Sexual harassment: an experimental test of sex-role spillover theory.

As sexual harassment has become more widely recognized in the workplace and in the courts as a form of discrimination against women, research questions about the nature and causes of harassment have arisen (Fitzgerald, 1993). And, in searching for answers, it has become increasingly clear that sexual harassment does not affect all women equally in all lines of work (Yoder, 1994). Survey research on sexual harassment has shown that prevalence rates are higher in workplaces and occupations in which women have been traditionally under-represented, particularly in blue-collar occupations (Gutek, 1985). In addition, sexual harassment of blue-collar women in nontraditional occupations appears to be more severe, with these women at greatest risk of physical harm (Fitzgerald, 1993). Although surveys and interviews have provided insights into the occupational context in which female employees are more likely to be harassed, few experimental studies have examined the social psychological factors that affect perceptions of harassment in different work contexts.

One theoretical framework that has implications for understanding the social psychological dynamics of sexual harassment in the workplace is sex-role spillover theory (Gutek, 1985). Sex-role spillover theory applies the research literature on stereotyping and social categorization to the phenomenon of sexual harassment. The theory suggests that women working in certain environments will be particularly vulnerable to sexual harassment. In workplaces with skewed sex ratios, the theory suggests that female employees will be categorized primarily in terms of their gender. The sex ratio can be skewed at the work-group level (a work group is the group of employees who regularly come into contact with each other) or at the level of occupation (occupations may be primarily male or female dominated, or may be integrated). In a work group that is male dominated, the female employee's gender role will be more salient, leading to a greater likelihood of sex-role spillover. In addition, male-dominated workplaces tend to be more sexualized, even when women are not present. Crude jokes, pornographic magazines, and sexual banter are often prevalent in male-dominated workplaces, and so, when a female employee works in this environment, she may become the target of the prevailing sexual climate (Gutek, 1985).

In general, when occupations have a skewed sex ratio, the gender-role expectations for the dominant sex may spill over into the occupational role. Women in female-dominated occupations, according to sex-role spillover theory, are often in situations in which attributes associated with their occupational role may overlap attributes associated with their gender role. Because the traditionally employed female's gender role is made more salient by the expectations of her occupational role, it may be difficult to perceive the extent to which she is being treated sexually and unprofessionally. For example, a woman in a female-dominated occupation, such as a secretary or a flight attendant, may be expected to act in accordance with the female stereotype. The prescription to be physically attractive and to be flattered by sexual attention may be perceived as part of her occupational role. By contrast, females in male-dominated occupations are in an environment, in which male gender-role attributes spill over into the occupational role. In this context, the female employee's gender attributes may be made salient by virtue of the contrast between her gender and her masculine work role. On the other hand, the female in a nontraditional workplace may be perceived to have assimilated male gender-role attributes and hence may be stereotyped as unfeminine, butch, lesbian, and so forth. Research on sex-role spillover theory, however, does not offer clear predictions on whether sexual harassment will be more likely when the female employee's sex and occupational gender role overlap (i.e., traditional occupation) or whether the likelihood of harassment will increase when the employee's sex and occupational gender role are incongruent (i.e., nontraditional occupation).

One of the few experimental studies to examine the way in which skewed occupational sex ratios would influence perceptions of sexual harassment was conducted by Sheffey and Tindale (1992). Subjects read vignettes of ambiguously harassing incidents in which the female target was employed in either a traditionally female, a traditionally male (nontraditional), or an integrated occupation, and rated the extent to which they felt the behavior was sexual harassment, the appropriateness of the behavior, and the frequency of the behavior's occurrence. Sheffey and Tindale found that participants rated the behavior as less harassing, more appropriate, and occurring more frequently when the target of the harassment was employed in a traditionally female occupation, as compared with a nontraditional or integrated occupation. These results were interpreted as support for sex-role spillover theory. Because the traditional female's gender role overlapped her occupational role, ambiguous sexual advances toward her were perceived as appropriate to the workplace. These same behaviors were seen as less appropriate when the target was employed in a nontraditional or integrated occupation, presumably because there was less overlap between the female employee's gender role and her occupational role.

In light of the survey data that show women in non-traditional occupations to be at greater risk of harassment (Gutek, 1985), Sheffey and Tindale's (1992) conclusion that sex-role spillover operates in female-dominated but not male-dominated occupations should be examined in the context of the type of sexual harassment contained in these scenarios. These scenarios consisted of relatively mild or ambiguous examples of unwanted sexual attention (e.g., "You must be doing a lot of running these days, your body looks terrific" and "Why don't we go where we can speak more privately?"). The use of incidents of this sort that are not typical of the more severe sexual harassment claims that are pursued in organizational and legal contexts, we hypothesize, limits the generalizability of Sheffey and Tindale's conclusion that sex-role spillover operates to a greater degree for targets of sexual harassment who are employed in a traditionally female occupation.

Thus, we would argue, type of harassment must be considered to improve our understanding of the dynamics of sex-role spillover. The most influential theory-based typology of sexual harassment in the psychological research literature was derived by factor analyzing women's reported experiences with sexual harassment (Fitzgerald & Hesson-McInnis, 1989). Three categories emerged: unwanted sexual attention, gender harassment, and sexual coercion. Unwanted sexual attention ranges from seductive behavior and "come-ons" to sexual assault; gender harassment is characterized by a "hostile environment" that can range from "insulting, misogynistic and degrading" remarks to life threats (e.g., a cup of water was placed above the battery of a woman's forklift so she would be electrocuted [Swan & Schneider, 1994]); and sexual coercion ranges from solicitation of sex by promises of rewards to explicit threats.

The present study was designed to examine how the occupation type of the female target will affect the way in which different types of sexual harassment are perceived. Because women are more likely to be sexually harassed when the work group is , male dominated (Gutek, 1985), women as targets of harassment in our scenarios were all employed in settings in which the work group would be perceived as primarily male (e.g., a loading dock or a steel mill). Finally, we chose to focus on harassment by a coworker rather than a supervisor because coworker harassment has been shown to be more prevalent, although less well studied (Fitzgerald, 1993).

In general, we expected that the extent to which sexual harassment was perceived would be affected by an interaction between the occupation of the target person and the type of harassment. Consistent with Sheffey and Tindale (1992), it was expected that incidents of unwanted sexual attention would less likely be perceived as sexual harassment when the target's occupation was traditionally female, because the attributes associated with the female gender role would spill over into the occupational role. However, survey and qualitative data describing the discriminatory environment that may confront female employees in nontraditional occupations (Carothers & Crull, 1984; Fitzgerald, 1993) suggest that gender harassment would less likely be perceived as harassing when the target was in a nontraditional occupation. In this case, it was believed that the male gender-role attributes associated with the nontraditional female employee's occupation would result in the female target being perceived as a role deviate, with gender harassment essentially being seen as part of the price she has to pay for access into a male-dominated field. Finally, we expected that the effects of sex-role spillover would be least likely when the type of harassment was sexual coercion because sexual coercion represents the type of behavior that most closely fits lay definitions of sexual harassment (Frazier, Cochran, & Olson, 1995). Hence the occupation of the female target should be less likely to influence judgments of harassment for incidents of sexual coercion.

We also included the dimension of severity to unconfound the effects associated with type of harassment with the effects associated with severity of harassment. Gutek and O'Connor (1995) make the argument that the ambiguity of incidents employed in scenario studies contributes to diverging perceptions of harassment. They argue that if incidents chosen were unambiguous, there would be a greater consensus in labeling the act, as sexual harassment. Because our prediction is based on the theory that diverging perceptions are affected by the occupational context surrounding the incident, rather than just some quality inherent in the incident, it is crucial that we are able to demonstrate that our results were not contingent on the choice of ambiguous incidents. Because harassment is more likely to be perceived when it is unambiguous and is considered to be more severe when it contains an element of physical contact by the harasser (Gutek, Morasch, & Cohen, 1983), the high-severe condition was operationalized by the presence of physical contact by the harasser, whereas the low-severe condition did not contain this element of physical contact. Furthermore, it was …