Sexual Harassment as a Gendered Expression of Power

by Christopher Uggen, Amy Blackstone
Citation
Title:
Sexual Harassment as a Gendered Expression of Power
Author:
Christopher Uggen, Amy Blackstone
Year: 
2004
Publication: 
American Sociological Review
Volume: 
69
Issue: 
1
Start Page: 
64
End Page: 
92
Publisher: 
Language: 
English
URL: 
Select license: 
Select License
DOI: 
PMID: 
ISSN: 
Abstract:

Sexual Harassment as a Gendered Expression of Power

Christopher Uggen Amy Blackstone

University of Minnesota University of Maine

Drawing on recent insights from the study of legal consciousness and gender relations, the authors test the generality of Catharine MacKinnon k theory of the sexual harassment of adult women. Survey and interview data from the Youth Development Study and the General Social Survey are analyzed to identzb a behavioral syndrome of sexual harassment for males and females during adolescence and young adulthood and to compare the syndrome against subjective reports of sexual harassment. A clear harassment syndrome is found for all age and sex groups and MacKinnon 5 predictions about the influence of workplace power and gender relations are generally supported. Financially vulnerable men as well as women are most likely to experience harassing behaviors, and men pursuing more egalitarian gender relationships are most likely to identzfi such behaviors as sexual harassment. Nevertheless, adult women remain the most frequent targets of classic sexual harassment markers, such as unwanted touching

and invasion ofpersonal space.

n her groundbreaking book, Sexual Harassment of Working Women, Catharine MacKinnon notes that men's victimization of women "is sufficiently pervasive in American society as to be nearly invisible" (1 979: 1). Since the publication of her book, sexual harassment has become increasingly visible, and variants of MacKinnon's broad sociocultural explanation

Direct correspondence to Christopher Uggen, Department of Sociology, University of Minnesota, 267 19th Avenue South #909, Minneapolis, MN 55455-04 12 (uggenOO 1@urnn.edu). Versions of this paper were presented at the 2001 annual meetings of the American Sociological Association in Anaheim, California, and the 2002 annual meetings of the Law & Society Association in Vancouver, British Columbia. We thank Jeylan Mortimer for support in several phases of this research and Angie Behrens, Kathy Hull, Justine Jones, Mike Massoglia, Jeremy Staff, Steven Smith, Melissa Thompson, and Sara Wakefield for comments and other help. This research was supported by grants from the National Institute of Child Health and Human Development (HD44138), the National Institute of Mental Health (MH42843) ("Work Experience and Mental Health: A Panel Study of Youth") and the University of Minnesota Life Course Center.

have gained broad acceptance (Schultz 2001; MacKinnon 2002; Tangri, Burt, and Johnson

1982; Welsh 1999). In reaction to evidence that at least some male and adolescent workers are targets of sexual harassment (Kalof et al. 2001; Talbot 2002; Thacker 1996), critics have begun to challenge feminist views of sexual harassment as an act committed by powerful adult males against "powerless females" (Patai 1998: 170) as founded on "unexamined notions of male 'power' and predatoriness" (Patai 199859; see also Francke 1997; Schultz 1998:95). Nevertheless, a systematic examination of the theory's basic propositions about gender and power has yet to emerge in the social science lit- erature. No empirical study of sexual harassment has appeared in the prominent general interest sociology journals American Journal of Sociology, American Sociological Review, or Social Forces (Sever 1996). In light of the impressive body of sociological theory around the phenomenon, a burgeoning research litera- ture in the top specialty journals, and the strong public interest it has generated, this void is sur- prising.

The neglect of sexual harassment in main- stream sociology also forestalls research that could have broad implications. In this paper we elaborate and test MacKinnon's sociocultural theory and address three such questions bear- ing on the generality of sexual harassment. For the sociology of gender, we consider how fem- inist models-originally designed to account for men's violence against women--can explain a diverse range of sexual harassment scenarios. For the sociology of law, we examine how peo- ple define their harassment experiences as sex- ual harassment. For life course studies, we consider the relative prevalence of sexual harass- ment in the adolescent and adult workplace and test whether the same individuals are likely to be targeted at different life course stages.

We first discuss existing theory and research, focusing on legal and sociological definitions of sexual harassment as a syndrome of related behaviors. We then develop a general concep- tual model of harassment experiences and legal consciousness, introducing hypotheses about sex and age differences in experiencing and perceiving harassment. We explain the targeting of men and adolescents by uniting R.W. Connell's notion of a gender system that privi- leges a particular vision of heterosexual mas- culinity with MacKinnon's power-based account of the sexual harassment of adult women. Next, we detail our survey and interview data sources and strategy of analysis. Because sexual harass- ment involves a complex of behaviors, we use latent class analysis to measure and assess group differences based on responses to survey items. We then present statistical results and interview excerpts, focusing on age and gender differ- ences in harassing experiences, their meaning to targets, and their relation to workplace power and gender relations. Finally, we take stock of MacKinnon's model in light of recent develop- ments in theory and the current project.

LEGAL AND SOCIOLOGICAL UNDERSTANDINGS OF SEXUAL

HARASSMENT

U.S. sexual harassment law has been heavily influenced by MacKinnon's (1979) argument that sexual harassment constitutes sex discrim- ination under Title VII of the 1964 Civil Rights Act (see, e.g., Cahill 2001; Saguy 2003). Although we focus on her sociological model rather than her legal arguments, the two are closely intertwined. Since Meritor v. Knson,

SEXUAL HARASSMENT 65

(477 U.S. 57 [1986]), the US. Supreme Court has recognized hostile work environment sex- ual harassment, which occurs when unwelcome sexual advances or a wide range of verbal or physical sexual conduct unreasonably interferes with a person's job or create an intimidating or offensive work atmosphere.' A "severe or per- vasive" legal standard applies to the definition of a hostile work environment, such that harass- ment may be established by a single serious incident or a pattern of less severe, but repeat- ed behaviors. Therefore, measures of sexual harassment must assess the overall pattern of diverse workplace behaviors as well as their severity.

As in other areas of law, interpretations of sexual harassment are shaped by "legal con- sciousness" or the cultural schemas guiding the understanding and use of law (Merry 1990; Ewick and Silbey 1998). Because consciousness of sexual harassment is likely to vary across social groups, a critical issue-for both legal and sociological purposes-is deciding whose per- spective should determine whether sexual harassment has occurred. In Harris v. Forklift Systems, Inc., 510 U.S. 17 (1993), the Supreme Court adopted a dual "objective/subjective" standard that takes into account both the specific "objective" behaviors that a "reasonable person" would find abusive, and the target's "subjective" impressions of the experiences. Some lower courts have held that such impressions of harass- ment depend on the gender of the intended tar- gets, rejecting the reasonable person standard in favor of a "reasonable woman" standard (Ellison

v. Brady, 924 F.2d 872 [9th Cir. 19911). Similarly, some federal appellate courts have recognized the concept of "gender stereotyping" in extending Title VII protections to males. For example, in Doe v. Belleville an adolescent male was physically harassed and threatened with sexual assault by older males because his appearance and behavior "did not conform to his co-workers' view of appropriate masculine behavior" (119 F.3d 563 [7th Cir. 19971).

Apart from legal definitions, social scien- tists have conceptualized and measured sexual harassment in a well-developed scholarly liter-

'The Court also recognizes "quid pro quo" harass- ment in which sexual demands are made a condition of employment or a basis for employment decisions (see Welsh 1999).

66 AMERICAN SOCIOLOGICAL REVIEW

ature. As MacKinnon noted in the 1970s, "lack- ing a term to express it, sexual harassment was literally unspeakable, which made a general- ized, shared, and social definition of it inac- cessible" (1979:27). Though the term is common today, we are still without a generally shared social definition that would help people who are targets of sexual harassment to readi- ly identify such behavior. MacKinnon (1979: 1) defined sexual harassment as "the unwanted imposition of sexual requirements in the context of a relationship of unequal power." Psychologist Louise Fitzgerald and colleagues later developed an influential Sexual Experiences Questionnaire to distinguish gender harassment, unwanted sexual attention, and sexual coercion (Fitzgerald et al. 1988; Gelfand, Fitzgerald, and Drasgow

1995). Others define sexual harassment more broadly as "repetitive, unwelcomed, and inher- ently coercive" acts (Katz et al. 1996:35), as one aspect of social sexual behavior (Gutek, Cohen, and Konrad 1990; Williams, Giuffre, and Dellinger 1999), or as generalized workplace abuse (Keashly 200 1 ;Richman et al. 1999).

Recent sociological research in this area links individuals' ideas about sexual harassment to broader structural relations and cultural sys- tems (e.g., Kalof et al. 2001; Katz et al. 1996; Morgan 1999; Padavic and Orcutt 1997; Rogers and Henson 1997; Rospenda, Richman, and Nawyn 1998). For example, Quinn (2002:389) attributes gender differences in interpreting sex- ual harassment to the acceptance of "normative ideas about women's inscrutability and indi- rectness and men's role as sexual aggressors." Another trend in sexual harassment research has been the attempt to differentiate consensu- al forms of workplace sexuality from sexual harassment (Dellinger and Williams 2002; Quinn 2002; Williams et al. 1999). Schultz (2003) argues that employers are "sanitizing" workplaces in pursuit of organizational effi- ciency, rooting out benign sexual conduct but ignoring sex segregation and inequality.

Despite differences across these literatures, most scholarly definitions of sexual harassment specify conduct that is "unwelcome or unso- licited, is sexual in nature? and is deliberate or repeated" (Barr 1993 :46 1). Although com- monalities exist, some argue that a "lack of conceptual clarity and specificity" continues to plague sexual harassment research (Fitzgerald and Shullman 1993: 19), suggesting the phe- nomenon is best conceptualized as a construct of multiple related behaviors (Gelfand et al. 1995) or as a process rather than an event (Quinn 2002:400). There is also disagreement regard- ing which behaviors constitute harassment (Sever 1999), how gender affects perceptions (Kalof et al. 2001), and whether "subjective" or "objective" behavioral measures are most appro- priate (Welsh 1999; Welsh and Nierobisz 1 997).2

FEMINIST

PERSPECTIVESAND MACKINNON'S SEXUAL HARASSMENT

MODEL

Feminist theories view sexual harassment as the product of a gender system maintained by a dominant, normative form of masculinity. In particular, Connell (1987; 1992; 2002) posits that gender-based inequalities and discrimina- tion are maintained and negotiated through interrelations among differently gendered (and therefore differently privileged) subjects with- in a larger gender system. Connell's constructivist theory introduced the concept of hegemonic masculinity-a gender system that privileges a singular vision of adult heterosex- ual masculinity over all forms of femininity and alternative ma~culinities.~

Connell's theo- ry acknowledges multiple masculinities and femininities (Martin 1998) and takes account of the subjective experience of gender and harass- ment within a larger gender system.

Major themes in Connell's recent theory are compatible with MacKinnon's earlier sociocul- tural conceptualization of the gender system and recent feminist scholarship that emphasizes the performative, relational, and socially con- structed nature of gender (Butler 1990; Kimmel 1994; West and Zimmerman 1987). For MacKinnon, gender and sexuality are similar- ly identified as systems of power and domina-

We should note that behavioral measures are also subjective; they rely upon individuals to perceive and record the behaviors (see Jaschik and Fretz 1991; Kalof et al. 2001).

The concept of hegemonic masculinity has been criticized for its representation of the Gramscian notion of hegemony (Demetriou 2001; Hall 2002; Jefferson 2002; see also Donaldson 1993; Lorber 1998) but lauded for its representation of multiple masculinities and femininities and its utility across research settings (Anderson 2002; Bird 1996; Donovan 1998; Gallagher and Smith 1999; Lee 2000; McGuffey and Rich 1999; Quinn 2002).

tion, with adult men wielding sexual power to assert and maintain dominance over women. MacKinnon thus locates cultural definitions of deviant and conforming sexual behavior in indi- vidual- and societal-level processes of gender socialization (1 979: 154) and in the imposition of power derived from the material economic sphere upon the sexual sphere (1 979:203,174). For MacKinnon, as for Connell, normative con- structions of masculinity disempower those who do not adopt these norms, either because their sex prevents them from doing so (e.g., for bio- logical females) or because they are men who do not adhere to the privileged normative model of heterosexual masculinity.

Therefore, men and women are likely to expe- rience and perceive sexually harassing behav- iors differently because of gender inequality and culturally prescribed expressions of sexu- ality. As Estrich (1987) notes, men and women are held to different standards of sexuality and these standards work to maintain the existing gender order. Women may perceive sexually harassing behaviors as threatening, in part, because they are taught from an early age to be concerned about their bodily safety and to pro- tect their sexuality (Burt and Estep 198 1). Men are taught less about the possibility of sexual predators than women and for good reason-tar- gets of sexual violence are overwhelmingly female and perpetrators are overwhelmingly male (U.S. Department of Justice 2002). Further, sexually harassing behaviors such as "girl watching" are themselves born of the social practices of masculinity (Quinn 2002). Because of these differences, we expect that the under- lying meaning of a sexual joke or a touch is gen- der specific, and men may be unlikely to consider themselves potential targets in a soci- ety that privileges masculinity (Kalof et al. 2001; see also Nelson and Oliver 1998). MacKinnon's attention to gender-based power differentials thus provides some insight into which males may be targets of sexual harass- ment, as well as how they might make sense of these experiences.

Surprisingly few empirical studies have test- ed MacKinnon's most basic proposition that "most sexually harassed people are women7' (1 979: 193). While research consistently shows that many adult women are sexually harassed at the workplace (Fitzgerald and Shullman 1993; MacKinnon 1979; U.S. Merit Systems Protection Board 1988), a handful of studies

SEXUAL HARASSMENT 67

have established lower but nontrivial rates of male-targeted harassment (Kalof et al. 2001; Kohlman 2003; Talbot 2002; Thacker 1996). The U.S. Equal Employment Opportunity Commission (2003) reports that males now file 15 percent of all sexual harassment charges4 and that the number of such charges has dou- bled in the past de~ade.~

Although age clearly is linked to power and gender relations (Connell2000; Thorne 1993), it has rarely been considered in studies of work- place sexual harassment (Gruber 1998). In fact, some have charged that a focus on gender-based rather than age-based, power differentials ignores women's agency and competence as adults (Patai 1998). For these critics, applying the same legal protections to adult females that are normally extended to children smacks of paternalism (Schultz 1998) or even the "infan-

One such claim occurred in the U.S. Supreme Court case establishing same-sex harassment as gen- der discrimination, in which Joseph Oncale testified to numerous sexual humiliations, attacks, and threats of rape by coworkers (Oncale v. Sundowner Offshove Services, 523 U.S. 75 [1998]). MacKinnon herself wrote an amicus curiae brief in support of Oncale (1997), observing that "sexual abuse of men by men is a serious and neglected social problem inextrica- bly connected to sexual abuse of women by men." In this brief, MacKinnon also tied sexual harassment to age-based power relations (Oncale was 21 years old), noting that adult men target those they have power over in society, including children and younger male coworkers. Nevertheless, the sexual harassment of men remains "understudied" (Welsh 1999: 185; Berdahl, Magley, and Waldo 1996), while virtually "no attention" has been directed to the sexual harass- ment of adolescent workers (Fineran 2002:953).

As in landmark cases with female plaintiffs, such as Meritor v. finson, (477 U.S. 57 [1986]; see, e.g., Marshall 1998), the disturbing details of the Oncale case establish the potential severity (though not the generality) of male-male sexual harassment. In addi- tion to frequent verbal harassment, Oncale was sex- ually attacked at work on multiple occasions, including an assault while showering. In Oncale's testimony, he said that one coworker lifted him off the ground while the other "grabs the bar of soap and rubbed it between the cheeks of my ass and tells me, you know, they're fixing to fuck me." After report- ing the incident to a supervisor, Oncale's coworkers confronted him and said, "You told your daddy, huh? Well, it ain't going to do no good because I'm going to fuck you anyway" (MacKinnon 1997: 13).

68 AMERICAN SOCIOLOGICAL REVIEW

tilization of adult women" (Patai 1998:xv, 69, 170). Such observations acknowledge that ado- lescents require special assistance in recogniz- ing or responding to sexual harassment, but there have been few studies of adolescent tar- gets outside the school setting (AAUW 2001; Fineran 2002; Kalof et al. 2001).

This omission is important because age struc- tures power relations in the workplace, and harassers may perceive young workers as vul- nerable or attractive targets (MacKinnon 1979:29).6 Relative to adults, adolescent work- ers are concentrated in a small number of occu- pations and industries, typically in restaurants and other service and retail settings (Mortimer 2003; CHSICL 1998). These jobs may be char- acterized by pleasurable or tolerable sexual behaviors, such as flirting and bantering, as well assexual harassment (Folgero and Fjeldstad 1995; Giufie and Williams 1994). In a study of a fast food restaurant, Reiter describes greater jostling, flirting, and teasing among adoles- cents than adult workers, noting that worker interactions are "clearly marked by age and gender" (1 991 :155). The socialization of ado- lescents into adult work roles thus includes learning the meaning and acceptability of var- ious workplace interactions (Mortimer 2003; Steinberg et al. 1981), including sexual harass- ment (see Schultz 2003). Younger workers may be increasingly aware of sexual harassment as an abstraction, but less experienced in distin- guishing between acceptable and problematic workplace conduct. To track these and other age-graded changes in the structure and mean- ing of sexual harassment, a recent authorita- tive review calls for longitudinal or life course studies (Welsh 1999).7 Our conceptual model

Paradoxically, most adolescents have an impor- tant source of countervailing power: unlike most adults, they need not work to support themselves or their families. In fact, employed adolescents come disproportionately from middle-class families (CHSICL 1998). Although we expect adolescent workers to experience high rates of harassing behav- iors (especially relative to the limited number of hours that they work), these expectations are tempered by adolescents' high rates of job satisfaction (Mortimer 2003:68) and their greater opportunities to exit potentially harassing workplaces.

In addition to gender and age, a line of empiri- cal research has linked sexual harassment with social class, race, and other factors (Cleveland and Kerst

and analysis explores the generality of sexual harassment and offers specific hypotheses about gender and consciousness of harassment over the life course.

CONCEPTUAL MODEL AND HYPOTHESES

Figure 1 shows an integrated conceptual model of power, masculinity, and sexual harassment, based on MacKinnon's theory of sexual harass- ment, Connell's theory of gender relations, and recent work on legal consciousness. Power arrangements-including the privileging of het- eronormative masculinity in the gendered work- place (Acker 1990) and age relations that give adult men rights and power over adolescents (MacKinnon 1997Faffect the extent to which individuals experience particular harassing behaviors. When these behaviors are severe or pervasive and concurrent in time and place, they constitute a syndrome of behavioral sexu- al harassment. The prevalence of the syndrome and even its constituent behaviors is likely to vary by gender and age, with power arrange- ments typically placing adult women at special risk.8

Although all adult women are culturally iden- tified as potential targets of sexual harassment,

1993; Kalof et al. 2001; Marshall 1998; Rospenda et al. 1998; Tangri et al. 1982; Vaux 1993). Researchers have identified particular dimensions of workplace power that place employees at greater risk of sexual harassment, including occupational status (Ragins and Scandura 1995; Richman et al. 1999), supervi- sory authority (Rospenda et al. 1998), and organi- zational factors (DeCoster et al. 1999; Gutek and Cohen 1987; Kohlman 2003; Mueller et al. 2001).

While adolescent females are doubly disadvan- taged by age and gender, adult women typically have less power to leave unpleasant work situations and generally work longer hours than adolescents. In the data to be analyzed, respondents worked far more hours as young adults than as high school students, although 93 percent of the sample worked for pay at some point during high school. Among working ado- lescents between 9th and 12th grade, the median hours worked ver week rose from 8 to 20 hours among boys and from 10 to 20 hours among girls. However, by age 24-25 years, men were working an average of 37 hours per week and women an average of 34 hours per week.

SEXUAL HARASSMENT 69

females >males adolescents >adults

L

females >>males adults >>adolescents

Figure 1. A General Model of Power, Masculinity, and Legal Consciousness in Predicting Sexual Harassment

male targets are likely to be those "perceived not to conform to stereotyped gender roles" (MacKinnon 1997:2; 1979: 178) or practicing "marginalized masculinities" (Connell 1995). Further, although Connell views masculinities as situation-specific "configurations of prac- tice" rather than fixed individual characteristics (1 995:8 I), males who consistently practice mar- ginalized masculinities are likely to be consis- tent targets of sexual harassment throughout the life course.

Of course, individuals do not automatically translate their experiences with harassing behav- iors into a global account of those experiences as sexual harassment. Theories of legal con- sciousness suggest that both culturally avail- able schemas and individual resources are

in the process labelingthe of behaviors as sexual harassment (Ewick and Silbe~ 1998:53). Although men in less power- ful positions may be targets of sexually harass- ing behaviors at the workplace, they are unlikely to interpret these behaviors as sexual harassment because they generally lack a cultural reference point that would give meaning to them as a uni- fied construct or phenomenon. This is the case for the "core markers" culturally associated with the sexual harassment of adult women, such as sexual touching and invasion of personal space, as well as more ambiguous behaviors, such as physical assault or offensive jokes. Because heteronormative masculinity encour- ages men to conceive of themselves as preda- tors or protectors rather than targets (or "victims7') of such harassing behaviors, men who experience the behavioral syndrome are less likely to identify it as sexual harassment than women who experience the behavioral syn- drome. Nevertheless, consciousness of sexual harassment is tied to consciousness of gender relations, such that men with more egalitarian attitudes and behaviors in gender relationships are most likely to recognize these experiences as sexual harassment.

We draw six hypotheses from this model, the first taken from MacKinnon's basic proposi- tion that ''most sexually harassed people are womenn (1 979: 193; 1987: 107).

Hypothesis I, Gender Dzference in Harassing Behaviors: More females than males will experience specific sexually harassing behaviors.

The second hypothesis specifies that a behav- ioral syndrome of sexual harassment will be observable for males and adolescents as well as

70 AMERICAN SOCIOLOGICAL REVIEW

adult women, but that its form will vary for adult females, adolescent females, adult males, and adolescent males.

Hypothesis 2, Syndromal Clustering: Sexually harassing behaviors will cluster in a syn- drome or construct of behaviors for all groups, but its structure will differ by age and sex.

The next two hypotheses address legal con- sciousness and the subjective interpretation of harassing behaviors. While sexual harassment exists within the cultural repertoire of events that might occur at the workplace among women, men are less likely to consider themselves poten- tial targets and therefore are less likely to per- ceive the behavioral syndrome as sexual harassment.

Hypothesis 3, Gender Difference in Subjective Appraisal: More females than males will perceive that they have been sexually harassed, as measured by a "subjective" self-appraisal.

Hypothesis 4, Gender Diference in Association between Behavioral and Subjective Harassment: General "subjective" percep- tions of sexual harassment will be more closely correlated with the behavioral sex- ual harassment syndrome for females than for males.

We similarly predict less continuity between adolescent and adult harassment experiences for females because all adult women are cul- turally identified as potential targets, whereas particular males will be consistently targeted based on their expressions of masculinity.

Hypothesis 5, Gender Diference in Life-Course Continuity:The correlation between behav- ioral sexual harassment in adolescence and adulthood will be lower for females than for males.

Our final hypotheses address power and mas- culinity, the predicted mechanisms linking age and sex to sexual harassment experiences.

Hypothesis 6a, Workplace Power: Men and women holding less workplace power are more likely to be targeted than men and women holding greater workplace power.

Hypothesis 6b, Gender Relationships: Adult men in more egalitarian gender relation- ships are more likely to perceive sexual harassment than adult men in less egali- tarian gender relationships.

As summarized in Figure 1 and the hypothe- ses, we suggest that gendered power relations result in females being more frequent harass- ment targets than males, and adolescents to be targeted at high rates relative to the number of hours they work. Because dominant cultural understandings of sexual harassment identify adult women as the most likely targets, howev- er, the adult men and adolescent males and females will be less likely than adult women to interpret their experiences as sexual harass- ment. Although our individual-level data can- not provide a critical test of the macro-level paths outlined in the figure, our analysis will offer evidence bearing on each of the hypo- theses.

DATA, MEASURES,
AND STRATEGY OF ANALYSIS

We adopt a quantitative survey approach, guid- ed by in-depth interviews undertaken with a subset of 33 survey respondents. Although methodological choices are contested in sexu- al harassment research as elsewhere (Arvey and Cavanaugh 1995; Gillespie and Leffler 1987; Smith 1994), our design is informed by an emerging measurement literature in the area: fol- lowing Fitzgerald and Shullman (1993), we inquire about a range of potentially harassing behaviors, posing more subjective questions about sexual harassment separately from ques- tions about specific behaviors; following Gelfand et al. (1995: 174), we conceptualize sexual harassment as "a construct, with multi- variate responses that are related" rather than as a simple event; and, following Welsh, we test harassment measures for multidimensionality (2000) and link survey data with intensive inter- views (1999).

DATAAND MEASURES

We analyze data from the Youth Development Study (YDS), a prospective longitudinal inves- tigation that began in 1988 with a random sam- ple of 1,010 ninth graders in the St. Paul, Minnesota public school district. The annual survey was administered in school from 1988 until 1991, with mail questionnaires used from 1992 until 2000. To our knowledge, no other data set contains a set of behavioral sexual harassment items for males and females in ado- lescence and adulthood. We placed these items on the 1 1 th survey wave in 1999, when respon- dents were 25-26 years old, and obtained data from 742 of the original 1,O 10 respondents

(73.5 per~ent).~

We measure sexual harassment with six dichotomous behavioral indicators, ranging from sexual content in the workplace (such as offensive remarks about the respondent) to physical assault. We began with a set of behavioral indicators because of their demon- strated reliability and validity (Welsh and Nierobisz 1997), using the high school period as a "contextual cue" (Horney and Marshall 1991) to help orient respondents. We asked whether they had experienced each harassing behavior while worlung in jobs since high school (the young adult period) and in jobs held during their high school years (the adolescent period).

Research on sexual victimization suggests that providing such checklists of specific behav- iors helps elicit accurate self-reports of early, averse sexual experiences (Miller, Johnson, and Johnson 1991). Because dichotomous indicators have generally proven most reliable in such ret- rospective accounts, all of our behavioral meas- ures are dichotomous. Although there is some potential for recall problems with the high school items, researchers in other contexts report impressive stability in self-reports of delin- quency and victimization for periods of up to 8 years (Hindelang, Hirschi, and Weis 198 1 :SO).

Our measures were written to reflect the breadth of harassment behaviors included in the Inventory of Sexual Harassment (Gruber 1992) and Sexual Experiences Questionnaire (SEQ) (Fitzgerald et al. 1988; Gelfand et al. 1995). We follow the SEQ in asking first about specific harassing behaviors (offensive jokes, remarks or gossip, intrusive questions, inva- sion of personal space, unwanted touching, offensive pictures or other materials, and phys- ical assault). Only then do we invoke the term "sexual harassment" using a more subjective global item (would you consider these experi-

The sample well represents the St. Paul commu- nity (Finch et al. 1991; Mortimer 2003). About 74 percent of the panel is white, 10 percent African- American, 5 percent Hispanic, and 4 percent Asian. For details onYDS sampling and panel retention, see Mortimer (2003:29-43).

SEXUAL HARASSMENT p

ences sexual harassment?). This allows us to examine hypotheses about gender differences in legal consciousness as well as sexually harass- ing behaviors.

Because we sampled from populations whose sexual harassment experiences have rarely been studied (e.g., males and adolescents), our YDS indicators differ from those on the SEQ and other instruments developed for adult female respondents. Most importantly, we avoided age- and gender-specific language. For example, the SEQ includes "leering" and "attempts to stroke or fondle" which may be less recognized by males, and "repeated requests for drinks or din- ner," which may be less relevant to adolescents. Instead, we asked about general conduct, such as "unwanted touching" and "invasion of per- sonal space,'' to develop inclusive measures that would not immediately exclude ambiguous behavior (such as a supervisor putting an arm around a subordinate) that may or may not indi- cate sexual harassment.

This approach may overstate the prevalence of sexual harassment by including sexual behav- iors that are not harassing (such as some types of sexual banter), as well as harassing behaviors that are not sexual (such as physical assault). We address this potential danger in four ways: (1) by examining the interrelation between more ambiguous and less ambiguous "core" items; (2) by considering the interrelation between all behavioral items and the global self-assessment of whether the behaviors constituted sexual harassment; (3) by examining models that exclude less serious items; and, (4) by con- ducting intensive interviews with a subset of sur- vey respondents, discussing the nature and context of their experiences with sexual harass- ment.

We selected YDS participants for intensive interviews based on their survey responses. We sent letters to 98 males and 86 females who had reported experiencing some form of harass- ing behavior at work, inviting them to discuss their experiences in a 60- to 90-minute inter- view. A total of 28 men and 30 women expressed interest in participating and we completed inter- views with 14 men and 19 women, who were each paid $40. The interviews took place at a location of the participant's choosing and were taped for later transcription. Participants were asked to describe their experiences in their own words and were not provided specific response categories. Our goal was to learn more about the

72 AMERICAN SOCIOLOGICALREVIEW

context of potentially harassing experiences, participants' own understandings of specific workplace interactions, and their ideas about sexual harassment more generally (GiuMe and Williams 1994; Stambaugh 1997). In choosing particular excerpts for inclusion in this paper, we looked for patterns across the interviews and selected those quotes that we thought best represented these patterns.

Following the analysis of the indicators and interviews described above, we examine the association between workplace power, gender relations, and sexual harassment using survey data from the YDS as well as the General Social Survey (GSS) (Davis, Smith, and Marsden 2003). To address concerns that the concept of power is poorly articulated or unmeasured in sexual harassment research (Patai 1998), we assess it directly as financial insecurity and supervisory authority. To meas- ure respondents' beliefs in and adherence to normative gender relations expectations, we consider their career expectations for them- selves and their partners (Morgan 1999) and behavioral indicators of these relationships, such as the share of housework they provide. Appendix 1 describes each of the measures used in this portion of the analysis.

We will first present descriptive statistics and simple t-tests to compare the rate of specific harassing behaviors across age and sex groups in ourYDS survey data. Analyzing these items individually could obscure important infor- mation about their covariation, yet combining them in a summative scale could conceal important group differences in the occurrence of particular behaviors. Restricting the analy- sis to one or two core items is also problem- atic, as sexual harassment is defined in part by its pervasiveness. To overcome these difficul- ties, we use a latent class approach (Dayton 1998; Lazarsfeld and Henry 1968; McCutcheon 1987) to examine these behaviors in combination and to test for the presence of a common syndrome of sexual harassment. Because there are several known indicators but no criterion "gold standard" that estab- lishes sexual harassment with certainty, latent class techniques are especially appropriate in this context. These methods test whether the covariation between each of the behavioral harassment items is due to their mutual rela- tionship to an unobserved or latent sexual harassment construct. If so, then specification of the latent sexual harassment variable should reduce this covariation among individual sur- vey items to the level of chance variation (McCutcheon 1987:5-6). Latent class analysis also allows us to establish whether there are distinct types of sexual harassment within age and sex groups and to impose equality con- straints to test whether a latent sexual harass- ment syndrome varies across groups.1° All latent class models are estimated using the CDAS-MLLSA program (Eliason 1997).

For our purposes, the greatest advantage of latent class analysis is that it helps reduce a complex set of response patterns among numerous intercorrelated nominal indicators to a rigorous but tractable set of ideal types with- out imposing a set of a priori assumptions about what counts as sexual harassment. This permits us to test hypotheses bearing on fun- damental substantive questions about the gen- erality of sexual harassment: (1) whether we can observe a syndrome of sexual harassment among men and adolescents; (2) whether this syndrome is the same as or different than the one MacKinnon identified among adult women; (3) how subjective perceptions of harassment are related to this behavioral syn- drome for different groups; and (4) whether the behavioral syndrome is related to workplace power and gender relations in the manner sug- gested by theories of sexual harassment. In addressing these questions, we also report illustrative examples of the nature and context of sexual harassment based on our intensive interviews with a subset ofYDS respondents.

RESULTS

PREVALENCE HARQSSMENTOF SEXUAL

Table 1 shows descriptive statistics and t-tests for each sexual harassment item. The social distribution of harassing behaviors varies with

lo For example, adult females who report offensive materials may also report unwanted touching, but teenage males may report offensive materials in iso- lation. If so, such materials may be less indicative of sexual harassment for adolescent males than for adult females.

SEXUAL HARASSMENT 73

Table 1. Percent Reporting Harassing Behaviors at Two Life Stages

At any job you have held during high school, have you experienced. . . Offensive jokes, remarks, or gossip directed at you?** Direct questioning about your private life? Staring or invasion of your personal space?+ Unwanted touching?* Pictures, posters, or other materials you found offensive?+ Physical assault by a co-worker, boss, or s~pervisor?~

Percent experiencing any of six behaviors
Percent experiencing touching or space (core indicators)*
At any job you have held since high school, have you experienced . . .

Offensive jokes, remarks, or gossip directed at you?
Direct questioning about your private life?
Staring or invasion ofyour personal space?**

Unwanted touching?**
Pictures, posters, or other materials you found offensive?
Physical assault by a co-worker, boss, or supervisor?

Percent experiencing any of six behaviors Percent experiencing touching or space (core indicatorsa)** Global Indicator Would you consider any of the above experiences during or since high school sexual harassment?**

Female Male t-value

22% 3 1% -2.814

22% 25% -.747

4% 7% -1.819

17% 11% 2.363 35% 3 7% -.504

Note: Sample sizes from 423 to 425 for females and from 3 14 to 318 for males. a Core indicators include unwanted touching and invasion of personal space.

'p < .lo; *p < .05; **p < .O1

the type (and perhaps severity) of each indi- cator. During high school, fewer females than males report offensive jokes but more females report invasion of personal space and unwant- ed touching than males. As adults, females face the highest rates of these core markers of sexual harassment-almost one third report unwanted touching or invasion of personal space. Contrary to the adolescent period, adult women also experience offensive jokes, remarks, or gossip about them at rates com- parable to adult men. Clearly, many adolescent workers experience these harassment behav- iors: 33 percent of females and 38 percent of males report at least 1 behavior, and 17 percent of females and 11 percent of males were sub- ject to at least 1 core marker. Although each individual behavioral item may tap some degree of non-sexual or non-harassing behav- ior (see, e.g., Keashly 2001; Richman et al. 1999), 33 percent of females and 14 percent of males reported that they considered their expe- riences with these behaviors to constitute sex- ual harassment.

Table 1 shows only partial support for our first hypothesis, that more females than males experience specific sexually harassing behav- iors. Consistent with expectations, females are more likely to face unwanted touching and violations of personal space, and we find a greater gender gap among these core markers for adults relative to adolescents. Yet males report similar rates of exposure to the other items. Adolescent males are somewhat more likely to report offensive materials and phys- ical assault than adolescent females. These differences are unlikely to be explained by gender differences in labor force participa- tion, because participation rates are similar

To summarize and test the basic behavioral pat- terns shown in Table 1, we also pooled the data and conducted a logistic regression analysis on the 6 harassment items, using age, sex, and their interac- tion as independent variables (table available from authors). In additive models, age is a positive pre- dictor of each behavior, and female gender has pos- itive effects on the two core items but negative effects on offensive jokes and physical assault. In interactive models, the product term only approaches statistical significance for the offensive jokes, remarks, and gossip item (p< .I).

74 AMERICAN SOCIOLOGICALREVIEW

for adolescent males and females (U.S. Census Bureau 2000:403). Nevertheless, results for the global indicator support our third hypoth- esis, showing that females are far more likely than males to report that they considered their experiences to be sexual harassment.12

l2 Although it is possible that some respondents could have read the global item as soliciting their opinions about the behaviors referenced rather than their own experiences, this does not appear to be the case. The global item followed immediately after the "have you experienced" items and immediately pre- ceded a question asking whether they had personal- ly consulted an attorney about these experiences. The rate of affirmative responses to the global item is also in line with estimates from other studies (surn- marized in Welsh 1999). Therefore, we believe that respondents were referencing their own experiences in answering this item.

Table 2. Fit Statistics for Latent Class Models

Statistic One Class
Adolescent Females L2 d f 359.31
  P BIC ID  
Adult Females L2 d f 305.12
  P BIC ID  
Adolescent Males L2 d f 309.59
  P BIC ID  
Adult Males L2 d f 190.66
  P BIC ID  

Note: L2: likelihood ratio chi-square statistic df: degrees of freedom BIC: Bayesian Information Criterion statistic ID: index of dissimilarity

Our second hypothesis predicts that the six individual indicators will be clustered as a syndrome of behavioral sexual harassment within each of the four sample groups (males and females at each life course stage). The first step in this analysis is to determine the number of latent classes needed to character- ize the harassment indicators. If a 2-class model accounts for the covariation among the behaviors, for example, this may provide evi- dence of a coherent syndrome. If 3-or 4-class models are needed, this may suggest multiple types of sexual harassment (see Fitzgerald et al. 1988; Gelfand et al. 1995). Summary sta- tistics for models specifying from 1to 5 latent classes are reported in Table 2: L2 is the like- lihood ratio chi-square test statistic relating

Two Class Three Class Four Class

44.13 40.63 23.41 5 1 48 43 ,741 ,766 ,994

-264.17 -249.53 -236.53 ,048 ,044 ,026

61.84 34.69 22.97 5 1 44 4 1 ,142 ,841 ,990

-245.85 -230.77 -224.38 ,094 .05 1 ,035

49.06 26.48 15.05 5 1 47 45 ,551 ,993 1.0

-243.99 -243.59 -243.53 ,064 ,044 ,033

54.38 29.03 20.08 50 45 39 .3 1 1 ,969 ,995

-233.41 -229.98 -204.39 ,084 ,047 ,036

the observed data to the latent model, the Bayesian information criterion (BIC) statistic helps identify the best fitting model when mul- tiple models provide an adequate fit based on the L2 criterion, and the index of dissimilari- ty (ID) between observed and estimated expected frequencies is the percentage of cases incorrectly classified by each model (see Dayton 1998; McCutcheon 1987).

The L2 values in Table 2 show the greatest improvement in fit when comparing 1 -class or independence models (which assume no asso- ciation among the indicators) to 2-class mod- els. For adolescent females, for example, the squared likelihood decreases from 359 to 44 when a second class is added, but when a third class is added, the squared likelihood decreas- es from 44 to 4 1. In general, a P-value greater than .10 and smaller BIC values indicate a more acceptable fit, so these statistics also favor a 2-class model: for each group, the 2- class model shows a P-value within the limits of chance variation and the smallest BIC value relative to 3,4, or 5-class models. As the num- ber of classes increases, the index of dissimi- larity shows that slightly fewer cases are misallocated. Because these gains are mar-

SEXUAL HARASSMENT 75

ginal and the other statistics point to a 2-class solution, we accept the 2-class models as pro- viding the best and most parsimonious fit to the data for all groups.

Although two classes describe sexual harass- ment within each group, the syndrome itself may differ by age and sex. In fact, when males and females are combined into a single sam- ple, we find that three adult classes and four adolescent classes are needed to characterize the data (tables available from authors). The nature of the behavioral syndrome for each group is more readily apparent from the latent class probabilities and conditional probabili- ties reported in Table 3. Latent class probabil- ities show the relative size of each class. Conditional probabilities show the likelihood of experiencing particular behaviors for indi- viduals within each class. The group-specific probabilities for adolescents and adults are shown in Table 3. The final conditional prob- abilities offer a clear interpretation. For each ageisex group the 64 possible response patterns among the 6 dichotomous indicators are clear- ly summarized by two classes that can be described as "high" and "low" sexual harass- ment, offering a parsimonious rendering of

Table 3. Estimates from Latent Class Models for Adolescence (N = 735) and adulthood (N = 733)

Male     Female
Low   High Low High
SH   SH SH SH

Adolescence Offensive jokes about you Questions about private life Invasion of personal space Unwanted touching Offensive materials Physical assault

Latent Class Probabilities N Conditional Latent Class Probabilities

Adulthood Offensive jokes about you Questions about private life Invasion of personal space Unwanted touching Offensive materials Physical assault

Latent Class Probabilities N Conditional Latent Class Probabilities

76 AMERICAN SOCIOLOGICAL REVIEW

the observed data.13 Those in the high harass- ment classes have higher probabilities of expe- riencing every behavior. Although it may be defensible to characterize these classes as mark- ing the presence or absence of sexual harass- ment, we label them "high" and "low" rather than "harassed" and "non-harassed" because many of those in the low classes have a non-zero probability of exposure to many of the behav- iors.

Among the adolescent females in Table 3,107 respondents (about 25 percent) are categorized within the high harassment class. In adulthood, this number increases to 13 1 women (about 3 1 percent). Female respondents in the high harass- ment classes in adolescence and in adulthood have a high probability (ranging from .50 to .81) of exposure to the first three behaviors, including offensive jokes, intrusive questions, and invasion of personal space. In addition, the severity of harassment seems to increase over time for females. As they enter adulthood, those in the high harassment classes have a much greater probability of experiencing core sexu- al harassment markers, such as invasion of per- sonal space, where the probability increases from .50 to .75, and unwanted touching, where the probability increases from .20 to .35. Although the overall number in the high harass- ment class increases, the probabilities of expe- riencing the other behaviors remain relatively stable from adolescence to adulthood.

Among the adolescent males in our survey data, 71 respondents (about 23 percent) are in the high harassment class, relative to 97 adult males (about 31 percent). Similar to females, males in the high harassment classes have a high probability of experiencing offensive jokes, intrusive questions, and invasion of personal space. Unlike females, however, the likelihood of experiencing the classic markers-invasion of personal space and unwanted touching- does not increase in the high harassment class as males enter adulthood. Yet, for males and females in the low harassment classes, the prob-

l3A combination of affirmative responses to three indicators cokes, remarks, and gossip about the respondent; intrusive questions; and invasion of per- sonal space) was the most common or second most common response pattern assigned to the high harassment class for all four groups.

ability of exposure to most behaviors increas- es in adulthood.

It is noteworthy that adults in the low harass- ment classes have a non-trivial probability of experiencing offensive jokes and unwanted questions about their private lives (.15 and .23 for females and .17 and .19 for males). While these behaviors may indeed be unwelcome, they often prove insufficient to legally establish a hostile work environment sexual harassment claim (Vento 2001). Nevertheless, offensive jokes or intrusive questions are often taken as evidence of hostile work environment sexual harassment when they occur among a "pletho- ra of offensive incidents" (Hall v. Gus Construction, 842 F.2d 101 5 [8th Cir. 19881). We found a similar pattern in our interviews: several participants described work settings in which sexual joking and intrusive questions were the norm but they did not consider these behaviors to constitute sexual harassment. Pam (names and other identifiers have been changed), who worked as a waitress during high school and afterward, reported offensive jokes, intrusive questions, and invasion of personal space on the job and was categorized in our high harassment latent class. She told us the fol- lowing:

It was just kind of accepted. There, people felt free to pretty much say whatever they wanted . . . I worked with a host, a male host, who was a lit- tle bit older and I got a lot of questions, he would ask me a lot of questions. And looking back at it, it wasn't appropriate.

When surveyed, Liz, a middle-class white woman, reported no harassing behaviors during her adolescent period. When interviewed, how- ever, she discussed working as a lifeguard in a sexually charged environment during this time:

There was tons of, lots of, lots of sexual talk throughout everything. But everybody enjoyed it and joked about it. I don't think anybody was offended, although it probably could have been offensive.

Though neither Pam nor Liz considered these experiences to be sexual harassment at the time, both suggest that, in retrospect, their experi- ences may have been problematic. This is con- sistent with research on the adolescent workplace as a setting for learning about adult work roles (Mortimer 2003). As Pam and Liz have gained age and experience, they have per- haps developed a more nuanced sense for dis- tinguishing problematic workplace conduct. While it is possible that Pam and Liz would still not define these particular experiences as sexual harassment, even if they happened today, the responses of both women indicate an increased awareness about potentially prob- lematic workplace conduct that seems to have come with age.

We should also note that males who experi- ence offensive joking may in fact interpret these jokes as a form of male bonding, or "doing masculinity," rather than harassment (Connell 1995; Quinn 2002). Nevertheless, our latent class results and interview data suggest that not all men experience offensive joking as an enjoy- able form of male bonding. Rick, who worked in a printing warehouse, reported sexual harass- ment on the survey and was classified in our high harassment latent class. He explained that male coworkers regularly joked in a way that made him uncomfortable: "There were lots of really awful jokes-gay jokes, sex jokes." Rick said his coworkers knew he was disturbed by the joking: "Sometimes they would just do it just to bug me." Rick said, "They wouldn't quit. They'd tell jokes just 'cause they knew it irritated the crap out of me." Rick said he would have protested more about the joking but, "they has- sled me enough as it was and I think if I'd said something it would have been even worse. So I think the reason I didn't really ever say anything was 'cause I just knew they'd even lay into it even more." Rick handled the offensive joking by listening to music all day with headphones. In Rick's words, he "literally tuned out." Rick's strategy of tuning out to avoid offensive inter- actions with coworkers suggests that the joking was not a bonding experience for him. Eventually, Rick quit his job out of concern that he "was gonna get beat up" by the cowork- ers who participated in the offensive joking.

Although the harassment did not become physical in Rick's case, other men reported more physical workplace conduct. Jerry is a white male grouped into our latent behavioral class after his survey responses indicated offen- sive jokes, intrusive questions, and invasion of personal space. He did not report that he con- sidered these behaviors sexual harassment. Nevertheless, he described his shock at being groped by a male coworker while working in a correctional facility:

[He] grabbed my butt. Like, not like a-like grabbed my-And I-. . .I freaked. I'm like, "You

SEXUAL HARASSMENT 7

don't-you know, you don't do that." You just don Y [do] that. One, we're at, this is the workplace. And two, you don't, you don't know me. You don't know anything about me. You don't. . . You don't do that. Well, I freaked out. And, you know, and like, my friends, like, God they were on me. . ..

When asked whether he would call the expe- rience sexual harassment, Jerry told us this:

I would say yes, because it wasn't like I was, I wanted that grab. I wasn't, like, advertising. I wasn't. ..I don't know. I guess my opinion is that the workplace should be the workplace. It shouldn't be . . . all the grab-assin'.

Like Rick, Jerry makes it clear that his expe- rience could not be described as consensual male bonding. Instead, the incident called Jerry's masculinity into question. Other male cowork- ers taunted him by saying (in a sing-song voice), "He likes you!" This response by Jeny's cowork- ers suggests that hegemonic masculinity is priv- ileged at the correctional facility. For Jerry, being the subject of an unwanted grab was just the first step in his harassment experience. As a result of the grab, his own adherence to the privileged model of masculinity was questioned.

?%S~NG SEX AND AGEDIFFERENCES

IN THE BEHAVIORAL SYNDROME

SEXUALHARASSMENT

Taken as a whole, Table 3 and our interviews suggest potentially important gender differ- ences as well as similarities in the behavioral sexual harassment syndrome. We next consid- er whether these differences are statistically or practically significant. We fit a series of simul- taneous latent class models to learn exactly how the behavioral sexual harassment syndrome varies by sex and age, following McCutcheon (1987). After first imposing a single equality constraint on each item individually (tables available from authors), we located group dif- ferences by successively adding constraints and examining the fit of the models.

We first compare males with females in ado- lescence and adulthood in Tables 4a and 4b. For the adolescents, all items differ signifi- cantly by sex except intrusive questions and exposure to offensive materials. Adult men and women have similar harassment experiences, however, with the important exception of two core indicators: invasion of personal space and unwanted touching. Women have greater expo- sure than men to these classic markers, even

78 AMERICAN SOCIOLOGICALREVIEW Table 4. Nested Latent Class Models Testing Sex and Age Equality in Sexual Harassment

a. Male/female dzfferences in adolescence

Unrestricted, Heterogeneous Model

Partial Homogeneity Models Questions about private life Offensive materials Invasion of personal space Physical assault Unwanted touching Offensive jokes about you

Restricted, Complete Homogeneity Model

b. Male/female dzfferences in adulthood

Unrestricted, Heterogeneous Model

Partial Homogeneity Models Offensive jokes about you Questions about private life Offensive materials Physical assault Invasion of personal space Unwanted touching

Restricted, Complete Homogeneity Model

c. Adolescent/adult dzfferences for females

Unrestricted, Heterogeneous Model

Partial Homogeneity Models Physical assault Unwanted touching Offensive materials Invasion of personal space Offensive jokes about you Questions about private life

Restricted, Complete Homogeneity Model

d. Adolescent/adult dzfferences for males

Unrestricted Heterogeneous Model

Partial Homogeneity Models Physical assault Offensive jokes about you Offensive materials Unwanted touching Invasion of personal space Questions about private life

Restricted, Complete Homogeneity Model

*p< .05; **p< .01; ***p< ,001

after they have already been categorized into high and low harassment classes. These results establish that the behavioral sexual harassment syndrome differs by sex and identify the behav- iors that distinguish the sexes. The final con- straint in the restricted complete homogeneity model provides a test of whether males and females are equally distributed across the two

d f L2 L2 d f Difference Difference

classes. Contrary to our expectations, we find no gender difference in the probability of assign- ment to the high harassment class. A similar pro- portion of males and females thus experience unwanted sexual content at work, but its char- acter differs by gender.

Tables 4c and 4d offer similar tests of across- age equality for females and males, respec-

SEXUAL HARASSMENT 79

tively. We find significant differences between adolescence and adulthood in the probability of intrusive questions and invasion of person- al space, and the probabilities increase most precipitously for females. Unlike males, females differ across life course stage in the conditional probability of exposure to offensive jokes (with the probability rising for those in the low harassment class). Perhaps most impor- tantly, the final tests show significant increas- es in the probability of assignment to the high harassment class between adolescence and

adulthood.

To summarize the results thus far, we iden- tify a coherent sexual harassment syndrome within all 4 groups, as indicated by the supe- rior fit of the 2-class models. Nevertheless, we find that this syndrome varies by age and sex, with females and adults more likely than males and adolescents to report core markers. Of course, these behavioral indicators cannot speak to our hypotheses about legal con- sciousness and the association between sub- jective and behavioral harassment. Our interviews suggest that some degree of work- place sexuality is common, though the extent to which it is understood as enjoyable or prob- lematic varies. Therefore, we next consider how our subjective sexual harassment item maps onto the statistical latent classes.

BEHAVIORAL AND SUBJE~

SYNDROME HARASSMENT

Table 5 cross classifies respondents' perceptions of sexual harassment with their statistical assign- ment to the behavioral latent classes. These results are not intended to establish the criteri- on validity of the latent syndrome, but rather they are meant to test hypotheses about gender differences in consciousness of sexual harass- ment. If men experience behavioral sexual harassment but do not count themselves among those eligible to name such experiences sexual harassment, this should be reflected in a lower correlation between the behavioral syndrome and the subjective harassment item for men rel- ative to women. As Table 5 shows, fewer than half of the respondents in either class defined their experiences as sexual harassment. About 41 percent of those in the high harassment class said that they would consider their experiences to be sexual harassment, relative to 14 percent in the low harassment class.14 Table 5 elaborates

l4 We also conducted a supplementary analysis defining behavioral harassment solely in terms of the two core indicators. This yielded a similar degree of association between subjective harassment and these core items (G = .6). Relative to the latent class approach, however, this stricter behavioral standard placed 47 percent more people in the low harassment category who told us in their surveys that they had been sexually harassed.

Table 5. Appraisal of Sexual Harassment by Latent Sexual Harassment Syndrome

Low Sexual High Sexual Self-appraisal of Harassment Harassment Total

Harassment

Overall* Was Not Sexual Harassment Was Sexual Harassment Total

Female** Was Not Sexual Harassment Was Sexual Harassment Total

Male*** Was Not Sexual Harassment Was Sexual Harassment Total

* chi square: 69.126 (1 df); gamma: .63

** chi square: 63.142 (1 do; gamma: .71 *** chi square: 5.356 (1 df); gamma: .36

n (YO) n (??) n (%)

80 AMERICAN SOCIOLOGICAL REVIEW

this table by sex to test hypotheses 3 and 4 regarding sex differences in legal conscious- ness and perceived harassment. Consistent with our third hypothesis, females in both latent classes are more likely than males to report that they have been sexually harassed. Among females experiencing the behavioral syndrome, 53 percent reported they were sexually harassed, relative to 2 1 percent of males who experienced the behavioral syndrome. The large gamma coefficient for females (.71, compared to .36 for males) indicates that their perceptions fit most closely with our statistical class assignments, consistent with legal consciousness arguments (Ewick and Silbey 1998) and our fourth hypoth- esis.

Building on these results, our interviews also suggest that many men lack a cultural catego- ry or reference point from which to understand their experiences with harassing behaviors. Men tended to talk around sexual harassment, oper- ating within a restrictive discourse of "accept- able" masculinity (Lee 2000). For example, rather than directly referencing sexual harass- ment, they often described the harassment in general terms, such as "socially unacceptable" and "a situation," or they described specific behaviors such as "grab-assin." In contrast, most women we interviewed understood individual harassing experiences as part of a broader com- plex of events or as indicative of a larger phe- nomenon having to do with gender and age relations.

Laurie, a white woman who worked as a wait- ress while attending high school, responded affirmatively to the core sexual harassment items in the survey as well as the subjective report of harassment. She attributed some prob- lems with older men to "generational" differ- ences in interaction, but she clearly identified other contact as sexual harassment:

There was physical contact, but it was part of the service, part of being hospitable, that-and there are just some that would pat, or want to grab your cheek, and your face, and that type of--or touch your hand over and over again. So sometimes it was just generational, sometimes it was just sexual harassment. [emphasis added]

The nonchalance with which Laurie explained that "sometimes it was just sexual harassment" parallels that of other female inter- view participants who tended to describe harass- ment as just another obstacle that they routinely confront in their workplaces. Although female participants were not discounting their own or others' actual harassment experiences, most of them were cognizant that sexual harassment is pervasive and were comfortable directly refer- encing it.

To test the statistical significance of the gen- der differences observed in the survey data, we estimated logistic regression equations to model the interaction of sex and class assignment. In model 1 ofTable 6, women are more likely than men to perceive that they have been sexually harassed. Under the interaction coding, however, women in the low harassment class are no more likely than men in the low harassment class to view their experiences as sexual harassment, as shown by the non-significant female effect in model 3. The significant product term in this model indicates sex differences in the effect of the behavioral syndrome on the likelihood of perceiving sexual harassment. This pattern of results is consistent with the idea that males lack a clear cultural reference point to translate the constellation of behaviors they report into a perception of sexual harassment.

An alternative explanation of these findings, however, is that the behavioral syndrome meas- ures something other than sexual harassment. Given the disjuncture between male rates of perceived harassment (14 percent) and the prevalence of the behavioral syndrome (23 per- cent and 31 percent in adolescence and adult- hood, respectively), there is some danger that our latent classes may capture non-sexual forms of harassment or sexual behavior that is not harass- ing. Therefore, we reestimated all models, leav- ing out the two most prevalent, but perhaps least serious, verbal behaviors--offensive jokes, remarks, or gossip directed at you, and direct questioning about your private life. Overall, our findings are robust: the 2-class model provides the best fit for all four of our groups, and those classes reveal a syndrome of sexual harassment (tables available from authors). Fewer people exhibit the behavioral syndrome in these mod- els (8 percent of adolescent males, 9 percent of adolescent females, 11 percent of adult males, and 16 percent of adult females), which tap more egregious sexual harassment: for example, the probability of unwanted touching was more than .4 for all groups and .7 for adult females. Yet the 4-item construct (or an exclusive reliance on the core touching and space indicators) over- looks instances where severe or persistent ver- bal harassment interferes with respondents'jobs or creates an intimidating atmosphere. Note also that these latent class probabilities are sig- nificantly lower than respondents' own appraisals of their situations. Nonetheless, with or without these items, the superior fit of the 2class model suggests that a coherent sexual harassment construct emerges for men and ado- lescents as well as adult women. l5

Table 6. Logistic Regression Estimates Predicting Subjective Sexual Harassment (N = 721)
  Model 1   Model 2   Model 3
  B SE B SE B SE
Female 1.084** .I93 0.997** ,202 0.440 .288
Latent Behavioral Syndrome     1.445** .I88 0.750* .329
Female*Latent Syndrome         1.023* ,403
Constant -1.810** .I63 -2.470** .201 -2.102** ,226

To help assess whether the behavior report- ed on the surveys was explicitly sexual, we relied on our intensive interviews. Male inter- view subjects told us about clients or cowork- ers attempting to kiss them, grab their buttocks,

l5 As an alternative to our latent class approach, an anonymous reviewer proposed estimating a logistic regression equation predicting the overall self- appraisal of harassment using the individual harass- ing behaviors as predictors. Although the individual behaviors are very closely correlated, this approach revealed some interesting patterns that mirror some of the gender differences in the behavioral syndrome reported in Table 5. For females, offensive jokes, invasion of personal space, unwanted touching, and offensive materials are all statistically significant positive predictors of subjective harassment. For males, offensive pictures and physical assault are significant positive predictors, and invasion of per- sonal space and unwanted touching are statistically significant at p < .l.

SEXUAL HARASSMENT 81

or touch them in other ways that made them uncomfortable, and telling sexually explicit, misogynistic, or anti-gay jokes. The women we interviewed similarly described coworkers and managers grabbing their buttocks or breasts, "wandering hands" on their knees and inner thighs, attempts to unsnap their bras, kiss, or rub against them, persistent questions about their sex habits and preferences, and other unwanted sex- ual behaviors. In one case, a female worker described regular parties hosted by the compa- ny chief executive officer (CEO) that involved strip poker, nude hot-tubbing, and erotic dancers.

In some instances, the interviews suggested that workers were themselves unsure whether specific workplace interactions were "sexual enough" to count as sexual harassment. James, a white man who has held a variety of blue col- lar jobs, told us his ambivalence in responding affirmatively to the survey question about "inva- sion of personal space" but not to the global item asking about sexual harassment. He then described situations when coworkers and boss- es made him uncomfortable by getting too phys- ically close to him, touching him on the arm or shoulder, or putting an arm around him. Erin, a working class white woman, reported a wide range of harassing behaviors on the surveys and was classified in the high harassment class in both adolescence and adulthood. She described a situation in which she was initially ambivalent about a fellow custodian's behavior toward her (which included hugs and massages) but came to understand it as sexual harassment as it started "getting kinda creepy" and he ignored her requests to stop. The behavior con- tinued even after a supervisor's warning:

He [the male custodian] said he was sorry and all that and then a couple days later or weeks later he touched me again and I was like, all right. So I told [my boss] again about it. And there was even two occasions where he actually unsnapped my bra while we were at work. Yeah, and I was like, all right, that's it, no more! . .. He would laugh about it and say, "Ha ha, lookit, I can unsnap your bra with one finger." I said, "I don't care how many fin- gers it takes, don't do it. Don't ever do it again."

Our survey data and latent class models thus complement the interviews, each revealing how a range of potentially harassing behaviors clus- ters into a syndrome of sexual harassment.

82 AMERICAN SOCIOLOGICALREVIEW

IN ADOLESCENCESEXUALHARASSMENT AND ADULTHOOD

Table 7 tests our fifth hypothesis regarding gen- der differences in the continuity of harassment during the life course. The table shows a strong association between sexual harassment at the two stages of the life course: 72 percent of respondents who experienced the behavioral syndrome in adolescence also experienced it as adults. Table 7 shows a correlation between behavioral harassment in adolescence and adult- hood that is stronger for males than for females. The same males are thus targeted at two stages of the life course, whereas more females are tar- geted for the first time at the adult stage. A hl1 43 percent of females experienced the syndrome at some point ((72 + 71 +35)1415 = .43)-a figure in line with estimates from previous stud- ies (Benson andThomson 1982; welsh 1999)- relative to 35 percent of males ((39 + 56 +

14)/312=.35).

To test the statistical significance of these differences, we again estimated logistic regres- sion equations, modeling the interaction between sex and adolescent class assignment with a product term in Table 8. Model 1 shows no significant sex differences in the likelihood of behavioral harassment in the adult stage. Model 2 shows that the odds of harassment in the adult stage are ten times higher (e2.326 = 10.2) for adolescents who experienced the behavioral syndrome than for adolescents who did not. Finally, consistent with hypothesis five, model 3 shows significant sex differences in the effect of adolescent class assignment on adult class assignment. The positive female effect in model 3 indicates that, among those not target- ed in adolescence, women are more likely than men to be targeted as adults. Men who are harassed as adolescents, however, are at great risk for harassment as adults: the odds for males in the high adolescent class to be in the high adult class are 20 times higher (e3.036 = 20.8) than the odds for males in the low adolescent class. The corresponding odds are lower among females, in part because there are many more new adult female targets. All adult women are at some risk of sexual harassment and more females than males in our sample were target- ed at some point in their lives.

MASCULINITY AND WORKPLACE

POWER

Although our individual-level survey data can-
not speak to macro-level relationships between power, masculinity, and the cultural meaning of sexual harassment, we can bring some evidence to bear on these relationships by examining pat- terns of association between measures of work- place power, gender relations, and sexual harassment. Tables 9 and 10 address our final hypotheses regarding power and masculinity, reporting logistic regression equations predict- ing the behavioral syndrome and subjective harassment. We consider perceived financial

Table 7. Adult Sexual Harassment Svndrome by Adolescent Sexual Harassment Svndrome

General, Adulta Low Sexual Harassment High Sexual Harassment Total

Female, Adultb Low Sexual Harassment High Sexual Harassment Total

Male, AdultC Low Sexual Harassment High Sexual Harassment Total

a chi square: 163.890 (1 df); gamma: .82 chi square: 66.677 (1 df); gamma: .74 chi square: 104.634 (1 df); gamma: .91

Adolescent
Low Sexual   High Sexual
Harassment   Harassment Total
n (%)   n (%) n (%)
440 (80%)   49 (28%) 489 (67%)
111(20%)   127 (72%) 238 (33%)
551 (100%)   176 (100%) 727 (100%)
237 (77%)   35 (33%) 272 (65.5%)
72 (23%)   7 1 (67Y0) 143 (34.5%)
309 (100%)   106 (100%) 415 (100%)
203 (84%)   14 (20%) 2 17 (70%)
39 (16%)   56 (80%) 95 (30%)
242 (1 00%)   70 (100%) 312 (100%)

SEXUALHARASSMENT 83

control and supervisory authority as dimen- sions of workplace power (Appendix 1 shows the wording of questions and the descriptive statistics for these items). Consistent with our model, those reporting less financial security are most likely to experience the behavioral syn- drome, although the positive effect of supervi- sory authority in model 2 is counter to our expectations.

Holly, a white woman who was the first and only female manager at her company, helped explain why female supervisors reported high levels of harassment in our survey. She noted that her male coworkers had specific expecta- tions about women's workplace roles. She describes her firm as follows: "[It is] an old- school company. It's mostly males. Part of the old boys' club." Holly also describes how male subordinates respond to her as a female man- ager:

They joke periodically about this is the first time a woman's been in a management position there directly under the owner and they'll joke and say, "If we had somebody with balls in this position we'd be getting things done."

A woman's authority does not immunize her from sexual harassment, at least withm a cultural context in which males hold greater power and authority. Although Holly reports a range of harassing behaviors and is classified among the high-harassment group in our statistical mod- els, she did not report in the survey that she was sexually harassed.

Apart from workplace power, the attitudes and behaviors regarding gender relations reveal one facet of the "marginalized masculinities" (Connell 1995:8 1) hypothesized to affect per- ceived harassment. As an attitudinal indicator, the survey asked the high school seniors to describe their beliefs about their spouse work- ing outside the home after they have children. As a behavioral measure, we also indexed the amount of indoor housework that a respondent does each week during the young adult period. Model 3 presents results from the additive model, and model 4 includes the hypothesized gender relations interactions and an interaction between supervisory authority and gender. Model 4 shows that women in supervisory posi- tions and men who do more housework are like- ly to experience the behavioral harassment syndrome.

In models predicting subjective harassment, we also include indicators of the two core behav- ioral harassment items. This helps isolate the independent effects of workplace power and gender relations on perceived harassment from the effects of exposure to the behaviors most commonly associated with sexual harassment. As noted above, heteronormative masculinity implies a construction of sex and gender in which males are predators or protectors rather than targets, which may partially account for low male perceptions of sexual harassment. Although the pattern is somewhat weaker in these models, the results suggest that men with more egalitarian attitudes toward their spouse's work patterns may have a broader frame or cul- tural reference point that gives meaning to their own harassment experiences as a unified con- struct.

To test the robustness and generalizability of these findings on an item that more explic- itly references sexual conduct, we report results of our General Social Survey (GSS) analysis in Table 10. Sexual harassment is measured in the GSS with a complex single item referencing sexual advances, physical contact, and sexual conversations (see Table 10). This item is use- ful for testing the robustness of YDS results because it is unlikely to tap nonsexual workplace conduct and because it was asked of a nation- ally representative adult sample. About 43 percent of women and 26 percent of men reported harassment on this GSS item. These numbers are

Table 8. Logistic Regression Estimates Predicting Sexual Harassment Syndrome in Adulthood (N=727) Model 1 Model 2 Model 3 B SE B SE B SE

Female 0.190 ,160 0.141 ,182 0.458* ,221
Adolescent Latent Syndrome     2.326** ,199 3.036** .346
Female* ~dolescen<~~ndrome Constant 4.829** ,122 -1.458** ,150 -1.137** -1.649** ,435 ,175

84 AMERICANSOCIOLOGICALREVIEW

somewhat higher than for our YDS global per- ceptions indicator and our latent classes.

The GSS models include an additional sta- tistical control for age because the age range of respondents is far wider than the YDS cohort study. Younger respondents were more likely to report harassment in all models despite their more limited work histories and reduced expo- sure time, suggesting that younger cohorts may have greater consciousness of sexual harass- ment. Otherwise, the GSS results generally par- allel those from the YDS. With regard to workplace power, people more vulnerable finan- cially and those with more egalitarian views were most likely to report harassment. With regard to behavior in gender relationships, men who do more housework were again most like- ly to report harassment. These findings lend some support to the argument that men who behave in a way that does not match cultural expectations of heteronormative masculinity may be targets of harassment. In sum, the sup- plementary GSS analysis bolsters our confi- dence in findings based onYDS data and refines conclusions about how sexual harassment is related to workplace power and gender relations.

TAKING STOCK OF MACKINNON'S SOCIOCULTURAL MODEL

The foregoing results suggest that power and masculinity are linked to harassing behaviors and perceived sexual harassment. MacKinnon (1979) posits that sexual harassment derives its meaning from the social context of power rela- tions in the workplace and in society. Though her analysis spurred important and credible work, sexual harassment research has been criticized for a lack of conceptual clarity (Fitzgerald and Shullman 1993; Foulis and McCabe 1997; Patai 1998) and methodological rigor (Welsh 1999). Building on MacKinnon's theory and recent work on gender and legal consciousness, we formulated and partially tested a general model of sexual harassment to explain both behav- ioral harassment and subjective perceptions of harassment. Our results suggest that difficulties in precisely defining sexual harassment arise because the structure and meaning of sexual harassment vary with age and gender. Though sexual harassment emerges as a clear behavioral syndrome across age and gender groups, we find important differences in the specific behav- iors that each group experiences.

MacKinnon's theory and more recent femi- nist work suggest that power and masculinity explain the social distribution of harassment experiences. We therefore tested her most basic prediction that females would experience more harassment than males, extended this model to consider age-based relations of power, and hypothesized that few males would define their harassment experiences as sexual harassment. Although male reports of harassment were high- er than we anticipated, we found general support for these hypotheses. The female rates of harass- ing behaviors increase in the transition to adult- hood, and male rates are comparatively stable. In fact, for males in the high harassment class- es, the likelihood of facing the classic markers actually declines in adulthood. This finding is consistent with our prediction that adolescents are often targeted, in part, because of their rel- ative lack of power in the workplace and the larger society.

Our GSS results and intensive interviews lend further support to this life course finding. Rachel, a working-class woman of color who worked in restaurants during high school, con- sulted an attorney after being harassed by an older male coworker:

He [the supervisor] came for me while I was stand- ing at the drive-through window and he came from behind and grabbed me. And rubbed up against me.

When asked why she thought her supervisor targeted her, Rachel said the following:

Well, for one I was young. And I was a young mother. My supervisor seemed to think, "You must be a freak or something because you have a child at a young age."

Age, race, class, and gender are interlocking dimensions of power that produce unequal social relationships and interactions (West and Fenstermaker 1995). As a young mother and working-class woman of color, Rachel's relative lack of power in the workplace, and in society more generally, may help explain her experience of sexual harassment. In fact, Rachel herself points to her status as a young mother as expla- nation for the harassment. Her supervisor felt free to harass her because as an adolescent and a mother she did not adhere to cultural expec- tations for females her age. She continued to experience harassing behaviors from older male supervisors after high school, when she worked as a telephone sales representative:

Table 9. Workplace Power, Gender Relations, and Sexual Harassment (1999 YDS Data): Logistic Regression Models Predicting Behavioral Syndrome and Subjective Appraisal of Sexual Harassment

Latent Behavioral Syndrome (N = 527) Subjective Harassment (N = 522)

Model 1 Model 2 Model 3 Model 4 Model 1 Model 2 Model 3 Model 4
Female .155 ,153 .061 -.717 1.133** 1.184** 1.033** ,662
  (.191) (.194) (.226) (.618) (.25 1) (.254) (.283) (.776)
Core Behavioral Items                
Invasion of Personal Space         1.024** 1.008** 1.046** 1.040**
          (.246) (.248) (.251) (.252)
Unwanted Touching         1.162** 1.197** 1.230** 1.241**
          (.343) (.347) (.350) (.352)
Workplace Power                
Perceived financial control                

Supervisory authority

Gender Relations and Interaction Terms

Housework by respondent (1 0+ hours)

Spouse should not work after kids

Female*Supervisory authority

Female*Spouse should not work Constant

86 AMERICAN SOCIOLOGICAL REVIEW

Table 10. Workplace Power, Gender Relations, and Sexual Harassment (1996 GSS Data): Logistic Regression Models Predicting Reported Sexual Harassment

Gender Age in years Workplace Power

Satisfaction with financial situation Supervisory authority Gender Relations Share of housework done by respondent

Model 1 Model 2 Model 3 Model 4
.896** .898** .814** 1.670**
(.124) (.124) (.130) (.573)
-.014** -.012** -.009* -.009*
(.004) (.004) (.004) (.004)
  -.270** -.271** -.275**
  (.081) (.081) (.081)
  .248+ .223 .04 1
  (.148) (.149) (.234)
    ,052 .167*
    (.055) (.082)
Wife should help husband's career rather than her own     -.254** -. 186
      (.087) (.153)
Female*Supervisory authority       .303
        (.305)
Female*Housework       -.204+
        (.I 10)
Female*Wife should help husband's career       -.093
        (.183)
Constant -.616** -,247 -.004 -.483
  (.186) (.234) (.336) (.485)

Note: The item is worded as follows: "Sometimes at work people find themselves the object of sexual advances, propositions, or unwanted sexual discussions from co-workers or supervisors. The advances sometimes involve physical contact and sometimes just involve sexual conversations. Has this ever happened to you?"

tp < .lo; *p < .05; **p < .O1

He [the supervisor] would make comments like 'Oh I'd love to see you at Playthings [a local strip club]' or just little smart comments regarding see- ing me in the nude.

Other interview participants described expe- riences that show how age is tied to workplace power dynamics. Erin, who also worked in restaurants during high school, was cornered by a male coworker who was "way older":

He asked me to go in the freezer and get him something. So I went in there and grabbed it and when I turned around with the box he was there and he tried to kiss me. And I was like, "Whoa! You can't kiss me! I don't like you! I'm only 14!" . . . I told him, I said, "No!" (Quotes indicate Erin h statements to her coworker.)

Erin explained that her reaction would be even stronger if the same thing happened today, now that she is in her mid 20s:

I think I'd probably smack the guy in his face maybe. And say, "Hey!" Being the age that I am. But then I didn't know any better. I mean, when you're 14 and there's an older man trying to kiss you, you're kind of flattered. You're like, "Whoa! He thinks I'm an older woman or something." But today I'd probably smack him in his face.

Though Erin is clear that she perceived the situation described above as inappropriate at the time, and would today as well, her descrip- tion indicates that her youth played a role in the event. Several of the males we interviewed also tied age to workplace power dynamics. Cam, an Asian man now in his twenties, noted invasion of personal space and subjective harassment in his survey responses. Cam was visibly upset (he had never discussed the incident before) when describing how an older female client ini- tiated sexual contact with him several years before:

She asked me to direct her to stop someplace and talk. . .And I didn't know what to do. So I stopped [the car] and, I mean, she intended to have a sexual relationship with me. She touched me [indi- cates by touching mouth, chest, and inner thigh],

SEXUAL HARASSMENT $

she asked me if I would kiss, and I said, "No. No, I don't want to do it because you are a married woman." And she said she didn't care because she has an older husband and she wants someone who is young and who takes chances. I told her, "No. I don't want to do that. It's wrong." . . .And she literally scolded me. She said, "Well, you are a very intelligent person in some ways, and very smart, but you're still dumb. You're a very dumb per- son."

Other interview participants reported ado- lescent work settings in which sexual joking, physical horseplay, and more serious harassing behaviors were common. As Rachel's case illus- trates, some adolescents pursue legal remedies when they identify such conduct as sexual harassment.

Although we observed sexual harassment among adolescents and males, we confirmed MacKinnon's prediction that adult women are most often targeted. Today, prevailing cultural understandings of sexual harassment point to adult women as the primary targets, and we found this group most likely to interpret harass- ing behaviors as sexual harassment. Though females experiencing harassment in adoles- cence were also quite likely to be targeted in adulthood, this pattern is even stronger for males. Men experiencing harassment appear to have less powerful workplace positions and more egalitarian gender relationships than other males, but further research is needed to under- stand whether the males who are targeted con- form to dominant cultural constructions of masculinity. Our GSS analysis and interview data suggest that Connell's conception of "mar- ginal masculinities" may prove useful in eluci- dating these relationships. For example, male interview participants discussed sexual orien- tation in relation to sexual harassment across a variety of contexts.

Our application of a latent class approach to the measurement of sexual harassment pursued Lazarsfeld and Henry's (1968:3) classic ques- tion: "whether the patterns of covariation we observe may not tell us something about the defining nature of a concept." Our analysis showed that sexual harassment experiences are best described by simple 2-class models that appear to identify a behavioral syndrome. About 25 percent of adolescents and 30 percent of adults were statistically assigned to the high sexual harassment classes in our 6-item models, with about 8 to 16 percent assigned to the high harassment class in the more restrictive 4-item models. Many others experience at least some degree of unwanted sexual behavior at work; for no group did we find a "non-harassed" latent class. We also observed that invasive behav- iors, such as unwanted touching, usually occur with more common workplace problems such as offensive jokes. Although males and females both experience sexual harassment, we find important differences in the latent structure of the behavioral syndrome. Moreover, the behav- ioral syndrome is itself more closely correlat- ed with "subjective" perceptions for females than for males.

Although MacGnnon initially suggested that harassment of males would be unlikely (1979), her observations led to the general proposition that expressions of gender connote different amounts of power in the social production of sexual harassment. Our interviews provide some important clues in this regard, though further qualitative work is needed to show how domi- nant expressions of gender, such as heteronor- mative masculinity, may be privileged in the workplace (Lee 2000; Quinn 2002). Nevertheless, our analysis resolves some of the tension in debates about whether men and women experience sexual harassment as a sim- ilarly cohesive behavioral syndrome and whether they are subject to similar forms of sexual harassment. Indeed more women appear to experience a virulent form of sexual harass- ment than men, as indicated by their greater likelihood of facing unwanted touching and violations of personal space, the classic mark- ers of sexual harassment. Yet, for those males who were targeted in adolescence, we also found a high degree of life course continuity and repeat harassment in adulthood.

CONCLUSION

Today, sexual harassment occupies a peculiar place in American culture. On the one hand, sexual harassment laws and policies represent the "great success story of contemporary fem- inism," forever altering workplace relations between the sexes and providing tangible insti- tutions for redressing grievances (Patai 1998:4). On the other, an undercurrent of patronizing skepticism often pervades discussions of the phenomenon. In private conversations and pub- lic discourse, some are asking whether regula- ticn of workplace sexual conduct has "gone too

88 AMERICAN SOCIOLOGICAL REVIEW

far" and whether the "whole paradigm of sex- ual harassment" should be reconsidered (Talbot 2002:95). Moreover, many feminist legal schol- ars are now challenging the very foundations of sexual harassment law and policy (Abrams 1998; Ehrenreich 1999; Schultz 1998,2003).

We suggest that the public skepticism, main- stream scientific neglect (Sever 1996), and legal dilemmas posed by the phenomenon are caused partly by untested assumptions about gender and power. We test some of these assumptions with an analysis of age and gender differences in the structure and meaning of harassment experiences and find evidence largely consistent with feminist models. We generalize MacKinnon's (1979) theory of the sexual harassment of working women to hypothesize that dominant constructions of heteronorma- tive masculinity also shape the harassment expe- riences of other workers. These expectations are largely supported, although more evidence is needed on several key points, particularly those relating to legal consciousness of sexual harassment.

In general, we found it difficult to tap men's experiences with sexual harassment, whether using intensive interviews or survey methods. In some instances in which sexual touching occurred, the men said that they had never dis- cussed these incidents with anyone prior to our interview. These men seem uncomfortable using existing cultural categories or vocabulary in discussing their experiences with harassing behaviors. We need better measures to learn how the workplace is gendered for men and women and to test whether men who do not

APPENDIX

adhere to dominant constructions of masculin- ity are more vulnerable to harassment.

In conclusion, the evidence supports MacKinnon's basic propositions that sexual harassment derives from power and masculin- ity-for males and adolescents as well as for adult women. Moreover, the high adolescent rates and clear harassment syndromes we observe across age and sex groups indicate that sexual harassment could be a general social phenomenon. Nevertheless, differences in expe- riences remain, and adult women are most sub- ject to classic markers. A model of workplace power and gender stereotyping appears most consistent with the social distribution of harass- ing behaviors as well as the age and sex differ- ences observed in their meaning to targets.

Christopher Uggen is Associate Professor of Sociology, Life Course Center afiliate, and McKnight Presidential Fellow at the University of Minnesota (www.soc. umn. edu/-uggen). He studies the sociolo- gy of crime, law, and deviance, and his currentproj- ects involve studies of felon voting rights, sexual harassment, and the work, family, and civic life offor- mer criminal offenders. Rth JefManza, he is coau- thor of Locked Out: Felon Disenfranchisement and American Democracy (Oxford University Press,

forthcoming).

Amy Blackstone is Assistant Professor of Sociology at the University of Maine. Her research interests include sociology of gender, social movements, and activism and other forms of civic participation. In addition to her work on the current project with Christopher Uggen on sexual harassment, her research includes a study of the social construction of gender andpolitics in the breast cancer and anti- rape movements.

Table 1. Variable Descriptions and Descriptive Statistics for Youth Development Study and General Social Survev Workvlace Power and Gender Relations Items

Mean

Variable Description Coding (s.d.)
Youth Development Study      
Workplace Power      
Perceived financial control I feel I am in control of my 1 = Strongly disagree 2.78
  financial situation. 4 = Strongly agree (.76)
Supervisory authority Do you supervise other work- 0 =No .29
  ers on your job? 1 =Yes (.46)

(Continued on next page)

SEXUAL HARASSMENT 89

Table 1. (Continued).

Mean
Variable Descrivtion Coding (s.d.)
Gender Relations    
Spouse should not work after kids If you have children, do you 1 =Yes 1.34
  think that your spouse will 2 = Maybe  
  work outside the home? 3 = No  
Housework by respondent (lo+ hours) Hours per week spent on 0 = Fewer than 10 41%
  indoor household chores 1 = 10 or more  
General Social Survey    
Workplace Power    
Satisfaction with financial situation So far as you and your family 1 = Not satisfied at all 2.01
  are concerned, would you 2 = More or less (.74)
  say that you are pretty well satisfied  
  satisfied with your present 3 = Pretty well  
  financial situation, more or satisfied  
  less satisfied, or not satis-  
  fied at all?  
Supervisory authority In your job, do you supervise 1 = No 20%
  anyone who is directly 2 =Yes  
  responsible to you?  
Age Respondent's age in years 45.38
Gender Relations     (16.97)
Wife should help husband's career rather It is more important for a wife 1 = Strongly disagree 2.01
than her own to help her husband's career 4 = Strongly agree than to have one herself. (.74)
Share of housework done by respondent How much of the work 1 =Very little or none 3.66
  around your home do you 5 =All (1.18)
  do (including cooking,  
  grocery shopping, and  
  doing little repair jobs)?  

REFERENCES Confluence of Authority Relations, Sexual Interest and Gender Stratification." Social Problems

Abrams, Kathryn. 1998. "The New Jurisprudence of 29:236-51.

Sexual Harassment." Cornell Law Review Berdahl, Jennifer L., Vicki J. Magley, and Craig R.

83:1169-230.

Waldo. 1996. "The Sexual Harassment of Men?:

Acker, Joan. 1990. "Hierarchies, Jobs, Bodies: A Exploring the Concept with Theory and Data."

Theory of Gendered Organizations." Gender &

Psychology of Women Quarterly 2052747.

Society 4:139-58.

Bird, Sharon R. 1996. "Welcome to the Men's Club: American Association of University Women Homosociality and the Maintenance of Hegemonic

(AAUW). 2001. Hostile Hallways: The AAUW

Masculinity." Gender & Society 10: 120-32.

Survey on Sexual Harassment in America's

Burt, Martha R. and Rhoda E. Estep. 1981.

Schools. Washington: AAUW Educational "Apprehension and Fear: Learning a Sense of Foundation.

Sexual Vulnerability." Sex Roles 7:5 11-22.

Anderson, Eric. 2002. "Openly Gay Athletes:

Butler, Judith. 1990. Gender Trouble: Feminism and

Contesting Hegemonic Masculinity in a

the Subversion of Identity. New York: Routledge.

Homophobic Environment." Gender & Society Cahill, Mia L. 200 1. The Social Construction of

16:860-77. Sexual Harassment Law. Aldershot, U.K.: Ashgate.

Arvey, Richard D. and Marcie A. Cavanaugh. 1995. "Using Surveys to Assess the Prevalence of Sexual

Cleveland, Jeanette N. and Melinda E. Kerst. 1993. Harassment: Some Methodological Problems." "Sexual Harassment and Perceptions of Power: Journal of Social Issues 5 1 :39-52. An Under-Articulated Relationship." Journal of

Barr, Paula A. 1993. "Perceptions of Sexual Vocational Behavior 42:49-67. Harassment." Sociological Inquily 63:460-70. Committee on the Health and Safety Implications of Benson, Donna J. and Gregg E. Thomson. 1982. Child Labor (CHSICL). 1998. Protecting Youth at "Sexual Harassment on a University Campus: The Work: Health, Safety, and Development of Working

go AMERICAN SOCIOLOGICAL REVIEW

Children and Adolescents in the United States.

Washington: National Academy Press.

Connell, R.W. 1987. Gender and Power: Society, the Person, and Sexual Politics. Stanford: Stanford University Press. . 1992. "A Very Straight Gay: Masculinity, Homosexual Experie~lce, and the Dynamics of Gender." America~z Sociological Review 57:735-51. . 1995.Masculinities. Berkeley: University of California Press. . 2000. The Men and the Boys. Berkeley: University of California Press. . 2002. Gender.Cambridge: Polity Press.

Davis, James A,, Tom W. Smith, and Peter V Marsden. 2003. General Social Surveys, 1972-2002: [Computer File]. Chicago, IL: National Opinion Research Center [producer]. Ann Arbor, MI: Inter- university Consortium for Political and Social Research [distributors]. Dayton, C. Mitchell. 1998. Latent Class Scaling Analysis. Thousand Oaks, CA: Sage. DeCoster, Stacy, Sarah Beth Estes, and Charles W. Mueller. 1999. "Routine Activities and Sexual Harassment in the Workplace." Work and Occupations 26:2149. Dellinger, Kirsten and Christine L. Williams. 2002. "The Locker Room v. the Dorm Room: The Cultural Context of Sexual Harassment in Two Magazine Publishing Organizations," Social Problems 49: 242-57. Demetriou, Demetrakis Z. 2001. "Connell's Concept of Hegemonic Masculinity: A Critique." Theory and Society 30:337-61.

Doe I,:Belleville, 119 F.3d 563 (7th Cir. 1997). Donaldson, Mike. 1993. "What is Hegemonic Masculinity?'Theory and Society 22:643-57. Donovan, Brian. 1998. "Political Consequences of Private Authority: Promise Keepers and the Transformation of Hegemonic Masculinity." Theory and Society 27:8 17-43. Ehrenreich, Rosa. 1999. "Dignity and Discrimination: Toward a Pluralistic Understanding of Workplace Harassment." Georgetown Law Journal 88: 144. Eliason, Scott. 1997. The Categorical Data Analysis System. Version 4.0 of MLLSA. Ellison v. Brady, 924 F.2d 872 (9th Cir. 1991). Estrich, Susan. 1987. Real Rape. Cambridge, MA: Hamard University Press. Ewick, Patricia, and Susan S. Silbey. 1998. The Common Place of Law. Chicago: University of Chicago Press. Finch, Michael, Michael J. Shanahan, Jeylan T. Mortimer, and Seongryeol Ryu. 1991. "Work Experience and Control Orientation in Adolescence." American Sociological Review 56:597-611. Fineran, Susan. 2002. "Adolescents at Work: Gender Issues and Sexual Harassment." Violence Against Women 8:953-67.

Fitzgerald, Louise F. and Sandra L. Shullman. 1993. "Sexual Harassment: A Research Analysis and Agenda for the 1990s." Journal of Vocational Behavior 42:5-27.

Fitzgerald, Louise F., Sandra L. Shullman, Nancy Bailey, Margaret Richards, Janice Swecker, Yael Gold, Mirni Ormerod, and Lauren Weitzman. 1988. "The Incidence and Dimensions of Sexual Harassment in Academia and the Workplace." Journal of Vocational Behavior 32: 152-75.

Folgero, Ingebjorg S. and Ingrid H. Fjeldstad. 1995. "On Duty-Off Guard: Cultural Norms and Sexual Harassment in Service Organizations." Organization Studies 16:299-3 13.

Foulis, Danielle and Marita P. McCabe. 1997. "Sexual Harassment: Factors Affecting Attitudes and Perceptions." Sex Roles 37:773-98.

Francke, Katherine. 1997. "What's Wrong with Sexual Harassment?" Stanford Law Review 49:691-772.

Gallagher, Sally K. and Christian Smith. 1999. "Symbolic Traditionalism and Pragmatic Egalitarianism: Contemporary Evangelicals, Families, and Gender." Gender & Society 13:211-33.

Gelfand Michele J., Louise F. Fitzgerald, and Fritz Drasgow. 1995. "The Structure of Sexual Harassment: A Confirmatory Analysis across Cultures and Settings." Journal of Vocational Behavior 47: 164-77.

Gillespie, Dair L. and Ann Leffler. 1987. "The Politics of Research Methodology in Claims-Making Activities: Social Science and Sexual Harassment." Social Problems 34:490-501. Giuffre, Patti A. and Christine L. Williams. 1994. "Boundary Lines: Labeling Sexual Harassment in Restaurants." Gender & Society 8:378-401.

Gruber, James. 1992. "A Typology of Personal and Environmental Sexual Harassment: Research and Policy Implications for the 1990s." Sex Roles 26:447-64. . 1998. "The Impact of Male Work Environments and Organizational Policies on Women's Experiences of Sexual Harassment." Gender & Society 12:301-20. Gutek, Barbara A. and Aaron Groff Cohen. 1987. "Sex Ratios, Sex Role Spillover, and Sex at Work: A Comparison of Men's and Women's Experiences." Human Relations 40:97-115. Gutek, Barbara A., Aaron Groff Cohen, and Alison

M. Konrad. 1990. "Predicting Social-Sexual Behavior at Work: A Contact Hypothesis." Academy of Management Journal 33:560-77.

Hall, Steve. 2002. "Daubing the Drudges of Fury: Men, Violence and the Piety of the 'Hegemonic Masculinity' Thesis." Theoretical Criminology 6:35-61.

Hall v. Gus Construction, 842 F.2d 1010 (8th Cir. 1988). Halley, Janet. 2002. "Sexuality Harassment." Pp.

SEXUAL HARASSMENT 91

80-1 04 in Left LegalisdLeft Critique, edited by

W. Brown and J. Halley. Durham, N.C.: Duke

University Press.

Harris v. Forklift Systems, Inc., 5 10 U.S. 17 (1 993).

Hindelang, Michael J., Travis Hirschi, and Joseph G. Weis. 198 1. Measuring Delinquency. Beverly Hills, CA: Sage.

Horney, Julie and Ineke Haen Marshall. 1991. "Measuring Lambda Through Self-Reports." Criminology 29:47 1-95.

Jaschik, Mollie and Bruce R. Fretz. 1991. "Women's Perceptions and Labeling of Sexual Harassment." Sex Roles 25:19-23. Jefferson, Tony. 2002. "Subordinating Hegemonic Masculinity." Theoretical Criminology 6:63-88. Kalof, Linda, Kimberly L. Eby, Jennifer L. Matheson, and Rob J. Kroska. 2001. "The Influence of Race and Gender on Student Self-Reports of Sexual Harassment by College Professors." Gender & Society 15:282-302. Katz, Roger C., Roseann Hannon, and Leslie Whitten. 1996. "Effects of Gender and Situation on the Perception of Sexual Harassment." Sex Roles 34:3542. Keashly, Loraleigh. 2001. "Interpersonal and Systemic Aspects of Emotional Abuse at Work: The Target's Perspective." Violence and Victims 16:233-68.

Kimmel, Michael S. 1994. "Masculinity as Homophobia: Fear, Shame, and Silence in the Construction of Gender Identity." Pp. 11941 in Theorizing Masculinities, edited by H. Brod and

M. Kaufman. Thousand Oaks, CA: Sage.

Kohlman, Marla H. 2003. "The Demographics of Sexual Harassment." Presented at the annual meet- ings of the Law and Society Association, June 5, Pittsburgh, PA.

Lazarsfeld, Paul F., andNeil W. Henry. 1968. Latent Structure Analysis. Boston: Houghton Mifflin.

Lee, Deborah. 2000. "Hegemonic Masculinity and Male Feminisation: The Sexual Harassment of Men at Work." Journal of Gender Studies 9:141-55.

Lorber, Judith. 1998. "Men's Gender Politics." Gender & Society 12:469-77. MacKinnon, Catharine. 1979. Sexual Harassment of Workrng Women: A Case of Sex Discrimination.

New Haven: Yale University Press.

. 1987. "Sexual Harassment: Its First Decade in Court," Pp. 103-1 6 in Feminism Unmodified: Discourses on Life and Law. Cambridge: Harvard University Press.

. 1997. Oncale v. Sundowner Offshore Services, Inc., No. 96-568, Amici Curiae Brief in Support of Petitioner, UCLA Women k Law Journal 8:9-44.

. 2002. "The Logic of Experience: Reflections on the Development of Sexual Harassment Law." Georgetown Law Journal 90:813-33.

Marshall, Anna-Maria. 1998. "Closing the Gaps: Plaintiffs in Pivotal Sexual Harassment Cases." Law and Social Inquiry 23:761-93.

Martin, PatriciaYancey. 1998. "Why Can't a Man Be More Like a Woman? Reflections on Connell's Masculinities." Gender & Society 12:469-77.

McCutcheon, Allan L. 1987. Latent Class Analysis. Newbury Park, CA: Sage.

McGuffey, C. Shawn and B. Lindsay Rich. 1999. "Playing in the Gender Transgression Zone: Race, Class, and Hegemonic Masculinity in Middle Childhood." Gender & Society 13:608-27.

Meritor v. Vinson, 477 U.S. 57 (1986).

Merry, Sally Engle. 1990. Getting Justice and Getting Even: Legal Consciousness among Working Class Americans. Chicago: University of Chicago Press.

Miller, Rowland S., James A. Johnson, and Judy K. Johnson. 199 1. "Assessing the Prevalence of Unwanted Childhood Sexual Experiences." Journal of Psychology and Human Sexuality 4:43-54.

Morgan, Phoebe A. 1999. "Risking Relationships: Understanding the Litigation Choices of Sexually Harassed Women." The Law and Society Review 33:201-26.

Mortimer, Jeylan. 2003. Working and Growing Up in America. Cambridge: Harvard. University Press.

Mueller, Charles W., DeCoster, Stacy, and Sarah Beth Estes. 200 1. "Sexual Harassment in the Workplace: Unanticipated Consequences of Modern Social Control in Organizations." Work and Occupations 28:41146.

Nelson, Andrea and Pamela Oliver. 1998. "Gender and the Construction of Consent in Child-Adult Sexual Contact: Beyond Gender Neutrality and Male Monopoly." Gender & Society 12554-77. Oncale v. Sundowner Oflshore Services, 523 U.S. 75 (1998).

Padavic, Irene and James D. Orcutt. 1997. "Perceptions of Sexual Harassment in the Florida Legal System: A Comparison of Dominance and Spillover Explanations." Gender & Society 11:682-98.

Patai, Daphne. 1998. Heterophobia: Sexual Harassment and the Future ofFerninism. Lanham, MD: Rowman & Littlefield.

Quinn, Beth A. 2002. "Sexual Harassment and Masculinity: The Power and Meaning of 'Girl Watching."' Gender & Society 16:386402.

Ragins, Belle Rose and Terri A. Scandura. 1995. "Antecedents and Work-Related Correlates of Reported Sexual Harassment: An Empirical Investigation of Competing Hypotheses." Sex Roles 32:429-55.

Reiter, Ester. 1991. Making Fast Food: From the Frying Pan into the Fryer. Montreal: McGill Queen's University Press.

Richman, Judith A., Kathleen M. Rospenda, Stephanie J. Nawyn, Joseph A. Flaherty, Michael Fendrich, Melinda L. Drum, and Timothy P.

ga AMERICAN SOCIOLOGICAL REVIEW

Johnson. 1999. "Sexual Harassment and Generalized Workplace Abuse." American Journal of Public Health 89:358-63.

Rogers, Jackie Krasas and Kevin D. Henson. 1997. "'Hey, Why Don't You Wear a Shorter Skirt?' Structural Vulnerability and the Organization of Sexual Harassment in Temporary Clerical Employment." Gender & Society 1 1 :2 15-37.

Rospenda, Kathleen M., Judith A. Richman, and Stephanie J. Nawyn. 1998. "Doing Power: The Confluence of Gender, Race, and Class in Contrapower Sexual Harassment." Gender & Society 12:4040. Saguy, Abigail. 2003. What is Sexual Harassment? From Capitol Hill to the Sorbonne. Berkeley: University of California Press. Schultz, Vicki. 1998. "Reconceptualizing Sexual Harassment." Yale Law Journal 107: 1683-32. . 2001. "Talking About Harassment." Journal of Law and Policy 9:417-33. . 2003. "The Sanitized Workplace." In Press. Yale Law Journal. Sever, Aysan. 1996. "Mainstream Neglect of Sexual Harassment as a Social Problem." The Canadian Journal of Sociology 21:185-202. . 1999. "Sexual Harassment: Where We Were, Where We Are and Prospects for the New Millennium." The Canadian Review of Sociology andAnthropo1og-y 36:469-97. Smith, Michael D. 1994. "Enhancing the Quality of Survey Data on Violence Against Women: A Feminist Approach." Gender & Society 8: 109-27. Stambaugh, Phoebe Morgan. 1997. "The Power of Law and the Sexual Harassment Complaints of Women." National Women k Studies Association Journal 9:2342.

Steinberg, Laurence, Ellen Greenberger, Maryann Jacobi, and Laurie Garduque. 198 1. "Early Work Experience: A Partial Antidote for Adolescent Egocentrism." Journal of Youth and Adolescence 10:141-57.

Talbot, Margaret. 2002. "Men Behaving Badly." New York Times Magazine, October 13, pp. 52-57,82, 84, 95.

Tangri, Sandra S., Martha R. Burt, and Leanor B. Johnson. 1982. "Sexual Harassment at Work: Three Explanatory Models." Journal of Social Issues 38:33-54.

Thacker, Rebecca A. 1996. "A Descriptive Study of

Situational and Individual Influences upon Individuals' Responses to Sexual Harassment." Human Relations 49: 1 105-122.

Thorne, Barrie. 1993. Gender Play: Girls and Boys in School. New Brunswick, NJ: Rutgers University Press.

U.S.
Census Bureau. 2000. Labor Force, Employment, and Earnings. Statistical Abstract of the United States. Washington, DC: U.S. Government Printing Office.
U.S.
Department of Justice. 2002. Sourcebook of Criminal Justice Statistics, 2001.Washington, DC:

U.S. Government Printing Office.

U.S.
Equal Employment Opportunity Commission. 2003. Sexual Harassment Charges EEOC & FEPAs Combined. Retrieved June 15, 2003 (http:llwww.eeoc.gov/statslharass.html) .

U.S. Merit Systems Protection Board. 1988. Sexual Harassment in the Federal Government.

Washington, DC: U.S. Government Printing Office.

Vaux, Alan. 1993. "Paradigmatic Assumptions in Sexual Harassment Research: Being Guided Without Being Misled." Journal of Vocational Behavior 42: 1 16-35.

Vento, Carol Schultz. 2001. "When is Work Environment Intimidating, Hostile or Offensive, so as to Constitute Sexual Harassment under State Law?" American Law Reports 5th 93:47-107.

Welsh, Sandy. 1999. "Gender and Sexual Harassment." Annual Review of Sociology

25: 169-90.

. 2000. "The Multidimensional Nature of Sexual Harassment: An Empirical Analysis of Women's Sexual Harassment Complaints." Violence Against Women 6: 1 1841.

Welsh, Sandy, and Annette Nierobisz. 1997. "How Prevalent is Sexual Harassment: A Research Note on Measuring Sexual Harassment in Canada." Canadian Journal of Sociology 22:505-22.

West, Candace, and Sarah Fenstermaker. 1995. "Doing Difference." Gender & Society 9:8-37. West, Candace, and Don H. Zimmerman. 1987. "Doing Gender." Gender & Society 1:125-5 1.

Williams, Christine L., Patti A. Giuffre, and Kirsten Dellinger. 1999. "Sexuality in the Workplace: Organizational Control, Sexual Harassment, and the Pursuit of Pleasure." Annual Review of Sociology 25:73-93.

Comments
  • Recommend Us