The Effect of Prior Experience, Gender, and Age
on the Identification of Sexual Harassment

Although numerous definitions of sexual harassment exist, they all tend to share the same key characteristics: the offender has power over the victim, the behavior is unwanted (as perceived by the victim), there is some negative harm or outcome, and the occurring behavior has a sexual component which can vary (McKinney 1992). Within the parameters of these criteria, a wide range of behaviors can be viewed as sexual harassment. These behaviors can be conceptualized as falling on a continuum ranging from fairly mild to severe (Adams, Kottke, and Padgitt 1983). McKinney (1990) has classified behaviors on this continuum into eight categories. These are sexist comments, undue attention, verbal sexual advances, body language, invitations, physical advances, sexual propositions, and sexual bribery. Classifications like this are useful because it allows us to determine where agreement and lack of consensus exist in the labeling of behaviors.

The variable that is most predictive of whether or not someone will label a behavior as sexual harassment is the gender of the labeler. Numerous researchers have found that men are less likely to consider a behavior to be sexual harassment, are more likely to be flattered by sexual overtures, and are more likely to think that a behavior that they do consider to be sexual harassment would nonetheless be flattering to a woman (see Riger, 1991). Some of the gender related findings have been explained using attribution theory. Pryor and Day (1988) have found that when evaluated from the victim’s perspective, more behaviors are likely to be labeled sexual harassment than when evaluated from the other perspective. Because women and men are likely to take different perspectives when evaluating behaviors, they label those behaviors differently. However, using attribution theory alone does not help explain all of the research on differences in labeling sexual harassment. For instance, while this theory can explain between gender differences, it does not explain within gender differences as well.

One factor that helps to explain the likelihood of men and women taking the actor’s or victim’s perspective is prior experience with sexual harassment. It has been established that women are much more likely to be victims of sexual harassment than men (e.g. Fitzgerald and Hesson-McInnis, 1989; Konrad and Gutek, 1986). We hypothesize that a person who has been a victim of sexual harassment is more likely to take the victim’s view, regardless of gender, than one who has not ever experienced sexual harassment.

A second question must be raised when considering the current findings on the labeling of sexual harassment. As cited above, women are more likely to say that a behavior is sexual harassment than men, and we believe victims would similarly be more likely to label a behavior sexual harassment than non-victims. Does this increased tendency to label a behavior sexual harassment imply that the labeling is inaccurate? We do not believe that this is necessarily the case. However, no study of which we are aware addresses this issue. We believe that prior experience with sexual harassment will sensitize a person to the subtleties of a behavior, and enable them to categorize a behavior more accurately.

Our Hypotheses

Consistent with prior studies, women will be more likely to label a behavior as sexual harassment, and label it accurately, than men.

Faculty will be more likely to label a behavior as sexual harassment, and label it accurately, than undergraduates, having been in the work-force with more opportunities for exposure to sexual harassment.

Victims of sexual harassment will be more likely to label a behavior as sexual harassment, and label it accurately, than non-victims.

There will be an interaction between the severity of the behavior depicted in the scenario and the labeling of the behavior by different groups: greatest variation in labeling will be seen for the least severe behaviors.

Participants

All of the personnel listed as faculty outside of the psychology department in the Binghamton University telephone book, N = 295, were emailed a request to complete a short, anonymous survey on sexual harassment on the Internet. Two days after the original mailing, a longer explanation of the purpose of the survey were emailed to faculty because of faculty distress. Several faculty members as well as representatives of the Faculty Senate had called the Chair of Binghamton University’s Human Subjects Research Committee expressing concern that the survey was an administrative initiative. The faculty were reassured that the survey was not connected to administration. Finally, one week later, faculty were emailed a second notice reminding them to complete the survey. A total of 42 faculty, or 14 percent of those receiving email requests, completed the survey. Thirty-eight percent of the faculty were female and 33 percent of faculty had some prior experience with sexual harassment. Sixty-three percent of female faculty and 18 percent of male faculty had experienced sexual harassment.

Undergraduate students (N = 105) were recruited from the Psychology Department subject pool to complete the survey, also over the Internet. Forty-nine percent of the students were female and 29 percent of the students had experienced sexual harassment. Fifty-three percent of female students and 6 percent of male students had experienced sexual harassment. (See Figure 1)

Materials

An eight item scale developed by the researchers was used. McKinney’s (1990) classification of sexual harassment behaviors was used to develop the scenarios. Fifteen advanced undergraduate students then independently evaluated the scenarios to determine how closely each scenario reflected McKinney’s definition. Scenarios were changed until the students agreed upon the classification of each scenario. The scenarios were also categorized into mild, moderate, and severe by the raters. Participants read each scenario and assigned them to one of four categories: no sexual harassment was depicted, or mild, moderate, or severe sexual harassment was depicted. The reliability of the scale, estimated by Cronbach’s alpha, was .71.

A 2 x 2 x 2 x 3, Gender x Victim Status x School Status x (Level of Scenario) analysis of variance was conducted. The results of the ANOVA are presented in Table 1 below (see Table 1). There was a trend for males to rate the scenarios less severely than females (F = 3.39, p = .07). Prior personal experience with sexual harassment interacted with gender (F = 3.90, p = .05) as follows. Among those who had never experienced sexual harassment, men rated the scenarios significantly less severely than women (F = 13.22, p = .0004). There was no difference in the ratings of men and women who had experienced sexual harassment (F = .006, p = .94). (See Figure 2)

The mean accuracy of the rating of the mild, moderate, and severe scenarios was computed. The difference between the participants labeling of the scenario from the actual label for the scenario was squared, then averaged, within categories. Thus, if a subject rated the severe scenario assessing sexual bribery as mild, we subtracted the actual rating (3) from the participant’s rating (1), and squared the difference. We then averaged the squared difference scores for each scenario in the severe category to obtain the subject’s accuracy score for the severe category. A 2 x 2 x 2 x 3, Gender x Victim Status x School Status x (Level of Scenario) analysis of variance was conducted on these mean squared error scores. The results of the ANOVA are presented in Table 2 below (see Table 2).

Women categorized the scenarios significantly more accurately than men (F = 3.95, p = .05). A planned comparison was conducted of men and women at each level of the victim status. Among those who had never experienced sexual harassment, women categorized the scenarios significantly more accurately than men (F =10.37, p = .002). There was no difference in the accuracy of classification of men and women who had experienced sexual harassment (F = .12, p = .73). However, the overall interaction between gender and victim status was not significant (F = 1.93, p = .17).  (See Figure 3)

Although the interaction between gender, victim status, and scenario severity was not significant (F = 1.46, p = .23), planned comparisons were conducted. There was never a significant difference in accuracy between men and women who had experienced sexual harassment (mild scenarios, F = .57, p = .45; moderate scenarios, F = .35, p = .56; severe scenarios, F = .04, p = .84). When rating mild scenarios, there was no significant difference in accuracy between men and women who had never experienced sexual harassment (F = .02, p = .89). Of those who had never experienced sexual harassment, men were significantly less accurate than women when rating the moderate and severe scenarios (F = 10.24, p = .002 and F = 5.84, p = .02, respectively). (See Figure 4)

This study extends earlier studies by finding that of those who have not experienced sexual harassment, women categorize moderate and severe scenarios, but not mild ones, more accurately than men. This suggests that when discussing prior studies on sexual harassment, differences in results across studies may be explained by the different levels of sexual harassment used by different researchers. The results are also consistent with those of earlier studies and adds support to the attribution hypothesis. Women are more likely to say a behavior is sexual harassment than are men. Male and female victims of sexual harassment do not differ in the severity of rating.

One limitation of this study which limits the generalizability of the findings is the paucity of male sexual harassment victims in the sample. Although significant differences were found in the expected direction, it would be prudent to consider them as suggestive and conduct further research with a larger sample. If this study is representative of the population from which we drew participants, a sizable effort would need to be put forth to identify and recruit males who have been victims of sexual harassment as less than nine percent of the males in our sample reported having been sexually harassed.

The results of the study have important implications to our current legal system. Prior experience with sexual harassment typically disqualifies potential jurors from serving in sexual harassment trials during voir dire. We have some evidence, however, that male victims were significantly more accurate than male non victims in identifying moderate and severe sexual harassment – the type of sexual harassment that may proceed to a jury trial. Jury panels composed of non-victims may be more likely to determine that sexual harassment did not occur when it did. In addition, males and female non-victims serving on jury panels are more likely to have conflicting conclusions about whether or not sexual harassment did occur in any given trial.

Adams, J.W., Kottke, J.L., & Padgitt, J.S. (1993). Sexual harassment of university students. Journal of College Student Personnel, 24 (6), 484-490.

Fitzgerald , L.F. & Hesson-McInnis, M. (1989). The dimensions of sexual harassment: A structural analysis. Journal of Vocational Behavior, 35 (3), 309-326.

Konrad, A.M. & Gutek, B.A. (1986). Impact of work experiences on attitudes towards sexual harassment. Administrative Science Quarterly, 31 (3), 422-438.

McKinney, K. (1992). Contrapower sexual harassment: The effects of students and type of behavior on faulty perceptions. Sex Roles, 27 (11-12), 627-664.

McKinney, K. (1990). Sexual harassment of university faculty by colleagues and students. Sex Roles, 23 (7-8), 421-438.

Pryor, J.B. & Day, J.D. (1988). Interpretations of sexual harassment: An attributional analysis. Sex Roles, 18 (7-8), 405-417.

Rigor, S. (1991). Gender Dilemmas in sexual harassment policies and procedures. American Psychologist, 46 (5), 497-505.

Table 1. The Effect of Gender, Victim Status, School Status, and Level of Scenario on Estimates of the Level of Sexual Harassment (Return to body of document)

Table 2. The Effect of Gender, Victim Status, School Status, and Level of Scenario on the Accuracy of Estimates of the Level of Sexual Harassment in a Scenario (Return to body of document)

Figure 1. Percent of Victims Per Category (Return to Body of Document)

Figure 2. Mean Sexual Harassment Score by Gender and Victim Status (Return to body of document)

Figure 3. Mean Square Rating Error by Gender and Victim Status (Return to body of document)

Figure 4. Mean Square Rating Error by Gender and Victim Status and Scenario Severity (Return to body of document)

Poster session presented at the 1998 Annual Meeting of the Eastern Psychological Association, Boston, MA. Please direct any inquiries to Roxanne.