Workplace Harassment Patterning, Gender, and Utilization of Professional Services: Findings from a US National Study

Abstract

This study constitutes the first national longitudinal survey to address the relationship between workplace harassment and service utilization. We examine how patterns of sexual harassment and generalized workplace harassment are linked to utilization of mental health, health, legal, spiritual, and work-related services, and whether and how gender influences these relationships. Data derive from a random digit dial telephone survey with a continental U.S. sample of employed adults. Eligibility criteria were being 18 years of age or over, and being employed at least 20 hours per week at some time in the 12 months prior to the wave 1 survey. Out of 4,116 households with eligible individuals, 2,151 agreed to participate at wave 1. 1,418 participated at wave 2, thus, the overall response rate was 34.5%. We show that the patterning of workplace harassment over two time points (chronic, remission, onset, never harassed) is associated with the use of different types of services. Gender partially moderated the relationship between workplace harassment and services.

One major source of life stress derives from one’s work environment, resulting in negative physical and mental health outcomes (e.g., Karasek & Theorell, 1990). Consequently, job stress researchers have paid increasing attention to the effects of workplace stressors and how individuals cope with these experiences (e.g., Latack & Havlovic, 1992). Most studies have focused on individuals’ utilization of informal resources, such as drawing support from family, friends, and co-workers. By contrast, there is limited knowledge of the extent to which individuals utilize more formal resources such as professional services.

This study derives from the stress and coping paradigm and focuses on social stress in the work domain and its relationship with the use of professional services to deal with workplace problems. In contrast to most studies which focus on task and role-related strains experienced at work (e.g., Larsson & Setterlind, 1990), this study centers on interpersonal workplace conflict involving sexual harassment and generalized workplace harassment. Both forms of workplace harassment have been linked to adverse mental health and job outcomes (Fitzgerald, Drasgow, Hulin, Gelfand, & Magley, 1997; Richman, Rospenda, Nawyn, & Flaherty, 1997; Rospenda, 1998) and are prevalent across occupational and social groups (e.g., Richman et al., 1999). However, the extent to which targets use professional services in response to workplace harassment and the types of services used have received limited attention (the term ‘target’ is preferred to ‘victim’ as the former term is felt to be more empowering for individuals because it allocates responsibility for the offensive behaviors to the perpetrator). The few existing studies on this relationship (Rospenda, 2002; Rospenda, Richman, & Shannon, in press) suggest linkages between interpersonal work stress and increased concurrent and future services usage. This research, however, is limited by its focus on a sample of university employees. Since professional resources are important for managing unpleasant work conditions (Knapp, Faley, Ekeberg, & DuBois, 1997), we build on Rospenda’s (2002) previous work by examining this relationship in a national longitudinal sample of working individuals. We examine 1) the relationship between workplace harassment (sexual and generalized) patterns across two waves of data collection and the use of different types of professional services at wave 2, and 2) whether and how gender influences the relationships among sexual harassment, generalized harassment and services usage.

Workplace Harassment: A Pathogenic Interpersonal Stressor

Workplace harassment encompasses varied forms of discrimination or interpersonal mistreatment that arise from different sources in the workplace (e.g., co-workers, supervisors, customers) and is experienced through direct or passive offenses to the targets. Although definitions of harassment vary, typical explanations refer to offensive behaviors that are unsolicited and unwelcome (Keashly & Jagatic, 2003; Sbraga & O’Donohue, 2000).

Sexual harassment, or unwelcome and offensive behaviors based on one’s sex, is the most recognized form of workplace harassment. In the U.S., sexual harassment is an illegal form of sex discrimination characterized by “unwelcome sexual advances, requests for sexual favors and other verbal or physical conduct of a sexual nature” which affects the terms, conditions, or employment decisions related to an individual’s job (“quid pro quo” harassment) or creates an “intimidating, hostile, or offensive working environment” (“hostile environment” harassment; Equal Employment Opportunity Commission (EEOC), 1980, 29 C.F.R. § 1604.11). It is estimated that more than half of women experience sexual harassment -type behaviors on the job (Fitzgerald et al., 1997; Ilies, Hauserman, Schwochau & Stibal, 2003) compared to between 14% and 19% of men (U.S. Merit Systems Protection Board (USMSPB), 1995). Experiences of sexual harassment tend to be less prevalent in European workforces than in American, especially where women have higher levels of social equity (Gruber, 1997). While most studies of sexual harassment have focused exclusively on women, among men it is becoming a more visible issue, as evidenced by the growth in claims filed with the EEOC by men from 9% in 1992 to 15% in 2004 (EEOC, 2005).

In contrast to sexual harassment, no U.S. laws currently prohibit more generic forms of interpersonal workplace problems such as generalized harassment. This type of harassment has been studied under various labels which differ in emphasis on aspects of the behaviors in question (i.e., duration, motivation, or power differences; see Keashly and Jagatic, 2003 for a review of the various labels and definitions). For example, European perspectives center on the concept of bullying which emphasizes the process whereby experiences of repeated offensive behaviors place targets in a powerless position (Einarsen, Hoel, Zapf & Cooper, 2003). We use the term generalized workplace harassment to describe offensive behaviors that encompass verbal aggression, disrespectful or exclusionary behavior, isolation/exclusion, threats or bribes, and physical aggression, without explicit reference to duration of experiences, perpetrator motivation, or the power relationship between perpetrator and target and that are not obviously related to legally protected characteristics (e.g., gender, race/ethnicity, age, disability). Prevalence rates of non-sexual hostile workplace behaviors have been found to be 30% or higher for men and 55% or higher for women (Björkqvist, Österman, & Hjelt-Bäck, 1994; Keashly, Trott & MacLean, 1994; Rospenda, 2002).

Workplace harassment remains a salient workplace problem for both genders. Sexual harassment negatively influences various mental health outcomes such as depression, anxiety, and hostility (Richman, Flaherty, Rospenda & Christensen, 1992) and alcohol and drug use (Richman et al., 1999). Similarly, generalized harassment has been linked to psychological distress (Marchand, Demers & Durand, 2005), and the use and abuse of alcohol and other drugs (Richman et al., 1999; Rospenda, 1998). The negative effects of harassment on mental health and job outcomes persist after controlling for other forms of job stress (Rospenda, Richman, Wislar, & Flaherty, 2000), suggesting that it is a unique work-related stressor.

The Patterning of Workplace Harassment

The negative effects of harassment are linked to the persistent nature of its occurrence, such as its timing, duration and/or severity. These factors condition the ways stressors are experienced and how they affect outcomes (Wheaton, 1997). Particularly, the negative effects of acute stressors with discrete onset and offset points may be limited compared to chronic stressors that persist over time. Chronic stressors have more deleterious effects on health and well-being (Avison & Turner, 1988). The persistence of unsolved issues contributes to higher levels of distress (Wheaton, 1997), and also brings about co-occurring problems that perpetuate the nature and severity of the stressor experienced (e.g., Richman at al., 1997). However, the extent to which the pattern of harassment over time conditions responses to it has been largely understudied due to a lack of longitudinal research. Thus, the present study examines the effects of sexual and generalized harassment patterns on the utilization of professional services. We conceptualize patterns of harassment within a gradient of severity ranging from never harassed (least severe) to chronic harassment (most severe). Since one of the strongest predictors of services use is degree of distress (Cooper-Patrick et al., 1999) and chronic stress is associated with higher levels of distress (Wheaton, 1997), we propose that longer duration of harassment will be associated with professional services utilization.

Coping Responses to Workplace Harassment

Studies on workplace harassment and coping show that targets rely on various resources but most often manage their experiences by trying to avoid and ignore the offensive behaviors (Gruber & Smith, 1995; Keashly et al., 1994). Targets also rely on different forms of social support from friends, family and co-workers (e.g., Richman, Rospenda, Flaherty, & Freels, 2001), including support from more formal associations (e.g., community or religious organizations).

Coping efforts have also been differentiated by their focus. Fitzgerald, Hulin, and Drasgow (1995) proposed that targets of sexual harassment use external strategies (behavioral efforts to solve the problem) or internal strategies (cognitive efforts to manage emotions related to the situation). Subsequent models have encompassed both the focus of response (self or initiator focused) and mode of response (self or supported responses) (Knapp et al., 1997). The focus of coping may be either the self (avoiding/denying behaviors or seeking support from friends/family) or the harasser/initiator (confronting one’s offender or utilizing organizational support). The mode may be a supported response where targets seek outside assistance or a self response where no outside help is used. More limited research has examined how targets of harassment deal with generalized harassment, but there is anecdotal evidence that the most frequent coping response has been to do nothing (Cox, 1991; Keashly et al., 1994).

Research on the links between workplace harassment and choice of specific coping strategies has primarily examined whether targets use internal or external strategies or whether targets confront or file complaints against their perpetrators. However, Knapp and colleagues (1997) view coping as a process whereby targets move from individual coping efforts to the use of supportive resources when the harassment continues and becomes more severe. We might expect targets to use professional services once they have exhausted informal resources. We are not able to test this theory directly in this study. However, given the importance of social support for dealing with life stressors (Thoits, 1995), it is plausible that support-related responses involving professional services will be beneficial to targets whether the coping efforts are self or initiator focused. Professional support may help targets cope with the consequences of harassment (in the case of counseling or mental health services) or help them confront and challenge their perpetrators (in the case of using human resource policies or gaining legal advocacy).

The use of professional services as a coping strategy, however, has received little attention in the workplace harassment literature. We argue that professional consultation may equip targets with the tools needed to effectively manage stressful workplace experiences. First, professionals can help to identify and legitimate the problem by labeling the experiences as harassment. Second, professionals may be knowledgeable about structured solutions for harassment (e.g., mediation or filing an EEOC complaint). Third, professional help seeking may instill or maintain a sense of control over one’s surroundings – a quality that has been linked with optimal health and well-being (Pearlin, Lieberman, Menaghan, & Mullan, 1981). Fourth, professional help seeking may help individuals to better manage their emotions and perspectives surrounding the stressful situations at work. Finally, the use of certain types of professional services may also lead to cessation of the problem.

The Role of Gender

Choice of coping responses may vary by gender, as gender influences the level of harassment experienced and the ways in which it is appraised. In the case of sexual harassment, women perceive a broader range of behaviors as offensive (Rotundo, Nguyen, Sackett, 2001), and report harassment at higher levels than do men (Magley, Hulin, Fitzgerald, & DeNardo, 1999). Although some research has shown that women and men may be equally likely to experience sexual harassment (Richman et al., 1999; Rospenda, 1998; Rospenda et al., 2000), there is theoretical and empirical support for the principle that women are more threatened by their perpetrators who are often men (EEOC, 1980), and who have greater social status and power compared to women (Cleveland & Kerst, 1993). Therefore, women may be harmed by sexual harassment more so than men. While there has been comparatively little research on gender differences in generalized harassment, existing research suggests that women more frequently experience these behaviors (Björkqvist et al., 1994) and experience more chronic forms (Rospenda et al., 2000) compared to men. Given gender differences in the experience and reporting of workplace harassment, it is likely that women will make greater use of professional services than men because these workplace problems are more prevalent and may overwhelm their individual coping capacity.

Gender also influences the ways in which individuals engage in social networks and decisions about what types of professional support to utilize. Societal gender role norms shape behaviors which are differentially deemed acceptable by women versus men. The masculine role embodies mastery and instrumentality, emphasizing self-reliance, while the feminine role embodies emotional sensitivity and expressiveness (Bem, 1981). Thus, women might be more open to receiving professional help, particularly from sources that underscore emotional expressiveness. The use of mental health or spiritual services may be greater among female than male targets of harassment because men may perceive greater stigma associated with acknowledging a problem. In the case of sexual harassment, men might feel reluctant to seek help for what they perceive to be a “women’s issue:” when men perceive problems to be unusual (i.e., non-normative), they are less likely to seek help (Addis & Mahalik, 2003). Thus, gender differences in the utilization of particular services may vary across different contexts.

Overview of the Present Study

Based on the literature outlined above, three hypotheses were tested:

Hypothesis 1

Chronic forms of sexual and generalized harassment will predict services use at wave 2 (W2) compared to no harassment, controlling for life stressors, general job stress, baseline services use and demographic characteristics.

Hypothesis 2

Sexual and generalized harassment will be differentially linked to types of services used, as sexual harassment is a legally constructed social problem as opposed to generalized.

Hypothesis 2a

Sexual harassment will be associated with initiator-focused services addressing the behaviors of the perpetrator (work-related or legal services).

Hypothesis 2b

Generalized harassment will be associated with self-focused services that help targets cope with the consequences of harassment (health, spiritual or mental health services).

Hypothesis 3

Gender will moderate the relationship between harassment and services use, such that harassed women will be more likely to seek services than harassed men. Particularly, women will be more likely than men to use mental health and spiritual services, given greater stigma associated with both harassment and use of emotionally based services among men.

Method

Sample

Data derive from a random digit dial telephone survey conducted by the second and third authors of this article and examined the prevalence and health and mental health outcomes of different forms of harassment, and the use of and satisfaction with professional services in a national (continental U.S.) sample of employed adults. The study population was drawn from households using the Troldahl-Carter-Bryant method of respondent selection (Bryant 1975, Troldahl & Carter 1964). Eligibility criteria were a) being 18 years of age or over, and b) being employed at least 20 hours per week at some time in the 12 months prior to the wave 1 (W1) survey. Ineligible (e.g., due to age, work status, nonworking numbers, nonresidential business, or foreign language) and unscreened (refusals, non-contacts) respondents were excluded from the sample. Out of 4,116 households with eligible individuals, 2,151 agreed to participate at W1 (52.3% response rate; 1,067 women and 1,083 men, with 1 person not specifying gender) and 1,418 participated at wave 2 (W2) (66% retention rate; 722 women and 696 men). Thus, the overall response rate was 34.5%.

The study was approved by the university Institutional Review Board. Pretest and main study interviews were completed in English or Spanish. Advance letters describing the study and information about who to call for further information were also sent to all cases with a listed address at both waves. W1 interviews were conducted from August 2003–February 2004, and respondents were re-interviewed one year later (W2). Respondents were paid $10 for participation at W1 and $20 at W2. Interviewing took place mostly on weekday evenings and weekends to increase the probability of successful contact and averaged 30 minutes in length. Twenty contact attempts were made, at different times and intervals, before determining a case as a noncontact. In the case of refusals, two callbacks were made by interviewers experienced at refusal conversion before finalizing a case as refusal.

Ninety-two percent of respondents were employed at both waves of data collection and worked an average of 41 hours per week (S.D. _W2 = 11). About half of study respondents are female and half are male with an average age of 43 years old (SD =12). The regional distribution was as follows: Northeast = 18%; South = 35%; Midwest = 25%; West Coast = 22%. The largest group, non-Hispanic Whites, represent 78% of study participants. The Black or African American population represents 9% of respondents, Asians represent 2%, and Latinos represent 7%, while those who report belonging to some other racial category represent 5% of study respondents.

Measures

Sexual Harassment (SH) was measured with a shortened and modified form of the Sexual Experiences Questionnaire (SEQ; Fitzgerald et al., 1988). The SEQ is a 19-item measure of three components of sexual harassment: gender harassment (comments or behaviors that demean the target’s gender), unwanted sexual attention (e.g., unwanted sexual touching, requests for dates), and sexual coercion (requests or demands for sexual favors that imply job-related consequences), and one item that measures sexual assault (Gelfand, Fitzgerald, & Drasgow, 1995). Nine items were selected (three items from each of the three components, α _W1 = .77 and α_W2 =.73) based on item-total correlations in the dataset reported in Richman et al. (2001). Respondents were asked about their experiences in the past 12 months and items were made applicable to both women and men.

Generalized Workplace Harassment (GWH) was measured by a shortened version of the 29-item Generalized Workplace Harassment Questionnaire, which was developed to assess general, non-sexual harassing experiences at work. It assesses five conceptual dimensions of generalized workplace harassment: verbal aggression (e.g., hostile verbal exchanges characterized by yelling or swearing), disrespectful behavior (e.g., demeaning experiences such as being publicly humiliated or talked down to), isolation/exclusion (e.g., having one’s work contributions ignored or being isolated from important work activities), threats/bribes (e.g., subtle or obvious threats or bribes to do things that are wrong), and physical aggression (e.g., being hit, pushed, or grabbed) (Rospenda & Richman, 2004). Eleven items were selected for inclusion in the present study (α_W1 =.85 and α _W2 = .83) using the same methods employed to create the modified Sexual Harassment scale.

Prior research demonstrates that the two questionnaire scales represent separate but related constructs (Fendrich, Woodward, & Richman, 2002). To examine whether the shortened versions of these scales performed similarly, confirmatory factor analysis was run using AMOS 5.0 (analyses not shown). Results were consistent with Fendrich et al. (2002), indicating that the reduced-item versions of the two questionnaires represent distinct, but correlated, constructs (Correlation =.58; model root mean squared error of approximation [RMSEA] = .047; 90% CI = .04–.05).

For each item, respondents indicated whether the experience occurred on their current job in the past 12 months: (1) never, (2) once, or (3) more than once. We counted a respondent as “harassed” (scored as 1) for a given wave if the respondent indicated experiencing any one of the behaviors “more than once”, or experiencing sexual coercion (for sexual harassment) or physical aggression (for generalized harassment) at least “once” (given their severity). Otherwise, they were scored as 0 (not harassed). These classifications were guided by evidence from stress research showing that acute and violent stressful events can be just as damaging as experiencing ongoing stress (Wheaton, 1997). For the purposes of this paper, respondents were then categorized into one of four separate harassment patterns for each type: (1) no harassment at waves 1 or 2, (2) harassment at W1 only (remission) (3) harassment at W2 only (onset), and (4) harassment at both waves (chronicity). For each type of harassment, the pattern was represented by three dummy variables in the analysis: chronicity, remission, and onset (no harassment served as the reference category). As the pattern variables are nominal and descriptive in nature, internal consistency estimates are not appropriate and were not calculated.

Use of Services

Respondents were asked at W1 to indicate whether they had seen any of the following professionals in the past 12 months: primary care physician, psychiatrist, psychologist, social worker, other health care professional, or non-health care professionals, such as a clergy person, legal professional or work-related professional. The questions at W2 were altered to ask specifically about use of services to deal with work stress. An indicator variable was created for overall use of services, coded 1 if respondents used any services and 0 if the respondent used no services. We also computed separate indicators for use of mental health services (saw psychologist, social worker, or psychiatrist), health services (primary care physician, or other health services professional), legal services (saw lawyer or legal professional), spiritual services (sought help from a clergy person) or work-related programs (e.g., affirmative action office, human resources, EAP, union). These indicator variables are nominal; therefore, internal consistency values do not apply.

Job Stressors

Job pressure and job threat were measured by a shortened 7-item version of the Stress in General Scale (Stanton, Balzer, Smith, Parra, & Ironson, 2001). The scale was developed for measuring the general experience of workplace stress so that it could be applicable across professions, industries, and cultures. Job pressure (3 items; α _W1 = .66; α _W2 =.65) assesses the extent to which one’s job is seen as pressured, hectic or relaxed and represents a sense of time pressure, which accounts for a substantial amount of general job stress. Job threat (4 items; α_W1 = .65; α _W2 = .64) measures the extent to which one’s job is seen as under control, nerve-wracking, hassled or smooth-running, reflecting an overall threatening or negative quality of the job. Response categories were “yes”, “no”, and “can’t decide”. Scales were computed such that higher numbers represent higher levels of job pressure and threat. An indicator for change in job stressors between waves 1 and 2 was also computed to capture the effects of change in level of job stress on services use.

Life Stressors

Life stress was assessed using the 12-item List of Threatening Experiences, an inventory of common stressful life events rated for long-term contextual threat (Brugha & Cragg, 1990). Included in the list are questions about past year illness, injury, separation from a significant other, friend, or family member through divorce or death, financial difficulties, and theft, among other stressful life events. Respondents indicated either “yes” or “no” to the life event specified, and responses were summed to create an overall indicator of life stress. The List has been demonstrated to be a reliable and valid instrument (Brugha, Bebbington, Tennant, Hurry, 1985; Brugha & Cragg, 1990). As no assumptions are made regarding the inter-correlation among individual life events, they do not meet the criteria for internal consistency statistics and therefore, Cronbach’s alpha values were not calculated (Streiner, 2003). An indicator for change in life stressors between waves 1 and 2 was also computed to capture the effects of change in level of life stress on services use.

Demographic control variables

The effects of the following demographic variables on services use were controlled statistically: age, income, gender (where appropriate), race, education, and baseline services use. Age was scored continuously in years, education was an ordinal scale ranging from 1 (8^th grade or less) to 8 (master’s degree or higher), and income was an ordinal scale ranging from 1 (less than $10,000) to 6 (greater than $70,000). Gender (1= women, 0= men), race (1= White, 0 = all other races) and baseline services (1 = utilized services, 0 = did not utilize services) were coded in a binary format.

In terms of demographic differences between completers and non-completers at W2, non-completers were more likely to be minorities (Black and Hispanic), to be younger and have lower mean income and educational levels than completers (p<.01). There were no differences between completers and non-completers in experienced generalized harassment, levels of job threat, or use of spiritual and work-related programs. Non-completers had higher mean scores for sexual harassment and life stressors and used mental and legal services more than did completers. Also, non-completers had lower mean job pressure scores, and used overall services and health services less frequently than did completers.

Statistical analysis

SPSS for Windows (version 13.0), which assumes simple random sampling when calculating statistics, was used for the analyses. This study utilizes a simple random sample; therefore, variance approximation methods were appropriate. Listwise deletion of missing data resulted in sample sizes which varied slightly by model, ranging from 1196–1199 individuals. Exploratory analyses showed that no more than 5% of values for each of the study variables were missing. Univariate descriptive statistics were calculated and Chi-square tests (for categorical variables) and t-tests (for continuous variables) were used to test whether there were gender differences in study variable characteristics.

To test hypotheses one and two, we computed a series of binary logistic regression analyses to examine the relationships between harassment patterns and use of services across two waves of data collection, controlling for age, income, race, education, job and life stressors and baseline services use. To test hypothesis three, we tested for interactions between harassment patterns and gender on service use. Where the interaction term was not significant, gender was retained as a control variable. For ease of interpretation, we conducted separate regressions using the sexual and generalized harassment pattern variables, given evidence that they represent separate but related constructs (Fendrich et al., 2002).

Support for hypothesis 1 would be demonstrated by an improvement in model fit upon entry of harassment variables and a main effect for the chronicity term. Support for hypothesis 2a and 2b would be demonstrated by an improvement in model fit upon entry of harassment variables and a main effect of harassment patterns on particular services, such that sexual patterns are linked to work-related or legal services, while generalized patterns are linked to health, spiritual or mental health services. Support for hypothesis 3 would be indicated by an improvement in model fit upon entry of the harassment by gender interaction term, and a significant harassment by gender interaction regression coefficient, such that harassed women exhibited greater use of services than harassed men. Results from logistic regression models are shown as odds ratios (OR) and 95% confidence intervals (CI). The Hosmer-Lemeshow test (Hosmer & Lemeshow, 1989) was used to assess the fit of the regression models.

All analyses reflect weighted data to reduce biases in estimates. Data were weighted in two stages: first, selection weights were calculated for each case to weight for different probability of selection (based on number of phone lines and number of eligible adults in the household). Second, post-stratification weights were calculated to ensure that the marginal distribution of several sample variables corresponded to 2003 Current Population Survey data. The variables used in the weighting procedure were age, gender, race, Census Bureau region, and education.

Results

Descriptive statistics for all variables are presented in Table 1, overall and separately for men and women. It can be seen that women were, on average, older and that a greater proportion of women than men utilized overall and specific types of services. Consistent with other studies on the relationship between gender and services use (e.g., Green & Pope, 1999; Rhodes, Goering, To, & Williams, 2002), women were more likely than men to report use of any service at both waves 1 and 2. When we examine the types of services utilized by men and women, we find that women were more likely to use mental health and health services at both waves. Women used more spiritual services at W1 and legal and workplace services at W2. Also, sexual harassment patterns were associated with gender, while generalized harassment patterns were not.

Table 1.

Weighted Means and Standard Deviations for Independent and Dependent Variables

	OVERALL		MEN		WOMEN
	n= 1382		n= 733 (53%)		n= 648 (47%)
n (observations)	Mean	S.D.	Mean	S.D.	Mean	S.D.	χ2	p
Sociodemographics
Age	43.17	12.48	42.69	12.40	43.70	12.55	-
Income
< $30,000	.21	-	.19	-	.23	-
$30,001 to $70,000	.47	-	.45	-	.49	-
> $70,000	.32	-	.36	-	.29	-	7.63	*
Race
Whites	.22	-	.77	-	.78	-
Non-Whites	.78	-	.23	-	.22	-
Education
High school graduate or lower	.38	-	.41	-	.36	-
Some college, Associate or Certificate	.30	-	.26	-	.34	-
Bachelor degree	.19	-	.20	-	.18	-
Some graduate work, Master's degree or higher	.13	-	.13	-	.13	-	10.32	*
Services Utilization
Overall
W1	.67	-	.64	-	.69	-	4.10	*
W2	.18	-	.15	-	.21	-	10.83	***
Mental Health
W1	.07	-	.05	-	.10	-	11.82	***
W2	.06	-	.04	-	.08	-	7.52	**
Health
W1	.61	-	.58	-	.64	-	6.14	*
W2	.12	-	.09	-	.15	-	9.02	**
Legal
W1	.11	-	.10	-	.13	-	2.47
W2	.01	-	.004	-	.02	-	5.67	*
Spiritual
W1	.07	-	.06	-	.08	-	3.21
W2	.03	-	.04	-	.03	-	.55
Work-Related
W1	.03	-	.04	-	.03	-	2.41
W2	.02	-	.01	-	.03	-	5.01	*
Life Stressors
W1	1.50	1.59	1.41	1.51	1.62	1.68	-	*
W2	1.39	1.44	1.32	1.43	1.47	1.44	-
Job Stressors
Job Pressure
W1	6.30	3.10	6.29	3.02	6.30	3.18	-
W2	6.15	3.16	6.02	3.18	6.29	3.13	-
Job Threat
W1	3.99	3.72	3.81	3.54	4.19	3.91	-
W2	3.56	3.60	3.43	3.45	3.70	3.76	-
Patterns of SH
None	.55	-	.56	-	.54	-
Chronic	.19	-	.18	-	.20	-
Remission	.15	-	.14	-	.17	-
Onset	.11	-	.13	-	.09	-	8.76	*
Patterns of GWH
None	.45	-	.44	-	.47	-
Chronic	.27	-	.27	-	.26	-
Remission	.16	-	.17	-	.13	-
Onset	.13	-	.12	-	.14	-	5.55

Note: Means for categorical variables represent proportion in that category; S.D. = standard deviation; χ² = chi-square statistic;

^*p<.05;

^**p<.01;

^***p<.001

Table 2 shows that those who experienced chronic forms of sexual harassment and those who experienced onset at W2 were significantly more likely than those who were not harassed to utilize legal and spiritual services, adjusting for control variables. Also, those who experienced remission of sexual harassment at W2 were more likely than those who were not harassed to utilize work-related services at W2. Thus, we obtain partial support for the hypothesis that chronic sexual harassment is linked to an increased odds of services use and that harassment patterns are linked to initiator-focused services.

Table 2.

Logistic Regression Predicting Services Use at Wave 2 (W2) from Sexual Harassment (SH) Patterns

	Overall (n=1196)		Mental Health (n=1198)		Health (n=1199)		Legal (n=1198)		Spiritual (n=1198)		Work-Related (n=1198)
Independent Variable	OR	95% C.I.	OR	95% C.I.	OR	95% C.I.	OR	95% C.I.	OR	95% C.I.	OR	95% C.I.
Chronic	1.45	.94–2.23	1.72	.81–3.64	.82	.48–1.41	9.17	1.23–68.52 *	5.27	1.98–14.03 **	1.37	.38–4.98
Remission	1.16	.72–1.89	.39	.12–1.24	.92	.51–1.66	1.55	.07–34.07	1.68	.50–5.67	4.46	1.58–12.60 **
Onset	1.08	.63–1.87	.98	.35–2.72	.99	.52–1.88	8.23	1.17–57.76 *	3.34	1.00–11.20 *	1.83	.44–7.72

Note. All OR in the table are the odds of the services use outcome among people who were harassed compared to those who were not harassed adjusted for age, income, gender, race, education, W1 services use, life stressors, change in life stressors, job stressors, and change in job stressors.

^*p < .05;

^**p < .01;

^***p < .001

Tables 3 and 4 present results for regressions predicting services use from generalized harassment patterns. Unlike sexual harassment, the relationship of generalized harassment to services use was moderated by gender in the case of overall and mental health services (Table 3), partially confirming hypothesis 3 regarding the differential gender effects of harassment on services use. Specifically, we find gender differences in the effect of generalized harassment patterns on service utilization. This is evidenced by a significant gender by chronicity interaction term once sociodemographics, baseline service use, and other forms of life and job stress were controlled. To examine the form of the interactions, we computed odds ratios for specific harassment patterns given either male or female gender, following the procedures outlined by Hosmer and Lemeshow (1989). In terms of overall services, we focused on those in the chronically harassed group, as only the gender by chronicity interaction term was significant. We found that women who are chronically harassed are more likely than women who were never harassed to seek services (OR = 1.94, 95% CI = 1.10– 3.42). For men, chronic generalized harassment did not result in increased odds of services compared to those who were never harassed. Regarding between- gender comparisons, chronically harassed women exhibited an increased odds of services use compared to chronically harassed men (OR = 3.74, 95% CI = 2.02– 6.93).

Table 3.

Logistic Regression Predicting Overall and Mental Health Services Use at Wave 2 (W2) from Generalized Workplace Harassment (GWH) Patterns

	Overall (n=1196)		Mental Health (n=1198)
Independent Variable	OR	95% C.I.	OR	95% C.I.
Gender* Chronic	2.95	1.28–6.80 *	7.67	1.03–57.11*
Gender* Remission	.66	.23–1.92	.26	.03–2.59
Gender* Onset	.39	.14–1.06	.68	.13–3.58

Note. All OR in the table are the odds of the services use outcome among people who were harassed compared to those who were not harassed adjusted for age, income, gender, race, education, W1 services use, life stressors, change in life stressors, job stressors, and change in job stressors.

^*p < .05;

^**p < .01;

^*** p < .001

Table 4.

Logistic Regression Predicting Services Use at Wave 2 (W2) from Generalized Workplace Harassment (GWH) Patterns

	Health (n=1199)		Legal (n=1198)		Spiritual (n=1198)		Work-Related (n=1198)
Independent Variable	OR	95% C.I.	OR	95% C.I.	OR	95% C.I.	OR	95% C.I.
Chronic	1.16	.68–1.97	.39	.05–3.05	1.80	.65–5.00	2.31	.74–7.19
Remission	.94	.49–1.82	.14	.01–2.69	1.48	.44–4.97	.87	.19–3.93
Onset	1.56	.84–2.87	.52	.07–3.97	4.00	1.37–11.67 *	2.8	.81–9.57

Note. All OR in the table are the odds of the services use outcome among people who were harassed compared to those who were not harassed adjusted for age, income, gender, race, education, W1 services use, life stressors, change in life stressors, job stressors, and change in job stressors.

^*p < .05;

^**p < .01;

^***p < .001

In terms of mental health services, we again focused on the gender by chronicity interaction. The results of our analysis were counter to our expectations. Specifically, men who were chronically harassed were less likely to use mental health services than men who were never harassed (OR = .14, 95% CI = .02– .93). When comparing men and women, we found that women who experienced chronic generalized harassment had a greatly increased odds of use of mental health services compared to men who were chronically harassed (OR= 10.57, 95% CI = 1.78– 62.68).

The analyses presented in Table 4 represent regressions where gender was used as a control variable, since initial tests showed that gender did not moderate the relationship between generalized harassment patterns and health, legal, spiritual and work-related program services. Those who experienced onset W2 were significantly more likely than those who were not harassed to use spiritual services at W2. Overall, life stressors were the best predictors of use of all types of services (data not shown), with the exception of spiritual and work-related services.

Discussion

This study examined the extent to which workplace harassment targets use professional services and how the pattern of their experiences influences professional support seeking over time. We also examined the role that gender plays in these relationships. In contrast to previous research on current and former university employees (Rospenda et al., 2000; Rospenda et al., in press), we did not find that women were more likely to experience chronic generalized harassment. Unlike the aforementioned studies, we found that chronic sexual harassment was more common among women.

The finding that women were more likely than men to report use of any service during the time frame of the study is consistent with other findings on gender differences in services utilization. We also found that women were more likely than men to have sought professional help from sources that could be considered self-focused (e.g., services that help individuals find relief related to the self).

Consistent with our previous work (Rospenda, 2002; Rospenda et al., in press), we found that both forms of harassment were associated with increased odds of services use. The present study extends this work by demonstrating in a national sample that different types of harassment are linked to different types of services use. Sexual harassment was associated with legal, spiritual and work-related program services, while generalized harassment was associated with overall, mental health, and spiritual services. Additionally, the present study extends prior research by considering the effects of other life stressors in addition to workplace harassment and job stress. This clarifies the relative importance of different types of stressors in predicting use of services. In most models, life stressors significantly contributed to model fit, suggesting that general life stress is more consistently related to the use of the different types of services. However, workplace harassment was linked to services use in the context of controls of life and job stressors. Thus, workplace harassment is sufficiently taxing to lead to professional help seeking. With the exception of health services, however, the fairly low overall prevalence of services use suggests that workers generally remain reluctant to seek these services. This may be true because workers, whether harassed or not, may try to minimize attention to their problems, hoping that they will go away on their own. Second, fears of retaliation leading to job loss may deter professional help-seeking. Third, workers may feel that the help available from professionals will not change the offensive work environment. Lastly, it is unknown to what extent help seeking from informal support systems encourages or discourages the use of professional services. Future studies should examine the conditions under which harassment and other job stressors lead to services use and whether particular types of services are viewed as helpful.

In terms of gender, with the exception of the effect of generalized chronicity on overall and mental health services, we found little evidence of differential use of services among male and female targets. The effect of harassment patterns on overall and mental health services, however, is greater for women than men. We found that men who experienced chronic generalized harassment were less likely than men who were never harassed to seek mental health services. Prior studies show that men, compared to women, are reluctant to seek mental health services for a variety of problems (e.g., Kessler, Brown, & Boman, 1981; Rhodes et al., 2002). It is posited that masculine gender role norms shape the contexts of men’s help seeking behaviors. Thus, help seeking is a function of whether particular problems are viewed as normative for one’s social group, are seen as central to one’s ego, and are within one’s domain of personal control (Addis & Mahalik, 2003). Men are less likely to use mental health services in general and may feel that they would be doubly stigmatized in acknowledging both being a victim of chronic generalized workplace harassment as well as needing mental health services.

Findings from the logistic regression analyses suggest the importance of considering patterns of harassment across time, as onset and chronicity of both types of harassment were most consistently related to a variety of types of services use. The models that show a significant and unique effect of harassment on services use suggest that acute and chronic sexual harassment lead targets to seek help from formal resources that may invoke policies to address harassment in the workplace. Chronic sexual harassment appears to influence decisions to seek support from sources that may help one take action against perpetrators and also help deal with the harassment emotionally. However, remission of sexual harassment was also linked to increased odds of work-related services use, indicating that the need for services does not end with the harassment. It is also plausible that the use of work-related programs may give workers the hope of effectively countering or preventing victimization. The experience of chronic generalized harassment was linked to a greater likelihood of services use particularly for women. However, the types of services used were more likely to be from self-focused sources such as a clergy person or mental health professional. The odds of spiritual services use were increased with the onset of generalized harassment at time 2, suggesting that when organizational policies prohibiting harassment are not in place, employees may not turn to the organization for help. Development of official policies prohibiting harassment might promote help seeking from sources better able to end the harassment (e.g., human resources).

Finally, our results suggest that sexual patterns were linked with initiator-focused coping responses while generalized patterns were not. This is likely related to the fact that generalized harassment, in the U.S., is not recognized as a serious social problem meriting policy initiatives, compared to sexual harassment, which is legally prohibited. The finding that generalized harassment patterns were associated with more emotionally and spiritually based support suggests that targets perceive fewer coping options when dealing with workplace problems that do not involve illegal discrimination. Services use may therefore be different in countries that do systematically prohibit generalized harassment.

Limitations of the study

Despite this research contribution, the lower than desired initial and succeeding response rates limit the generalizability of the findings. Levels of non-response should be considered when interpreting these findings. First, we know little about respondents’ reason for refusals, unanswered calls, or item nonresponse. Given the rigorous data collection procedures employed, it is likely that our study was affected by the challenges common in telephone interviewing. The rise in use of call screening devices and answering machines and the increased aversion to telephone surveys (partly due to the proliferation of telephone marketing and concerns for identity theft) have produced sharp declines in survey response rates (Hartge, 1999; Curtin, Presser, & Singer, 2005). The presence of these trends in this study was evidenced by respondents citing the “Do Not Call Registry” as a reason to refuse the interview.

Second, there are also concerns regarding the relatively low reliabilities of the study’s job stress variables, indicating that the shortened version of the Stress scale may not have adequately captured varied experiences of job stress, possibly underestimating its true effects on services. Future studies should replicate and extend our analyses by using a more complete measure of job stress, preferably including a variety of specific stressors.

A third issue concerns sample attrition and bias. Time 2 non- responders tend to be of lower social statuses (e.g., minorities, lower levels of education and income, younger ages), also limiting the generalizability of the results. Research shows that these groups are less likely to seek services in general (e.g., Gallo, Marino, Ford, & Anthony, 1995). Thus, the finding that W2 non-completers were less likely than completers to use one or more of any type of service at W1 indicates that W2 non-completers may have felt that the study lacked personal relevance.

These limitations can be partially overcome in future research where special efforts are incorporated to achieve a more representative sample. The low response rates do suggest that multi-methods, rather than relying on random dialing alone, may decrease sample bias. It may also be necessary to over sample to compensate for projected attrition and implement more intensive tracking efforts. Future research should also examine the factors related to the likelihood of nonresponse in studies of this nature, given the sensitive topics addressed and the overall low use of professional services for harassment in general. The duration and severity of one’s experiences, the power relationship between the harasser and target, the mental health status of respondents, social desirability and negative affectivity all are likely to impact reporting. Also, the role that attitudes toward professional help seeking and use of alternate coping efforts (e.g., ignoring offenses or seeking informal support) play in impacting services use remains unexplored.

Due to these limitations, we caution the reader to consider potential biases when interpreting the findings. However, it is unclear what differences would be observed in a survey with significantly higher response rates or at what response rate level nonresponse errors become problematic (see Keeter, Miller, Kohut, Groves, & Presser, 2000 and Curtin et al., 2005 regarding new evidence that lower response rates do not necessarily lead to respective increases in nonresponse bias). These questions are particularly germane since this study constitutes the first national study on workplace harassment and services use among a diverse workforce. Thus, the study’s limitations should not obscure the relative value of the study.