Sexual Harassment in the Federal Workplace

Author(s): Meredith A. Newman, Robert A. Jackson, Douglas D. Baker

Source: Public Administration Review, Vol. 63, No. 4 (Jul. - Aug., 2003), pp. 472-483
Published by: Blackwell Publishing on behalf of the American Society for Public Administration
Stable URL: http://www.jstor.org/stable/977402 .
Accessed: 26/01/2011 05:19

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=black. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Blackwell Publishing and American Society for Public Administration are collaborating with JSTOR to
digitize, preserve and extend access to Public Administration Review.

http://www.jstor.org
MeredithA. Newman
WashingtonStateUniversity-Vancouver
RobertA. Jackson
FloridaStateUniversity
DouglasD. Baker
WashingtonStateUniversity

Sexual Harassment in the Federal Workplace

Whatfactorsinfluencethelikelihoodthata federalworkerwillreceiveunwantedsexualattention?
Who is mostlikelyto be accusedof sexual harassment?Whatfactorsinfluencefederalworkers'
perceptionsof the effectivenessof agency sexual harassmenttraining?Usingthe raw data file of
the U.S. MeritSystemsProtectionBoard'smostrecentsurveyon sexual harassment,the authors
find that workercharacteristicsare the principalinfluenceon the likelihoodthat a workerwill
receive unwantedsexual attentionand whetheran agency'ssexual harassmenttrainingis per-
ceived favorably.Contextualfactorsdemonstratelesser influence.Theirconclusionslead the au-
thorsto believe thata reevaluationof trainingprogramsis in order.A one-size-fits-alltraining
approachmay no longerbe tenable,if it ever was.

Introduction
Sexual harassmentcontinues to be an ingrained fea-
ture of the workplacefor many of America's workersde-
spite some three decades of research into its causes and
effects, and despite a precedentialruling in 1976 (Will-
iams v. Saxbe) classifying sexual harassmentin the work-
place as a form of sex discriminationand a violation of
TitleVII of the Civil RightsAct of 1964. Since the Clarence
Thomas confirmationhearings fixed the nation's atten-
tion on this topic some 10 years ago, incidents of sexual
harassmentin the public workplacehave continuedapace.
The Navy's Tailhookscandal,the Army'sAberdeenProv-
ing Groundsconvictions, and the abuse of female cadets
at the Citadelraisethe specterthatsexual harassmentmay
be the price of integratingwomen into traditionallymale
bastions, at least in the militarycontext (Newman 1999).
Allegations surroundingPresidentClinton's behaviorto-
wardhis female subordinatesrekindledthe nation'sinter-
est in the issue.
Within the public sector more broadly,the numberof
sexual harassmentcharges filed with the Equal Employ-
mentOpportunityCommission(EEOC)andwith stateand
local fairemploymentpracticesagencies aroundthe coun-
trytotaled 15,836 in 2000, up from 10,532 in 1992 (EEOC
2001). A multistatestudy of upper-levelpublic adminis-
tratorsin state agencies reportedthat 6 percentto 16 per-
cent of administrative-levelfemale public administrators
have experienced unwanted sexual advances in connec-
tion with theirwork.The same studyreportedthatup to 24
percentof the women had experiencedrequestsfor sexual
favors,up to 36 percentoffensive physical contact,and up
to 57 percentoffensive verbalbehavior(Kelly 1995, 196).
At the federallevel, the incidence of sexual harassment
is widespread and pervasive across agencies. The latest
MeredithA. Newman is an associateprofessorof publicaffairsat Wash-
ingtonStateUniversity-Vancouver. Herarticleson publicmanagementhave
appearedin a numberof scholarlyjournals,includingPublicAdministra-
tion Review,AmericanReviewof PublicAdministration,Administrative
Theoryand Praxis,Womenand Politics,PublicIntegrity, PublicPerformance
and ManagementReview,and Reviewof PublicPersonnelAdministration.
Newmanis a past NationalCouncilrepresentative of ASPAand past chair
of the Section for Women in PublicAdministration.Email:newman@
vancouver.wsu.edu.
RobertA. Jacksonis an associateprofessorof politicalscienceat Florida
StateUniversity.Hisresearchon votingbehaviorand electionshas appeared
in a varietyof journals,includingAmericanPoliticsQuarterly,Journalof
Politics, and Political Research Quarterly. Email: riackson@garnet.
acns.fsu.edu.
DouglasD. Bakeris a professorof managementand viceprovostfor aca-
demicaffairsat Washington StateUniversity.
Hisresearchinterestscomprise
boththe individualand organizationalleveland includechangein organi-
zationalstructureand strategy,decisionmaking,goal setting,sexualharass-
ment,and teachingand learningeffectiveness.Someof his workhas been
publishedin Acadfemyof ManagementJournal,DecisionSciences,Group
and OrganizationStudies,HumanPerformance, Journalof AppliedPsy-
chology,Journalof Organizational andSexRoles.Email:ddbaker@
Behavior,
wsu.edu.

472 Public Review* July/August

Administration 2003,Vol.63, No.4
studyby the U.S. MeritSystems ProtectionBoard(MSPB
1995) examined the nature and extent of sexual harass-
ment in the federalgovernment.By any measure,its find-
ings arestaggering 44 percentof female respondentsand
19 percent of male respondentsreported some form of
unwanted sexual attention during the preceding two years.
These rates are similar to those reportedin the MSPB's
1980 and 1987 studies.'The 1995 MSPB study defined
such sexual attentionbroadly and, beyond strictly legal
parameters,as occurringin situationsrangingfrom actual
or attemptedrape or assault to uninvited and unwanted
sexual teasing, jokes, remarks,or questions.2More nar-
rowly, 15 percentof all respondents(25 percentof women
and 6 percentof males) in the 1995 MSPB studyreported
they had "personallybeen the victim of sexual harassment
on a job in the FederalGovernment"(question39, empha-
sis added).3
We note with some alarmthat the incidence of sexual
harassmentapparentlyhas not decreased over time, de-
spite researchinto and trainingon sexual harassment.It
wouldseemthatsexualharassmenttraining,whichhas been
implementedacrossa wide rangeof public-sectoremploy-
ment settings, remains largely unable to effect positive
change.The MSPB study's authorsconclude, in part,that
despiteie verywidespreadtrainingandinformationefforts
that have successfully raised workforcesensitivity to the
issues surroundingsexual harassment,the persistence of
this amountof unwantedsexual attentionin the Federal
workplace suggests that the government's programs to
eradicate the problem need some serious reexamination"
(MSPB 1995, viii, emphasisadded).
The purposeof this articleis to begin such a reexamina-
tion. Using the most recent datafrom the U.S. Merit Sys-
tems ProtectionBoard,we explore a numberof questions
aboutsexual harassmentin the federalgovernmentwork-
place to shed light on core concerns and to lay a founda-
tion for subsequentresearchin this area.Ourfindingsiden-
tify the types of workerswho are most likely to be targets
of sexual harassment,the types of workersmost likely to
be accusedof harassment,andfactorsthatinfluencework-
ers' perceptionsof the effectiveness of agency sexual ha-
rassmenttraining.We conclude, in part, that a one-size-
fits-all trainingapproachmay no longer be tenable.
Researchon Sexual Harassmentand
Sexual HarassmentTraining
Throughoutthe 1980s and 1990s, researchdocumented
the range and scope of sexual harassment(Baldridgeand
MacLean 1980; Fitzgerald et al. 1988; Reese and
Lindenberg1999; MSPB 1981, 1988, 1995). Researchon
sexualharassmenthas been undertakenfrommultiplefoci.
A numberof studies have examinedworkers'perceptions
of sexual harassment(Terpstraand Baker 1986; Baker,
Terpstra,and Larntz1990; Cohen and Gutek 1985; Pryor
and Day 1985; Pryor 1985; Baugh 1997). This research
reveals a numberof factors-such as sex, religiosity, and
the proportionof women in the work group-that affect
the perception of harassing behaviors and that, in turn,
may lead to conflict andmisunderstanding(Biviano 1997).
Moreover,the relationshipbetween sexualharassmentand
power, aggression (Pryor 1992; Ely 1995; Bargh et al.
1995; Moorman and Fountroy 1992; Pryor and Stoller
1994), and gender (Riger 1992; Kelly and Stambaugh
1992; Guy 1993; Newman 1995; Berdahl, Magley, and
Waldo 1996; Baker,Terpstra,and Cutler 1990) have been
examined rigorously.
Much of this literatureshares a common prescriptive
conclusion-namely, that trainingshould be a core com-
ponentof any sexualharassmentpolicy. Sexualharassment
trainingtakes manyforms.Forexample,the differencesin
perceptionbetweenthe sexes areusuallyaddressedin train-
ing programsthat seek to establish a clear definition of
what will and what will not be toleratedin the workplace
(LindenbergandReese 1996). Variationsin emphasisnot-
withstanding,most sexual harassmenttrainingoccurs as a
one-time event and follows a relatively standardscript,
reflecting organizationalconcern for the legal liabilities
associated with the occurrenceof behavior that violates
EEOC guidelines.
Several studies have been conducted on the effects of
sexual harassmenttrainingprogramson awareness,per-
sonalprevention,behavioralchanges(Barak1994;Gilbert,
Heesacker,and Gannon 1991; Licata and Poovich 1987),
and perceptualexpertise (Moyer and Nath 1998; Blakely,
Blakely, and Moorman 1998). In a recent study, Biviano
(1997) evaluatedthe effectiveness of trainingprogramsin
developing an understandingof the definition of sexual
harassmentand the roles of employers, supervisors,and
employees in its prevention.In his report, Biviano sug-
gests that a consensus on what constitutes sexual harass-
ment would likely result in improvedcommunicationand
behavior,in turnleading to fewer incidents of sexual ha-
rassment.Indeed,much of the literatureon sexual harass-
menttrainingindicatesthattrainingshouldestablisha clear
definitionof sexual harassmentand promotethe develop-
ment of an organizationalconsensus on what constitutes
harassingbehaviors.This creation of a common starting
point may be the principalbenefit of training.
While the argumentsfor sexual harassmenttrainingare
reasonable,they aresubjectto challengeandremainlargely
unsupported by empirical evidence (Fitzgerald and
Shullman 1993; Newman 1995; Pryor and McKinney
1995). Only a handfulof publishedstudies have assessed
the effects of sexual harassment training empirically
(Beauvais 1986;Thomann,Strickland,andGibbons 1989;
Workplace473
intheFederal
SexualHarassment
Maurizo and Rogers 1992). However, these pioneering
studies permit only limited conclusions because of their
design and other limitations(Moyer and Nath 1998).
Despite nagging questions aboutthe efficacy of sexual
harassmenttraining,privateand public organizationsare
spending a great deal of time and money on it. For ex-
ample,FordMotorCompany'srecent$7.75 million settle-
mentwith as manyas 900 womencalls for sensitivitytrain-
ing at Fordplants across the nation, at a cost estimatedby
the EEOC to be $10 million (Robinson 1999). Further-
more, at every organizationallevel, every federalgovern-
ment agency provides sexual harassmenttraining(MSPB
1995, xi). While most agencies know whethersuch train-
ing is popularor unpopularwith participants,they gener-
ally do not know what kind of trainingworks best, what
parts of the training curriculumare effective, why inci-
dence ratesvary by agency,or whetherany of the training
has a negative effect on some employees (MSPB 1995,
xi-xii). Traininggoes on with the presumptionthat it is
doing some good. In recognitionof this apparentdilemma,
the U.S. MeritSystems ProtectionBoardrecommendsthe
"trainingcontent should be revised if it is found to make
no appreciabledifferencein preventingor stoppingsexual
harassment"(MSPB 1995, xii). Generic programsaimed
at an agency's entireworkforcemay be missing the mark.
Althoughresearchindicatesthat sexual harassmenttrain-
ing may heighten awareness and clarify definitions, the
questionsof whetherit detersharassmentor whethertrain-
ees think it is helpful remainlargely unanswered.
ResearchQuestions
In the present study, we focus on a set of basic ques-
tions about sexual harassmentin the federal government
workplace:
1. What factors influence the likelihood that a federal
workerwill receive unwantedsexual attention?We as-
sess the influence of workercharacteristics,agency lo-
cation, daily work setting, and the nature of agency
sexual harassmenttraining.
2. What types of workers-defined by demographicand
job characteristics-are most likely to be accused of
sexual harassment?
3. Whatfactorsinfluencefederalworkers'perceptionsof the
effectivenessof agency sexual harassmenttraining?We
assessthe influenceof bothworkercharacteristicsandre-
ceiptof unwantedsexualattentionin the workplace.
Methodologyand Data
We obtainedthe rawdatafile of the U.S. MeritSystems
ProtectionBoard's 1994 survey (MPSB 1995).4In April
1994, the board sent survey questionnairesto a random
sample of almost 13,200 federalemployees situatedin 22
474 Public
Administration
Reviewa July/August
2003,Vol.63, No.4
departmentsand agencies nationwide. The response rate
was 61 percent, with a total of 8,081 questionnairesre-
turned.The survey methodology is detailed in the U.S.
MeritSystems ProtectionBoard'sstudy,andthe appendix
contains a brief statistical overview of variablesthat we
incorporatein our analyses.Whereasthe board'sreportis
based on frequencyanalyses and descriptive,bivariateex-
aminations,we presentmultivariatemodels thatenable us
to assess the relative importanceof a numberof indepen-
dent variables.
From a social science perspective,the majorlimitation
of the report is its failure to account for the concept of
statisticalcontrol.To illustrate,women aremorelikely than
men to be victimsof sexualharassmentin the federalwork-
place. No doubt,this descriptivefact is tremendouslyim-
portant.But we also know thatthe genderof federalwork-
ers is correlatedwith such factors as pay grade and job
type. Whataccountsfor female workers'greaterlikelihood
of receiving unwantedsexual attention?Is it primarilya
function of being a woman in the federal workforce,re-
gardless of other characteristics?Alternatively,women's
elevatedrisk may be largelya by-productof theirtendency
to be located in lower-statusand lower-payingjobs. Fail-
ing to accountfor multiplefactorssimultaneously(thatis,
failing to control for other variables), bivariateanalyses
cannotprovidethe answer.To betterunderstandwhy some
types of workersare more likely to be sexually harassed,
we need to migrateto multivariatemodels.
Dependent Variables
We createda dichotomousvariable,SexuallyHarrassed
(based on question20 of the 1994 MSPB survey),to indi-
cate whetherthe respondenthad received "uninvitedand
unwantedsexual attention"duringthe last 24 monthsfrom
someone where he or she worked in the federal govern-
ment.5This question's treatmentof "uninvited and un-
wanted sexual attention"refers to the range of behaviors
delineatedpreviously.Thirty-threepercentof respondents
reportedsome form of "uninvitedand unwanted sexual
attention"in the previous24 months.
We also created a dichotomousvariable,Accused Ha-
rasser (based on question 40 of the 1994 MSPB survey),
to indicatewhether"anyonein the last 24 monthssaid that
the respondenthad sexually harassedthem."Only 1.2 per-
cent of respondentsindicatedthat someone had accused
them of harassment.
Finally,we createda four-pointordinalvariable,Effec-
tiveness of Sexual HarrassmentTraining(based on ques-
tion 15 of the 1994 MSPB survey) to gauge respondents'
perceptionsof the extent to which they feel theiragency's
sexualharassmenttraininghelpsto reduceorpreventsexual
harassmentin theirorganization.The answersrangedfrom
"no extent"(coded 0) to "a greatextent"(coded 3).
 In our examinationsof influences on the dichotomous Results
variables,we estimatedbinarylogit models. In ourexami-
nationsof influenceson the ordinalvariable,we estimated
Binary Logit Models of Sexual Harassment
orderedlogit models.6 Table 1 presentsthe results of a baseline, multivariate
logit model of sexual harassment(see model L.A).9This
Independent Variables model assesses the relationshipbetween workers'demo-
WorkerCharacteristics. Our specifications take into graphic characteristics(sex, education,age, maritalstatus,
accounta varietyof workercharacteristics:sex, education, pay grade, and job type) and their likelihood of having
age, maritalstatus, pay grade, and job type.7Female is a been harassedin the previous 24 months. The principal
dichotomousvariablecoded 1 for women. For education, finding is thatwomen are much more likely to reporthav-
age, maritalstatus,pay grade, andjob type, respectively, ing been harassedthan are men, controllingfor other fac-
we coded a seriesof dichotomous(dummy)variablesbased tors.'0To illustratethe magnitudeof this effect, the esti-
on the respondentcategoriesin the MSPB survey: mated probabilityof a "typical"female workerreporting
* Education-Less than High School, High School (high that she has received unwanted sexual attentionis 0.51,
school diplomaor GED),High SchoolPlus (high school whereas the estimatedprobabilityof a male counterpartis
diplomaplus some technicaltrainingor apprenticeship), only 0.22.1
SomeCollege, College (B.A., B.S., or some otherbache- Not surprisingly,otherimportantinfluencesareage and
lor's degree), College Plus (some graduateschool), and marital status. Those who are 16-24 years old are most
AdvancedDegree (graduateor professionaldegree) likely to reporthaving been harassed,and those 55 years
* Age- 16-24 YearsOld, 25-34 YearsOld, 35-44 Years of age and over are least likely to reporthaving been ha-
Old, 45-54 YearsOld, and55 Plus YearsOld rassed.Generallyspeaking,youngerworkersreceivemore
* Maritalstatus-Married, Single,Divorced,andWidowed unwantedsexualattention.Otherthingsbeing equal,single
* Pay grade-Pay Grade1-4, Pay Grade5-10, Pay Grade and divorced workers report a higher level of unwanted
11-12, Pay Grade 13-15, andPay Grade 15 Plus sexual attention.The other workercharacteristics(educa-
* Job type-Trainee, Blue Collar (blue-collaror service tion, pay grade, and job type) also demonstratenotewor-
job), Clerical (office, clerical, or technician), Profes- thy influence. It appearsthat,relativeto theirless educated
sional (professional or scientific), Administrative counterparts, those with some college educationandabove
(administrativeor management),and OtherJob. are more likely to reporthaving been harassed.Those at
Agency Location. We createda series of dichotomous higher pay gradesarealso associatedwith an elevatedlike-
variablesto accountfor the agencyin whichthe respondent lihood. Finally,traineesand those in blue-collarpositions
works:Agriculture,Commerce,AirForce,Army,Navy,Other reported higher levels of unwantedsexual attentionthan
DOD (Departmentof Defense), Education, Energy,EPA did those in otherjobs, controllingfor otherfactors.
(Environmental ProtectionAgency),GSA(GeneralServices Most of these findings are consistent with the conven-
Administration), HHS (HealthandHumanServices),HUD tional wisdom aboutsexualharassment.Who is most likely
(Housing and Urban Development),Interior,Justice, La- to receive unwantedand uninvitedsexual attentionin the
bor,NASA(NationalAeronauticsandSpaceAdministration), federalworkplace?A young, single or divorcedwoman in
OPM(Officeof PersonnelManagement),SBA(SmallBusi- a low-statusjob. Despite the increased levels of training
ness Administration),State, Transportation,Treasury,VA and the attentionthat has been focused on sexual harass-
(VeteranAffairs),andOtherAgency. mentin recentdecades,all too predictablepatternsremain.
WorkSituation. We created a series of dichotomous However,thepictureof harassmentin the federalworkforce
variablesto capturesome aspectsof a respondent'simme- revealssome nuancedcausalpatterns.Controllingfor other
diateworkenvironment.Specifically,we createda dichoto- factors, highly educated and well-paid workersare more
mous variablefor male supervisor(Male Supervisor)and likely to relay that they have received "unwantedand un-
a series of dichotomousvariablesto gauge the ratioof men invited"sexual attention.Likely these reportsare a func-
to women in the respondent'sdaily work setting(All Men, tion of heightenedsensitivity to these types of behaviors.
More Men, Equal Men, More Women,andAll Women). Gaugingthe relativeimportanceof the explanatoryfactors
Agency Training.Finally,we createda seriesof dichoto- in model 1.A, differenceof chi-square(log-likelihood ra-
mous variables to gauge which workersreceived sexual tio) tests indicatethat sex is the most importantinfluence
harassmentawarenesstrainingin the respondent'sagency: on the likelihood of receipt of unwantedsexual attention,
SupervisorTraining(only supervisorsand managersre- followedin orderby age, maritalstatus,education,job type,
ceive training),EmployeeTraining(only nonsupervisory and pay grade.
employees receive training),Full Training(all employees Does the level of sexual harassment vary markedly
receive training),andNo Training.8 across agencies? For example, scandals in recent years

Workplace475
intheFederal
SexualHarassment
 Table1 BinaryLogitModelsof Sexual Harassment
Model 1.A Model 11.B Model 1.C Model 11.D
Independent variable Coefficient t-value Coefficient t-value Coefficient t-value Coefficient t-value
Female 1.29** 20.24 1.30** 20.18 1.31 ** 20.28 1.31 ** 17.98
High school .251 .80 .184 .59 .171 .54 .210 .60
High school plus .457 1.48 .381 1.22 .364 1.16 .304 .88
Some college .772** 2.52 .713** 2.32 .695* 2.25 .630* 1.85
College .756** 2.42 .695* 2.21 .683* 2.16 .541 1.56
College plus .889** 2.75 .830** 2.56 .799** 2.45 .709* 1.97
Advanced degree .711* 2.23 .657* 2.05 .666* 2.07 .508 1.43
25-34 years old -.318* -1.81 -.332* -1.87 -.308* -1.74 -.418* -1.99
35-44 years old -.327* -1.86 -.355* -1.99 -.289 -1.64 -.391 * -1.86
45-54 years old -.704** -3.93 -.736** -4.05 -.668** -3.70 -.854** -4.00
55 plus years old -1.31 ** -6.66 -1.35** -6.75 -1.28** -6.45 -1.37** -5.92
Single .405** 5.54 .392** 5.29 .404** 5.48 .458** 5.48
Divorced .511 ** 6.61 .507** 6.50 .517** 6.63 .580** 6.57
Widowed .114 .55 .089 .42 .096 .46 .095 .40
Pay grade 5-10 .120 .94 .178 1.37 .127 .98 .252* 1.65
Paygrade 11-12 .142 .98 .218 1.47 .145 .99 .252 1.48
Pay grade 13-15 .255 1.62 .338* 2.08 .236 1.47 .481** 2.60
Pay grade 15 plus .726* 2.02 .834* 2.31 .684* 1.89 .883* 2.12
Blue collar -.225 -.83 -.252 -.92 -.306 -1.11 -.428 -1.33
Clerical -.685** -2.62 -.695** -2.64 -.690** -2.61 -.906** -2.90
Professional -.832** -3.17 -.859** -3.24 -.869** -3.27 -1.02** -3.28
Administrative -.654** -2.47 -.690** -2.59 -.640** -2.39 -.874** -2.79
Other job -.871** -2.82 -.892** -2.87 -.883** -2.83 -1.06** -2.87
Male supervisor -.140* -2.04
More men .044 .31
Equal men -.170 -1.16
More women -.281* -1.86
All women -.305 -.86
Employee training -.178 -.74
Supervisor training -.286* -2.02
Fulltraining -.417** -3.49
Agriculture -.187 -1.02
Commerce .189 .99
AirForce .136 .75
Army .232 1.29
Navy .171 .88
Other Departmentof Defense .151 .84
Education .075 .38
Energy -.046 -.26
EnvironmentalProtectionAgency .227 1.19
General Services Administration .303 1.64
Healthand HumanServices -.142 -.72
Housing and Urban Development .294 1.51
Interior .298* 1.68
Justice .129 .73
Labor .113 .59
NASA .023 .13
Office of Personnel Management .270 1.36
Small BusinessAdministration -.044 -.23
State .219 1.06
Transportation .161 .88
Treasury .338* 1.93
VeteransAffairs .318* 1.78
(constant) -1.1 9** -2.82 -1.29** -2.95 -.937* -2.10 -.501 -1.00
LRchi-square(d.f.)= 863.4** (23) 882.1 ** (45) 876.9** (28) 730.2** (26)
N1= 6,496 6,464 6,434 5,010
l Significant
atp < 0.01 (1-tailed)l
atp < 0.05 (1-tailed)
~Significant
Note:Forpurposes of estimationandstatistical
comparisoninthevarious
models,thereferencecategories
aremale,lessthanhighschool,16-24yearsold,married,
pay
grade1-4, trainee,
otheragency,femalesupervisor,allmen,andnotraining.l

476 Public
Administration
Review* July/August
2003, Vol.63, No. 4
suggest the possibility that workers (female workers in Binary Logit Model of Accusation as a Sexual
particular)in defense agencies may be located in an espe- Harasser
cially hostile environment.To assess whether the likeli- Table 2 reportsthe results of a baseline logit model of
hood of harassmentvariesfrom agency to agency,we esti- whether a worker has been accused of being a harasser
mated a model that introducesthe series of dichotomous within the previous 24 months.15Again, the independent
variablesfor agency location into the baseline specifica- variablesgauge workers'demographiccharacteristics:sex,
tion outlined previously (see model 1.B). The results do education,age, maritalstatus,pay grade,andjob type. The
notreveala tremendousamountof agency-to-agencyvaria- emergentpicture is largely consistent with received wis-
tion, controlling for other factors. Providing an overall dom. The results indicate that, other things being equal,
evaluativesummary,a differenceof chi-square(log-likeli- men are more likely to have been accused of harassment.
hood ratio)test between a "full"model with agency dum- Anothersignificantfinding is that, relativeto marriedin-
mies (model 1.B) and a "restricted"baseline model with- dividuals,those who are divorcedand widowed are more
out (model 1.A) does not reveal that the set of agency likely to have been accused of harassment.Interestingly,
dummies provides significant explanatoryleverage. Al- single people do not differ from marriedindividuals in
though there is some evidence to suggest that the likeli- terms of theirlikelihood of having been accused, control-
hood of harassmentis slightly lower in such departments ling for otherfactors.Some significantdifferencesalso exist
as AgricultureandHealthandHumanServicesandslightly between workersin differentage categories.For example,
elevatedin such departmentsas the TreasuryandVeterans relativeto those who are 55 years of age and older, those
Affairs, our overarchingconclusion reflects a null find- who areeither 16-24 years old or 35-44 years old are sig-
ing-agency context is not a principaldeterminantof ha- nificantly more likely to have been accused. The sets of
rassmentlikelihood. Perhapsof greatestinterest,there is variablesassociatedwitheducation,pay grade,andjob type
no evidence to supportthe conclusion that defense agen- do not emerge as significantpredictors.16Again, perhaps
cies are outliers.
The third model of table 1 (model 1.C) builds on the Table2 BinaryLogitModelof Accusationas a Sexual
baseline specificationand introducesseveralindicatorsof Harasser
immediateworkenvironment-specifically,a dichotomous
Independentvariable Coefficient t-value
variablefor a male supervisoranda series of dichotomous Female -1.1 8** -4.26
variablesfor the ratio of men to women in the daily work Highschool .529 .49
setting. Takenas a set, these workplacevariablesdo not Highschoolplus -.568 -.51
demonstratedramaticinfluence,otherthingsheld constant. Somecollege -.019 -.02
If anything,workersare a bit less likely to reporthaving College -.294 -.27
Collegeplus .078 .07
been harassedif they have a male supervisor.Althoughthe
Advanceddegree -.070 -.06
coefficients gauging male-to-female ratio are consistent 25-34 yearsold -1.18 -1.45
with the conclusionthatwork settingswith proportionally 35-44 yearsold -.532 -.68
largernumbersof women are associatedwith less harass- 45-54 yearsold -.920 -1.16
ment, these variablesdemonstratemarginalimpact."2 55 plusyearsold -1.44* -1.67
Single -.097 -.27
Finally, table 1 presents the results of a fourth model
Divorced .731** 2.46
(model 1.D) that introduces measures on the nature of Widowed 1.45* 2.31
agencysexualharassmenttraining-specifically, measures Paygrade 5-10 -.390 -.68
of what types of employees, if any, within the agency re- Paygrade 11-12 .444 .72
ceive training.13The coefficientestimatessupportthe con- Paygrade 13-15 .462 .70
clusion that trainingreduces the likelihood that agency Paygrade 15plus 1.23 1.24
Bluecollar .660 .84
workerswill receiveunwantedsexualattention,otherthings .257 .33
Clerical
being equal. As proponentsof trainingwould hope, "full Professional .516 .68
training"(thatis, trainingof both employees and supervi- Administrative .366 .48
sors) appearsto be the most effective option.Forexample, (constant) -3:72** -2.55
relativeto thatof hercounterpartin an agencywithouttrain- LRchi-square(d.f.)= 56.5** (22)
ing, the estimatedprobabilityof a typical female worker N= 6,380
reportingunwantedsexual attentionin an agency with full at p < 0.01 (1-tailed)
**Significant
at p < 0.05 (1-tailed)
*Significant
trainingis 0.10 lower.14 Note: Forpurposes of estimationand statisticalcomparison, the reference categories
are male, less thanhigh school, 16-24 years old, married,pay grade 1-4, and
trainee/otherjob.
Sinceno traineeswereaccusedof beinga harasser,we combinedthoseworkersin
the"otherjob"categorywith"trainees" to formthesuppressedreferencecategory.

SexualHarassment Workplace477
intheFederal
the most importantoverarchingfinding is that very few tively high pay grades, and who are divorcedto perceive
federal workers(only 1.2 percent)reportedthat they had agency trainingas less effective. Although there appears
been accused of sexual harassment. to be some tendency for those most likely to be harassed
(for instance, women and younger and divorcedworkers)
Ordered Logit Models of Perception of to perceive the effects of trainingless positively, this gen-
Effectiveness of Agency Sexual Harassment eralizationdoes not extend to single workers.
Training We also estimateda follow-up model (model 3.B) that
Finally, we provide some initial examinationof influ- introducesan independentvariablethat is a dichotomous
ences on federal workers' perceptions of whether their measure of whether the respondent received unwanted
agency's sexual harassmenttraining"helps to reduce or sexual attentionin the previous24 months.As one would
preventsexual harassment."Table3 presentsthe estimates expect, the harassmentdummyexertsa powerfulinfluence
from an orderedlogit model in which the perceptionof on the perceivedeffectiveness of training.Those who re-
effectiveness of agency training(measuredon an ordinal, port having been harassedratherrecently do not believe
four-pointscale) is regressedon the series of dichotomous that agency training is effective. Controllingfor harass-
variables that gauge worker characteristics (see model ment,the directinfluenceof sex disappears,indicatingthat
3.A).17On average,men andolderworkersperceiveagency the influenceof sex on theperceivedeffectivenessof agency
training as being more effective than do women and trainingis principallyindirect.Womenare more likely to
youngerworkers,otherthings being equal. There appears be harassedand, subsequently,are more likely to perceive
to be some tendencyfor those with moreeducation,at rela- agencytrainingas ineffective.However,once respondents'
recentexperienceswith harassmentaretakeninto account,
Table3 OrderedLogitModelsof Perceptionof women's perceptionsof trainingeffectiveness do not dif-
Effectivenessof Agency Sexual HarassmentTraining fer markedlyfrom those of men. Somewhat surprisingly,
Model3.A model 3.B the previously noted influences of education,pay grade,
IndependentvariableCoefficient t-value Coefficient t-value and age remainintact.For example,regardlessof theirre-
Sexuallyharassed -.748** -12.03 cent personalexperienceswith harassment,older workers
Female -.216** -3.60 -.028 -.45 reportmore favorablyon the effectiveness of training.Fi-
Highschool -.449* -1.66 -.471* -1.69 nally, other things being equal (including the harassment
Highschoolplus -.524* -1.97 -.512* -1.87 indicator),single workersaremostlikely to perceiveagency
Somecollege -.713** -2.72 -.671 ** -2.48
College -.694** -2.58 -.646** -2.33
training in a positive light.
Collegeplus -.918** -3.29 -.881 ** -3.07
Advanceddegree -.757** -2.75 -.745** -2.64
25-34 yearsold .344* 1.89 .211 1.12
Discussion
35-44 yearsold .543** 2.98 .419* 2.23 Our analyses have shed some light on basic questions
45-54 yearsold .946** 5.12 .778** 4.07 pertainingto the environmentof sexual harassmentin the
55 plusyearsold 1.12** 5.67 .881 ** 4.33 federal governmentworkplace.Who is most likely to be
Single .114 1.52 .173* 2.28
Divorced -.128 -1.62
harassed?Workercharacteristicsarethe principalinfluence
-.051 -.64
Widowed -.018 -.08 .066 .30 on the likelihoodthata workerwill receiveunwantedsexual
Paygrade 5-10 -.31 1** -2.39 -.31 1** -2.34 attention.All too predictably,young, single or divorcedfe-
Paygrade 11-12 -.491** -3.40 -.467** -3.16 males in low-statusjobs aremost likely to receivethis type
Paygrade 13-15 -.720** -4.63 -.684** -4.30 of attention.Broadlyspeaking,powerin the workplace(and
Paygrade 15 plus -.546 -1.64 -.442 -1.31 the lackthereof)still matters.Thisprofileatthe federallevel
Bluecollar -.212 -.81 -.290 -1.06
Clerical .033 .13 -.095 -.36
is roughlycomparableto its state-levelcounterpart(Kelly
Professional -.132 -.52 -.283 -1.07 1995). For example,our resultsregardingthe role of mar-
Administrative .078 .31 -.058 -.22 riage in the federal workforceare consistentwith Kelly's
Otherjob -.073 -.24 -.258 -.84 (1995, 200) state-levelconclusions:"Althoughbeing mar-
_cut 1 -3.75 -4.19 rieddid not at all precludethe possibilityof beingharassed,
_cut2 -1.51 -1.92 not being marriedclearly raisedthe probabilityof experi-
_cut3 .754 .406
encingmoreseveresexualharassment. Apparently,marriage
LRchi-square(d.f.)= 217.8** (23) 362.0** (24) still connotesthata woman 'belongsto' someoneelse and,
N= 4,866 4,745
at p < 0.01 (1-tailed)
**Significant
therefore,addssome additionalprotectionfromharassment
*Significant
at p < 0.05 (1-tailed) by the men with whom women work."
Note:Forpurposesof estimation and statistical
comparisonin themodels,the refer- Perhapsa functionof heightenedsensitivityto unwanted
ence categoriesare male, less than high school, 16-e24years old, married,pay
grade 1-4, trainee,and notsexuallyharassed. attention,highly educatedand well-paid workersare also

478 Public
Administration
Review* July/August
2003, Vol.63, No.4
more likely to reportits receipt, other things being equal.
Those responsible for designing training responses to
sexual harassmentneed to be cognizant of the types of
workers most likely to be targets and structuretraining
programsaccordingly.Such restructuringrepresentsa de-
parturefrom traditionaltrainingapproachesthat tend to
follow a relativelystandardscriptthat emphasizes avoid-
ance of litigation.
Contextualfactors demonstratelesser influence. Con-
trollingfor otherfactors,the likelihood of workerharass-
ment does not varymarkedlyfrom agency to agency.Fur-
thermore,the effects of variousaspectsof day-to-daywork
environment(for instance,sex of supervisorand male-to-
female ratioin daily work setting) are not dramatic.How-
ever,we find thatthe type of sexual harassmenttrainingin
the agency exertsa noteworthyinfluenceon the likelihood
of harassment.Our empiricalevidence supportsthe con-
clusion that an inclusive approach(thatis, the trainingof
all workers)is the most effective. That workersat lower
organizationallevels experience more sexual harassment
thando those at supervisoryand managementlevels rein-
forces the point thattrainingneeds to take place at all lev-
els and encompassall types of workers.
Who is most likely to be accused of harassment?A di-
vorced, 40-year-old male fits the profile of the federal
workermost likely to be accused.Obviously,the structure
andmissionof trainingshouldtakethis into account.How-
ever,we suspectthatmost personswho subjectcoworkers
to uninvitedsexual attentionare never accused of sexual
harassmentand thus would not be capturedby the MSPB
measure.The sheer magnitudeof the level of harassment
suggests thata narrowsegmentof the workforceis not the
problem-although some segments, no doubt, contribute
much morethantheirproportionalshare.Forexample,we
doubtthatthe 1 percentor so of workerswho reporthav-
ing been accused of harassmentare responsiblefor all of
the harassment.Again, this conclusion points to the im-
portanceof widespreadtraining.
Unfortunately,despite the federalgovernment'sexten-
sive educationand trainingefforts to eliminate sexual ha-
rassmentin the workplace, sexual harassmentcontinues
to cloud the work experienceof many federalemployees.
What factorsinfluence workers'perceptionsof the effec-
tiveness of agency sexual harassmenttraining?As one
would suspect, there is a markedtendencyfor those who
are most likely to receive unwanted sexual attention to
perceive that trainingprogramsare not effective. For ex-
ample,young female workersare less inclined to perceive
agencytrainingas being effective.Furthermore,those who
reportthey actuallyhave received unwantedsexual atten-
tion in recentmonthsperceivetrainingprogramsas being
less effective.
Both our conclusions and those of the 1995 MSPB re-
port underscorethe need to continue to work on and re-
searchthe problem.Again, the percentageof federalwork-
ersreportingtheyhavebeen sexuallyharassedhasremained
nearly the same in each of the MSPB's studies over the
past 20 years (MSPB 1995). The results of these cross-
sectional surveyspoint to a dilemma in evaluatingsexual
harassmenttrainingover time. Do reportsof increasedlev-
els of sexual harassmentstem from an increase in harass-
ment or from a heightenedsense of awarenessof the prob-
lem and an increasedwillingness on the partof victims to
reportoccurrences(MSPB 1995)? Viewed from the latter
perspective,the fact thatthe rate of sexual harassmentre-
portshas not decreasedcan be perceivedas a sign of train-
ing effectivenessandprogresstowarda less genderedwork-
place environment. Perhaps researchers should view
favorablyan increase in the amount of reportedharass-
ment as an indicatorthatthe workforcehas been educated
as to what constitutes sexual harassmentand is now re-
portingit (MSPB 1995).
This caveat notwithstanding,our focused look at the
most recentcross-sectionof dataraisesconcernsandleads
us to believe that a reevaluationof trainingprogramsis
clearly in order.A one-size-fits-all trainingapproachmay
no longer be tenable, if it ever was. We believe that our
delineation of the relative importanceof the factors that
structurethe likelihood that a federal workerwill receive
unwantedsexual attentionand those factorsthat structure
workers'perceptionof the effectiveness of agency train-
ing shouldbe takeninto accountin such a reconsideration.
In summary,this articlehas attemptedto raiseourunder-
standingof sexual harassmentand the trainingresponse.
Effortsto revise trainingprogramswould benefitfromrec-
ognitionof ourmajorfindings.To review,these include:(1)
singleordivorced,youngfemalesin low-statusjobs arethose
workersmostlikelyto be harassed;(2) however,even among
highly educatedand well-paid workers,women are more
likely thanmen to be harassed;(3) a divorced,40-year-old
male fits the profileof the federalworkermost likely to be
accusedof harassment;(4) men andolderworkersperceive
agency training as more effective than do women and
younger workers;and (5) trainingof both employees and
supervisorsappearsto be the most effective option.
The public sector's investment in training continues
unabated.Although the severity of occurrencesof sexual
harassmentmay have decreased, their numbershows no
signs of diminishing.Whenone takesinto accountthe like-
lihood of underreportingof such occurrences(DuBois et
al. 1999), the pervasivenessof the problembecomes even
more pronounced.This situationcalls for a reassessment
of the efficacy of existing trainingprogramsand for new
and creative ways of thinking about sexual harassment
training.This article provides a step towardmeeting this
challenge.
Workplace479
intheFederal
SexualHarassment
 Acknowledgments
This researchwas fundedin partby a grantfrom the Section tance, KatherineNaff for access to the U.S. Merit Systems Pro-
forWomenin PublicAdministration, AmericanSocietyfor Public tection Board's data file, and Nicholas Lovrich for his helpful
Administrationand is gratefully acknowledged. The authors comments on this article.
would like to thankCathyRineer-Garberfor her researchassis-
Notes
1. The sexual harassmentrates of the 1980 MSPB study are
42 percentfor women and 15 percentfor men (MSPB 1981,
3). The rates of the 1987 MSPB are 42 percentfor women
and 14 percentfor men (MSPB 1988, 2).
2. Between these two extremes, "unwantedsexual attention"
also occurs in any of the following situations(outlined in
question20 of the 1994 MSPB survey):stalking;uninvited
and unwantedpressurefor sexual favors;uninvitedand un-
wantedletters,telephonecalls, or materialsof a sexual na-
ture;anduninvitedandunwantedpressurefor dates(MSPB
1995, 61).
3. This question was not included in the 1980 and 1987 sur-
veys. The distinctionbetween "unwantedsexual attention"
and"sexualharassment"is noteworthy.Indeed,the language
of the threestudiesis somewhatinconsistentandpotentially
confusing to respondents.For example, in attemptingto
capturethe frequencyand type of sexual harassmentin the
workplace,thefollowinglanguageis includedin one or more
of the three survey instruments:"sexualharassment,""un-
invited behaviors and actions," "uninvitedand unwanted
sexual attention,""unwantedsexual attention,"" uninvited
sexual attention,"and "being sexually botheredor sexually
botheringothers."
4. These arethe most recentandcomprehensivedataon sexual
harassmentcollected by the MSPB. Our literaturesearch
did not uncoverany researchderivedfrom these data.
5. Given the natureof some of our later models and tests, we
focus on reportsof "uninvitedand unwantedsexual atten-
tion" in workers'currentagency.
6. Witha dichotomousdependentvariable,a binarylogit model
is preferableto a linear regressionmodel (that is, a linear
probabilitymodel). Linearprobabilitymodels suffer from
heteroscedasticity,inefficientestimates,biased standarder-
rors (and thus incorrecttest statistics), non-normallydis-
tributederrors,and the possibility of nonsensicalprobabil-
ity predictionsthatare negativeor greaterthan one. Binary
logit models avoidthese problems.Althoughmost research-
ers use the linearregressionmodel when they have an ordi-
nal dependentvariable,the orderedlogit model is a prefer-
able alternative.Whereasthe linearregressionmodel makes
the implicitassumptionthatintervalsbetween adjacentcat-
egories of the dependentvariableareequal,the orderedlogit
model-which is designedexplicitlyfor ordinaloutcomes-
does not make this assumptionand avoids the (potential)
bias that is its by-product(McKelvey and Zavoina 1975;
480 Public
Administration
Review* July/August
2003,Vol.63, No.4
Winshipand Mare 1984; Long 1997).
7. Obviously,we could only considerthosecharacteristicsmea-
sured in the survey.For example, there were no questions
on race or ethnicity.
8. These variablesare based on question 11 of the MSPB sur-
vey and reflect the respondent'sreportof agency training.
9. For purposes of our data analysis and write-up of results
fromthe 1994 MSPB survey,we equate"sexualharassment"
with receipt of "uninvitedand unwanted sexual attention
duringthe last 24 monthsfromsomeonewhereyou work(ed)
in the FederalGovernment"(question20 of the 1994 MSPB
survey).
10. We also estimated robust standarderrors for each of the
models. Without exception, the t-values that accompany
these robuststandarderrorsarequite similarto the conven-
tional t-valuesreportedin tables 1-3. Perhapsmost impor-
tant, reliance on robust standarderrorswould not change
any of our substantiveconclusions.
11. This "typical"workeris a married,college-educated40-year-
old with a pay grade of 12 and who is in an administrative
position. For some types of workers, the estimated prob-
ability of "uninvitedand unwanted"sexual attentionis ex-
traordinarilyhigh. For example, this estimatedprobability
is a staggering0.80 for a single, 20-year-old female with
some college educationand who is in a trainingprogramat
a pay gradeof 5.
12. However, a difference of chi-square(log-likelihood ratio)
test reveals that, consideredas a set, the variablesgauging
the ratio of male to female co-workersprovide significant
explanatoryleverage (p < 0.01).
13. The smallerN for this model reflects the sizable numberof
respondentswho "didn'tknow"or were "unsure"for whom
theiragency providestraining.
14. It is worth highlighting that these negative coefficients
emerge despite the likelihood that trainingalso heightens
the awarenessof and sensitivity to unwantedsexual atten-
tion of both employees and supervisors.Furthermore,it is
plausiblethat"full training"effortsmay have been focused
on agencies with a poor recordon this issue. It follows that
these likely are conservativeestimates of the actual effec-
tiveness of employee training.The typical worker in our
illustrationis a 40-year-old,marriedfemale with a college
educationand who is in an administrativeposition at a pay
gradeof 12.
15. Because only 78 of the 6,380 respondents(1.2 percent)re-
portedthey had been accused of being a harasser,this de-
pendentvariableis quite skewed and is associatedwith re-
stricted variance.There are likely some respondentswho
had been accusedof harassmentbut failed to indicateso on
the survey.However,those selected to take partin the sur-
vey were guaranteedstrictconfidentialityandwere encour-
aged to be "frankandhonest"in theirresponses.In termsof
concernsabout(potential)biases in our multivariateresults
due to underreporting, we do not have any indicationor rea-
son to suspect that certaintypes of respondentsamong the
universeof those accusedof harassmentwere moreinclined
References
Baker,Douglas, David Terpstra,and Bob Cutler.1990. Percep-
tions of Sexual Harassment:A Re-examinationof Gender
Differences.Journal of Psychology 124(4): 409-16.
Baker,Douglas, David Terpstra,and Kinley Larntz. 1990. The
Influence of IndividualCharacteristicsand Severity of Ha-
rassing Behavior on Reactions to Sexual Harassment.Sex
Roles 22(5/6): 305-25.
Baldridge,Kathy,andGaryMcLean.1980.How Muchof a Prob-
lem Is It ... Really?Journalof BusinessEducation55(April):
294-97.
Barak,Azy. 1994. A Cognitive-BehavioralEducationalWork-
shop to CombatSexual Harassmentin the Workplace.Jour-
nal of Counselingand Development72(6): 595-602.
Bargh,John,PaulaRaymond,JohnPryor,andFritzStrack.1995.
Attractivenessand the Underling:An AutomaticPower-Sex
AssociationandIts Consequencesfor SexualHarassmentand
Aggression. Journal of Personality and Social Psychology
68(5): 768-81.
Baugh, S. Gayle. 1997. On the Persistence of Sexual Harass-
mentin theWorkplace.Journalof BusinessEthics 16(9):899-
908.
Beauvais, Kathleen. 1986. Workshopsto Combat Sexual Ha-
rassment:A Case Study of ChangingAttitudes.Signs 12(1):
130-45.
Berdahl, Jennifer,Vicki Magley, and Craig Waldo. 1996. The
Sexual Harassmentof Men? Exploring the Concept with
TheoryandData.Psychologyof WomenQuarterly20(4): 527-
47.
Biviano, David. 1997. TrainingEffectiveness of a Sexual Ha-
rassment Awareness and Prevention Program. PhD diss.,
GonzagaUniversity.
Blakely, Gerald,EleanorBlakely, and RobertMoorman. 1998.
The Effects of Trainingon Perceptionsof SexualHarassment
Allegations.JournalofAppliedSocial Psychology28(1): 71-
83.
Cohen, Aaron, and BarbaraGutek. 1985. Dimensions of Per-
ceptions of Social-Sexual Behavior in a Work Setting. Sex
Roles 13(5/6): 317-27.
to underreport.Furthermore,our findings pertainingto the
influenceson the likelihoodof havingbeen accuseddo have
face validity.
16. A difference of chi-square(log-likelihood ratio) test con-
firmsthis conclusion for each of the associatedsets of vari-
ables.
17. Long (1997) providesan accessible introductionto ordered
logit. The markedlysmallerN for these models reflect the
sizable numberof respondentswho "didn'tknow"or were
"unsure"as to whethertraininghelps to reduce or prevent
sexual harassmentin their organization.
DuBois, Cathy,RobertFaley,GaryKustis,and DeborahKnapp.
1999. Perceptions of OrganizationalResponses to Formal
Sexual HarassmentComplaints.Journal of Managerial Is-
sues XI(2): 198-212.
Ely, Robin. 1995. The Power in Demography:Women's Social
Constructionsof GenderIdentityat Work.Academyof Man-
agementJournal 38(3): 589-634.
Fitzgerald,Louise, and SandraShullman.1993. Sexual Harass-
ment:A ResearchAnalysis andAgenda for the 1990s. Jour-
nal of VocationalBehavior 42(1): 5-27.
Fitzgerald,Louise, SandraShullman,Nancy Bailyn, et al. 1988.
The Incidence and Dimensions of Sexual Harassment in
Academia and the Workplace.Journal of VocationalBehav-
ior 32(2): 152-75.
Gilbert,Barbara,MartinHeesacker,and Linda Gannon. 1991.
Changingthe SexualAggression-Supportive Attitudesof Men:
A Psychoeducational Intervention.Journal of Counseling
Psychology 38(2): 197-203.
Guy, MaryEllen. 1993. Women'sIntegrationinto Public Man-
agement.Public AdministrationReview 53(4): 285-92.
Kelly,RitaMae. 1995.OffensiveMen,DefensiveWomen:Sexual
Harassment,Leadership,andManagement.In GenderPower,
Leadership,and Governance,editedby GeorgiaDuerst-Lahti
and Rita Mae Kelly, 195-209. Ann Arbor,MI: Universityof
MichiganPress.
Kelly, Rita Mae, and Phoebe Stambaugh.1992. Sexual Harass-
ment in State Government.In Womenand Men of the States,
edited by MaryEllen Guy, 109-24. New York:M.E. Sharpe.
Licata, Betty, and Paula Poovich. 1987. PreventingSexual Ha-
rassment:A ProactiveApproach.Trainingand Development
Journal 41(5): 34-38.
Lindenberg,Karen,and LauraReese. 1996. Sexual Harassment
Policy: What Do Employees Want?Policy Studies Journal
24(3): 387-403.
Long, J. Scott. 1997. Regression Models for Categorical and
Limited Dependent Variables. Thousand Oaks, CA: Sage
Publications.
Maurizo, Sandra,and Janet Rogers. 1992. Sexual Harassment
andAttitudesin RuralCommunityCareWorkers.Health Val-
ues 16(4): 40-45.
Workplace481
intheFederal
SexualHarassment
McKelvey, Richard,and William Zavoina. 1975. A Statistical U.S. Merit Systems Protection Board (MSPB). 1981. Sexual
Model for theAnalysisof OrdinalLevel DependentVariables. Harassmentin theFederalWorkplace:Is It a Problem?Wash-
Journal of MathematicalSociology 4(1): 103-20. ington, DC: U.S. GovernmentPrintingOffice.
Moorman,Mary, and Wilma Fountroy. 1992. The 1990s Pre- . 1988. Sexual Harassmentin the Federal Government:
scriptionfor a WorkplaceEnvironmentFreeof Sex Discrimi- An Update.Washington,DC: U.S. GovernmentPrintingOf-
nation/Harassmentand the Glass Ceiling. In Sex and Power fice.
Issues in the Workplace:Conference Proceedings. Seattle, . 1995. Sexual Harassment in the Federal Workplace.
WA:NorthwestWomen'sLaw Center. Trends,Progress, ContinuingChallenges. Washington,DC:
Moyer, Robert, and Anjan Nath. 1998. Some Effects of Brief U.S. GovernmentPrintingOffice.
TrainingInterventionson Perceptionsof SexualHarassment. Winship,Christopher,andRobertMare. 1984. RegressionMod-
Journal of AppliedSocial Psychology 28(4): 333-56. els with OrdinalVariables.American Sociological Review
Newman,Meredith.1995. Sexual HarassmentandProductivity: 49(4): 512-25.
It's Not Just a U.S. Problem.Public Productivityand Man-
agementReview 19(2): 172-87.
.1999. A WorkForceunderSiege: A GenderedPerspec-
tive on Women In Military Service. Public Administration
and Management:An InteractiveJournal 4(3). Available at
http://www.pamij.com/. Accessed February24, 2003.
Pryor,John. 1985. The Lay Person's Understandingof Sexual
Harassment.Sex Roles 13(5/6): 273-78.
. 1992. The Social Psychology of Sexual Harassment:
Personand SituationFactorsWhich Give Rise to Sexual Ha-
rassment.In Sex and PowerIssues in the Workplace:Confer-
ence Proceedings. Seattle, WA: Northwest Women's Law
Center.
Pryor,John, and J.D. Day. 1985. Interpretationsof Sexual Ha-
rassment:An AttributionalAnalysis. Sex Roles 18(7/8):405-
417.
Pryor,John,andKathleenMcKinney.1995. Researchon Sexual
Harassment:Lingering Issues and FutureDirections.Basic
and AppliedSocial Psychology 17(4): 605-11.
Pryor,John, and Lynette Stoller. 1994. Sexual Cognition Pro-
cesses in Men High in the Likelihood to Sexually Harass.
Personalityand Social Psychology Bulletin 20(2): 163-69.
Reese, Laura,andKarenLindenberg.1999.ImplementingSexual
HarassmentPolicy. Challengesfor the Public Sector Work-
place. ThousandOaks, CA: Sage Publications.
Riger,Stephanie.1992. GenderDilemmasin SexualHarassment
Policies andProcedures.In Sex and PowerIssues in the Work-
place: Conference Proceedings. Seattle, WA: Northwest
Women'sLaw Center.
Robinson,Mike. 1999. FordAgreesto Pay $7.75 Millionto Settle
Sexual Harassment Case. The Oregonian (Portland, OR),
September8, D2.
Terpstra,David, and Douglas Baker. 1986. Psychological and
DemographicCorrelatesof Perceptionsof SexualHarassment.
Genetic,Social, and GeneralPsychologyMonographs112(4):
459-78.
Thomann,Daniel, D.E. Strickland,and J.L. Gibbons. 1989. An
OrganizationalDevelopmentApproachto PreventingSexual
Harassment.College and UniversityPersonnelAssociation
Fall: 34-43.
U.S. Equal Employment OpportunityCommission (EEOC).
2001. Sexual HarassmentCharges.EEOCand FEPAsCom-
bined: FY 1992-FY 2000. Available at www.eeoc.gov.Ac-
cessed November 12, 2001.

482 Public
Administration
Review* July/August
2003,Vol.63, No.4
Appendix StatisticalSummaryof Variables
Variable Observations Mean Std.Dev.Minimum
Maximum
16-24 yearsold 7,292 .026 0 1
25-34 yearsold 7,292 .201 0 1
35-44 yearsold 7,292 .318 0 1
45-54 yearsold 7,292 .323 0 1
55 plusyearsold 7,292 .132 0 1
Accusedharasser 7,602 .012 0 1
Administrative 7,171 .240 0 1
Advanceddegree 6,886 .156 0 1
Agriculture 7,803 .052 0 1
AirForce 7,803 .050 0 1
All men 7,202 .054 0 1
Allwomen 7,202 .007 0 1
Army 7,803 .050 0 1
Bluecollar 7,171 .128 0 1
Clerical 7,171 .283 0 1
College 6,886 .212 0 1
Collegeplus 6,886 .087 0 1
Commerce 7,803 .038 0 1
Divorced 7,178 .159 0 1
Education 7,803 .033 0 1
Effectiveness
of
training 5,785 1.91 .789 0 3
Employeetraining 6,080 .024 0 1
Energy 7,803 .055 0 1
Environmental
ProtectionAgency 7,803 .036 0 1
Equalmen 7,202 .301 0 1
Female 7,819 .545 0 1
Fulltraining 6,080 .768 0 1
GeneralServices
Administration 7,803 .045 0 1
Healthand Human
Services 7,803 .038 0 1
Highschool 6,886 .096 0 1
Highschoolplus 6,886 .133 0 1
Housingan Urban
Development 7,803 .036 0 1
Interior 7,803 .050 0 1
Justice 7,803 .050 0 1
Labor 7,803 .037 0 1
Lessthanhighschool 6,886 .014 0 1
Malesupervisor 7,136 .739 0 1
Married 7,178 .621 0 1
Moremen 7,202 .391 0 1
Morewomen 7,202 .246 0 1
NASA 7,803 .046 0 1
Navy 7,803 .039 0 1
No training 6,080 .066 0 1
Officeof Personnel
Management 7,803 .032 0 1
Otheragency 7,803 .051 0 1
OtherDepartment
of Defense 7,803 .049 0 1
Otherjob 7,171 .032 0 1
Paygrade 1-4 7,776 .058 0 1
Paygrade 5-10 7,776 .460 0 1
Paygrade 11-12 7,776 .275 0 1
Paygrade 13-15 7,776 .199 0 1
Paygrade 15 plus 7,776 .007 0 1
Professional 7,171 .305 0 1
SmallBusiness
Administration 7,803 .039 0 1
Sexuallyharassed 7,738 .331 0 1
Single 7,178 .200 0 1
Somecollege 6,886 .302 0 1
State 7,803 .029 0 1
Supervisortraining 6,080 .138 0 1
Trainee 7,171 .012 0 1
Transportation 7,803 .045 0 1
Treasury 7,803 .051 0 1
VeteransAffairs 7,803 .048 0 1
Widowed 718 .20 1

Workplace483
intheFederal
SexualHarassment