likely nested within the methodological moder-
ator categories that we used. The type of behaviors listed on behavioral
experiences surveys, for example, will certainly moderate the reported
incidence rate (i.e., gender harassment behaviors do happen more often
than sexually coercive behaviors; e.g., Fitzgerald et al., 1988; Richman
et al., 1999; Schneider et al., 1997). Such effects would explain the dif-
ferences in the primary incidence rates that were not explained by the
moderator analyses conducted in this study. Unfortunately, as it is often
REMUS ILIES ET AL. 627

the case in conducting meta-analyses,we are unable to address such po-


tential nesting effects because we are limited to the information reported
in the primary studies (e.g., studies often do not report item-level data
or are not consistent in reporting item- or behavioral category-level re-
sults). We recommend that authors of primary studies report incidence
data for the specific behavioral categories included in the SEQ in or-
der to enable future integrative research to investigate moderator effects
more fully.

Conclusions

Sexual harassment is an important social and organizational phe-


nomenon with broad implications for individuals, organizations, and so-
ciety. This paper contributes to the literature on sexual harassment by
calibrating the impact of methodological moderators on the reported in-
cidence of sexually harassing behaviors in organizations, and providing
suggestive evidence for impact of the type of work environment on the
incidence rates reported in different organizations. Researchers, man-
agers, and policy makers should find the present results useful for design-
ing future studies, selecting measurement instruments of sexual harass-
ment, comparing results from different studies, evaluating the extent of
sexual harassment in various settings, and studying the effect of the law.

REFERENCES
References marked with an asterisk indicate studies included in the meta-analysis.

Alger A, Flanagan WG. (1996). Sexual politics: Sexual harassment in the workplace.
Forbes, 157,106-110.
Arvey RD, Cavanaugh MA. (1995). Using surveys to assess the prevalence of sexual
harassment: Some methodological problems. Journal of Social Issues, 51,3!%52.
'Bails CI:(1995). Females' reactions to serual harassment in the workplace and the impact
of race. Unpublished doctoral dissertation, Temple University.
'Baker NL. (1990). S m l harassment and job satlsfactwn in traditionaland nontraditional
industrial occupations. Unpublished doctoral dissertation, California Schoolof Pro-
fessional Psychology.
*BaldridgeK, McLean G. (1980). Sexual harassment: How much of a probbm is it.. .really?
Journal of Business Education, 55,294-297.
'Blakely GL, Blakely EH, Moorman RH. (1995). The relationship between gender, per-
sonal experience,and perceptions of sexual harassment in the workplace. Empfoyee
Responsibilities and Rights Journal, 8,263-274.
Blumenthal JA. (1998). The reasonable woman standard: A meta-analytic review of
gender differences in perceptions of sexual harassment. Law and Human Behavior;
22,33-57.
Braine JD, Bless C, Fox PMC. (1995). How do students perceive sexual harassment? An
investigation on the University of Natal, Pietermaritzburg campus. South African
Journal of Pychology, 25,140-149.
628 PERSONNEL PSYCHOLOGY

*Bremer BA, Moore Cr,Bildersee EE (1991). Do you have to call it “sexual harassment”
to feel harassed? College Student Journal, 25,258-268.
*Brooks L, Perot AR. (1991). Reporting sexual harrassment: Exploring a predictive
model. Psychology of Women Quarter& 15,3147.
Bums T, Stalker GM. (1961). The management of innovation. London: Tavistock Press.
‘Cardinale V. (1992,March 9). Sexual harassment on the job. Drug Topics, 136,pp. 40,
43-46,49-50,53-54.
‘Carol1 L, Ellis KL. (1989). Faculty attitudes toward sexual harassment: Survey results,
survey process. Initiatives: Journal of the National Association for Women Deans,
Administrators,and Counselors, 52,3541.
“CulbertsonAL, Rosenfeld F! (1993). Understanding sexual harassment through organiza-
tional surveys. In Rosenfeld P, Edwards JE (Eds.), Improvingorganizationalsurveys:
New directions, methods, and applications (pp. 164-187). Newbury Park, C A Sage.
*DanskyBS, Kilpatrick DG. (1997). Theoretical explanations for the effects of sexual ha-
rassment. In O’DonohueW (Ed.), Sexual harassment theory, research, and treatment
(pp. 152-174). Needham Heights, M A Allyn & Bacon.
‘Daugherty S,Baldwin DC, Rowley BD. (1998). Learning, satisfaction, and mistreatment
during medical internship: A national survey of working conditions. J A M : Journal
of the American Medical Association, 279,11944199.
‘DeMayo RA. (1997). Patient sexual behavior and sexual harassment: A national survey
of female psychologists. Professional Psychology Research and Practice, 28,58-62.
‘Dey EL, Korn JS, Sax LJ. (1996). Betrayed by the academy: The sexual harassment of
women college faculty. Journal of Higher Education, 67,149-173.
*Duldt BW. (1982). Sexual harassment in nursing. Nursing Outlook, 30,336-343.
‘Fitzgerald LF, Shullman SL, Bailey N, Richards M, Swecker J, Gold Y,Ormerod AJ,
Weitzman L. (1988). The incidence and dimensions of sexual harassment in aca-
demia and the workplace. Journal of Vocational BehavioG 32,152-175.
Foulis D, McCabe MF! (1997). Sexual harassment: Factors affecting attitudes and per-
ceptions. Sex Roles, 37,773-798.
Gardner J, Allen E (1996). Sexual and gender harassment at university: Experiences and
perceptions of Australian women. Australian Psychologist, 31,210-216.
*Goldenhar LM, Swanson NG, Hurrell JJ Jr., Ruder A, Deddens J. (1998). Stressors and
adverse outcomes for female construction workers. Journal of OccupationalHealth
Psychology, 3,19-32.
*Goodwin MP, Roscoe B, Clarkson SE, Rose M. (1989). Sexual harassment experiences
of university employees. Initiatives: Journal of the National Associationfor Women
Deans, Administrators,and Counselors, 52,25-33.
*Grant SM. (2000). Incidence and risk factors of sexual harassment among working women:
Implications for counseling psychology and the workplace. Unpublished doctoral
dissertation, University of California.
‘Grauerholz E. (1989). Sexual harassment of women professors by students: Exploring.
the dynamics of power, authority, and gender in a university setting. SexRoles, 21,
789-801.
*Grieco A. (1987). Scope and nature of sexual harassment in nursing. Journal of Ser
Research, 23,261-266.
*Griffin-ShelleyE. (1985). Sexual harassment: One organization’s response. Journal of
Counselingand Development, 64,72-73.
Gruber JE. (1990). Methodological problems and policy implications in sexual harass-
ment research. Population Research and Policy Review, 9,235-254.
‘Gruber JE, Bjom L. (1982). Blue-collar blues: The sexual harassment of women au-
toworkers. Work and Occupations, 9,271-298.
REMUS ILIES ET AL. 629

GutekBA. (1995). How subjectiveis sexual harassment? An examination of rater effects.


Basic and Applied Social Pychology, 17,447-467.
*Gutek BA, Cohen AG, Konrad AM. (1990). Predicting social-sexual behavior at work
A contact hypothesis. Academy of Management Journal, 33,560-577.
*Gutek BA, Nakamura CY, Gahart M, Handschumacher I. (1980). Sexuality and the
workplace. Basic and Applied Social Psychology, 1,255-285.
'Harder VJ. (1995). Sexual harassment: Employee reactions and organizational conse-
quences. Unpublished doctoral dissertation, Marquette University.
*Hargrow AM. (1997). Speaking to our realities: From speculation to truth concerning
Afican American women's experiences of sexual harassment. Unpublished doctoral
dissertation, The Ohio State University.
*Hay MS, Elig Tw.(1999). The 1995 Department of Defense sexual harassment survey:
Overview and methodology. Militay Pychology, 11,233-242.
Hunter JE, Schmidt FL. (1990). Methods of meta-analysis: Correcting emor and bias in
researchjindings. Newbury Park, C A Sage.
*Kelly RM, Stambaugh PM. (1992). Sexual harassment in the states. In Guy M (Ed.),
Women and men of the states public administrators at the state level (pp. 109-124).
Armonk, Ny:Sharpe.
'Komaromy M, Bindman AB, Haber RJ, Sande MA. (1993). Sexual harassment in
medicine. New England Journal of Medicine, 328,322-326.
'Lafontaine E, Tredeau L. (1986). The frequency, sources, and correlates of sexual ha-
rassment among women in traditional male occupations. SexRoles, 15,433-442.
Lengnick-Hall ML. (1995). Sexual harassment research: A methodological critique.
PERSONNEL PSYCHOLOGY,48,841-864.
*Loy P, Stewart L. (1984). The extent and effects of the sexual harassment of working
women. Sociological Focus, 17,31-43.
Luthar HK. (1995). The neglect of critical issues in the sexual harassment discussion: Im-
plications for organizational and public policies. Journal of Individual Employment
Rights, 4,261-0276.
*Magley VJ, Hulin CL, Fitzgerald LF, DeNardo M. (1999). Outcomes of self-labeling
sexual harassment. Journal of Applied Psychology, 84,390-402.
*Mansfield PK, Koch PB, Henderson J, Vicary JR. (1991). The job climate for women in
traditionally male blue-collar occupations. Sex Roles, 25,63-79.
*Martindale M. (1990). Sexual harassment in the militay: 1988. Arlington, VA Defense
Manpower Data Centers.
'Maypole DE. (1986). Sexual harassment of social workers at work: Injustice within?
Social Work 31,2%34.
*Maypole DE, Skaine R. (1982). Sexual harassment of blue collar workers. Journal of
Sociology and Social Weyare, 9,682-695.
*McIntyreDI, Renick JC. (1982). Protecting public employees and employers from sexual
harassment. Public Personnel Management, 11,282-292.
'McKinney K. (1990). Sexual harassment of university faculty by colleagues and students.
Sex Roles, 23,421-438.
*Morris A. (1996). Gender and ethnic differences in social constraints among a sam-
ple of New York City police officers. Journal of OccupationalHealth Psychology, I ,
224-235.
*Morrow PC, McElroy JC, Phillips CM. (1994). Sexual harassment behaviors and work
related perceptions and attitudes. Journal of Vocational Behavior, 45,295-309.
*MumellAJ, Olson JE, Frieze IH. (1995). Sexual harassment and gender discrimination:
A longitudinal study of women managers. Journal of Social Issues, 51,139-149.
Munson LT,Miner AG, H u h C. (2001). Labeling sexual harassment in the military: An
extension and replication. Journal of Applied Psychology, 86,293-303.
630 PERSONNEL PSYCHOLOGY

*Newell CE, Rosenfeld P, Culbertson AL. (1995). Sexual harassment experiences and
equal opportunity perceptions of Navy women. Sex Roles, 32,159-168.
Niebuhr RE, Boyles WR. (1991). Sexual harassment of military personnel: An exam-
ination of power differentials. International Journal of Intercultural Relations, 15,
445-457.
*Piotrkowski CS.(1998). Gender harassment, job satisfaction, and distress among em-
ployed White and minority women. Journal of Occupational Health Pychology, 3,
33-43.
*Powell GN. (1983). Definition of sexual harassment and sexual attention experienced.
Journal of Psychology, 113,113-117.
Pryor JB. (1995). The phenomenology of sexual harassment: Why does sexual behav-
ior bother people in the workplace? Consulting Psychology Journal: Practice and
Research, 47,160-168.
*Pryor JB, LaVite CM, Stoller LM. (1993). A social psychological analysis of sexual
harassment: The persodsituation interaction. Journal of Vocational Behavior; 42,
68-83.
Pryor JB, McKinney K. (1995). Research on sexual harassment: Lingering issues and
future directions. Basic and Applied Social Psychology, 17,605-611.
*Ragins BR, Scandura TA. (1995). Antecedents and work-related correlates of reported
sexual harassment: An empirical investigation of competing hypotheses. Sex Roles,
32,429-455.
*Richman JA, Rospeda KM, Nawyn SJ, Flaherty JA, Fendrich M, Drum ML, Johnson
T€?(1999). Sexual harassment and generalized workplace abuse among university
employees: prevalence and mental health correlates. American Journal of Public
Health, 89,358-363.
'Rose11 E,Miller K, Barber K. (1995).Firefighting women and sexual harassment. Public
Personnel Management, 24,339-350.
'Rosen LN, Martin L. (1998). Incidence and perceptions of sexual harassment among
male and female U S . Army soldiers. Military Psychology, 10,239-257.
*Rosenberg J, Perlstadt H, Phillips WR.(1993). Now that we are here: Discrimination,
disparagement, and harassment at work and the experience of women lawyers.
Gender and Society, 7,415433.
Rotundo M, Nguyen DH, Sackett PR. (2001)A meta-afialytic review of gender percep-
tions of sexual harassment. Journal of Applied Psychology, 86,914-922.
*Safran C. (1976). What men do to women on the job: A shocking look at sexual harass-
ment. Redbook, 149,217-223.
*Sandroff R. (1992). Sexual harassment: The inside story. Working Woman, 17,47-51.
*Sangamon State University. (1979). Study of unwanted sexual attention received by female
state employees: Preliminary data. Springfield, IL: Center for Policy Studies and
Program Evaluation.
*Schneider m, Swan S, Fitzgerald LE (1997). Job-related and psychological effect's of
sexual harassment in the workplace: Empirical evidence from two organizations.
Journal ofApplied Psychology, 82,401415.
Scott WR. (1998). Organizations: rational, natural, and open systems (4th ed.). Upper
Saddle River, NJ: Prentice Hall.
*Sigler RT, Johnson IM. (1986). Public perceptions of the need for criminalization of
sexual harassment. Journal of Criminal Justice, 14,229-237.
*Silverman D. (1976).Sexual harassment: Working women's dilemma. Quest: A Ferninkt
Quafie*, 3,15-24.
*SkinnerKM, Kressin N, Frayne S, 'Itipp TJ, Hankin CS, Miller DR, Sullivan LM. (2000).
The prevalence of military sexual assault among female Veterans' Administration
outpatients. Journal of Interpersonal Violence, 15,291-310.
REMUS ILIES ET AL. 631

*Stank0 B, Werner CA. (1995). Sexual harassment: What is it? How to prevent it?
National Public Accountant, 40, 14-16.
*Stedham Y, Mitchell MC. (1998). Sexual harassment in casinos: Effects on employee
attitudes and behaviors. Journal of Gambling Studies, 14,381-400.
'Stockdale MS, Vaux A. (1993). What sexual harassment experiences lead respondents to
acknowledge being sexually harassed? A secondary analysis of a university survey.
Journal of Vocational Behavioc 43,221-234.
Stockdale M S , Vaux A, Cashin J. (1995). Acknowledging sexual harassment: A test of
alternative models. Basic and Applied Social Pychology, 17,469-496.
*United States Merit Systems Protection Board. (1981). Sexual harassment in the federal
government: Is it aproblem? Washington, DC: United States Government Printing
Office.
*United States Merit Systems Protection Board. (1987). Sexual harassment of federal
workers: An update. Washington,D C United States Government Printing Office.
*United States Merit Systems Protection Board. (1994). Sexualharassment in the federal
workplace: Trends, progress, and continuing challenges. Washington, DC: United
States Government Printing Office.
Viswesvaran C, Schmidt FL. (1992). A meta-analytic comparison of the effectiveness of
smoking cessation methods. Journal of Applied Pychology, 77,554-561.
*Vukovich MC. (1996). The prevalence of sexual harassment among female family prac-
tice residents in the United States. Violenceand Victims, 11,175-180.
'Walsh-Childers K, Chance J, Herzog K. (1996). Sexual harassment of women journalists.
Journalism and Mass Communication Quarter& 73,559-81.
*White JE. (1991). Sexual harassment in the workplace. Unpublished doctoral dissertation,
Kansas State University.
Whitener EM. (1990). Confusion of confidence intervals and credibility intervals in meta-
analysis. Journal ofApplied Pvchologx 75,315-321.
'Wolfe J, Sharkansky EJ,Read JP, Dawson R, Martin JA, Ouimette PC. (1998). Sexual
harassment and assault as predictors of PTSD symptomatology among U.S. female
Persian Gulf War military personnel. Journal of Interpersonal Violence,13,40-57.
'Wong MM. (1984). Serual harassment at work: Female police oficers. Unpublished
doctoral dissertation, University of Pittsburgh.
'Wyat GE, Riederle M. (1995). The prevalence and context of sexual harassment among
African American and White American women. Journal of Interpersonal Violence,
10,306-319.
*Yoder JD, Aniakudo F! (1996). When pranks become harassment: The case of African
American women firefighters. Sex Roles, 35,253-270.

View publication stats

Vous aimerez peut-être aussi