Participants

Study participants were recruited from Amazon Mechanical Turk (MTurk). MTurk is an online labor market that allows individuals to complete brief tasks, or HITs (Human Intelligence Tasks), for pay. MTurk workers are aged at least 18 years and are based in a variety of countries. US MTurk workers are largely representative of the United States general population, but tend to skew younger, be slightly more educated, and are more likely to identify as European-American and Asian-American [29–31]. Approximately 57% MTurk workers report currently being employed, working nearly 37 hours on average per week [31]. Workers also are employed in a variety of industries that are distributed relatively closely to large-scale national, stratified surveys (e.g., the Congressional Cooperative Election Survey; [30]). To ensure data quality, we only recruited MTurk workers who had a HIT acceptance rate (HAR) of greater than or equal to 95%, along with at least 50 approved HITs. These steps help ensure data quality and have been used previously in a number of studies using the MTurk platform (e.g., [31, 32]).

To ensure data integrity, participants were also required to answer no more than one “attention check” question incorrectly (e.g., “I will choose ‘Agree’ if I am paying attention”) or their participation would be discontinued. During initial data collection, participants completed four attention check questions. Additional, more sophisticated attention check questions were systematically added throughout data collection when concerns arose about possible participation of non-human actors that attempt to mimic human responses to survey measures, commonly referred to as “bots.” The majority of the sample was collected using six attention check questions (74.4%; n = 599). The remainder of the sample answered either answered five (13.2%; n = 106) or four (12.4%; n = 100) attention check questions.

A total of 971 participants were recruited and completed the study in full. Inclusion criteria for participation in the current study were an age of at least 18, current full-time employment for at least six months, and residence in the United States. Participants who did not meet these criteria were excluded from the study (n = 10), as well as participants who provided illogical responses (e.g., reporting an age of 30 and also reporting military service during the Vietnam War; n = 54), nonsensical responses (e.g., unrelated content in open-ended responses; n = 11), or “straight-lined” responses (n = 11). Participants were also excluded for failing a manipulation check question regarding an experiment that was included in the study (not reported on in this paper [33]; n = 36), or for failing to report demographic or employment information (n = 44). Of the 971 participants who completed the study, 805 participants provided data that met our data inclusion criteria and were used in data analysis.

Participants in the final sample were an average age of 32.35 years (SD = 10.18), and women and men were approximately equally represented. The majority of participants identified as heterosexual/straight (87%) and White (75.5%). A plurality of participants had a bachelor’s degree (41.2%), with 90.4% of the sample reporting an income range above the federal poverty line. The modal income level for this sample was at the $50,000 and $59,000 range. and the average yearly salary was $47, 908 (SD = $28,052). Participants worked in a range of industries and departments, and the majority were employed in the private for-profit sector (75.8%). The majority of participants identified as non-management, individual contributor employees (57.1%). On average, participants had been at their current place of employment for approximately six and a half years (M = 6.51, SD = 5.48) and worked on average just over 40 hours per week (M = 41.85, SD = 5.35). See Table 1 for full demographic information and Table 2 for full employment information for both the full sample (N = 305) and the subsample used analyses (n = 317). The demographic characteristics of the subsample reporting sexual harassment was relatively comparable compared to the full sample.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographic information of full sample (N = 805) and subset of sample (n = 317) used in analyses.

https://doi.org/10.1371/journal.pone.0278830.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Employment information for full sample (N = 805) and subset of sample (n = 317) used in analyses.

https://doi.org/10.1371/journal.pone.0278830.t002

Workplace sexual harassment

Workplace sexual harassment was assessed using the Sexual Experiences Questionnaire-Department of Defense Workplace Gender Relation Survey (SEQ; [34]). The SEQ is a 23-item measure that assesses experiences of different types of sexual harassment at work during the past year using four subscales. Subscales include: sexist hostility (e.g., “treated you ‘differently’ because of your sex [for example, mistreated, slighted, or ignored you]”); sexual hostility (e.g., “repeatedly told sexual stories or jokes that were offensive to you”); unwanted sexual attention (e.g., “touched you in a way that made you feel uncomfortable”); and sexual coercion (e.g., “treated you badly for refusing to have sex”). In addition to the standard items, we included four additional items that are part of revised versions of the SEQ [34]. Participants responded to all items using a 5-point Likert scale (1 = “Never,” 5 = “Many times”). The SEQ was scored by creating a total summed score of all items and calculating incidence rates (i.e., the percentage of individuals experienced at least one instance of sexual harassment). The SEQ is one of the most widely used measures of sexual harassment [5] and demonstrates good internal consistency [34]. In the current study, this measure demonstrated good internal consistency (α = .97).

Institutional betrayal

Institutional betrayal was assessed with the Institutional Betrayal Questionnaire-Version 2 (IBQ; [21]). The IBQ is a 12-item measure that assesses different experiences of institutional betrayal. Given that experiencing institutional betrayal requires first experiencing an index event (e.g., an employee is sexually harassed by her boss [the index event] and is subsequently punished in some way for reporting the experience [the subsequent institutional betrayal]), participants were presented with the IBQ following completion of the SEQ. Importantly, and given the necessity of experiencing an index event, only participants who endorsed at least one item on the SEQ were shown the IBQ. Example items include: “covering up the experience” and “punishing you in some way for reporting the experience (e.g., loss of privileges of status).” Response options included: “Yes,” “No,” and “Not applicable.” Scores on the IBQ were determined by summing the number of items endorsed with “Yes,” with a theoretical score range of 0–12. As with the SEQ, we calculated incidence rates (i.e., the percentage of individuals who have experienced at least one instance of institutional betrayal).

Institutional courage

Institutional courage was assessed with the Institutional Courage Questionnaire–Specific (ICQ-Specific; [35]). The ICQ-Specific is an 18-item measure designed to assess experiences of institutional courage that was created for the purposes of this study. The ICQ-Specific was presented to participants along with the IBQ following their completion of the SEQ. This measure, like the IBQ, is designed to be completed only by those participants who have experienced workplace sexual harassment. Participants were given the same instructions and response options as the IBQ. Example items include: “creating an environment where continued employment was not more difficult for you than before the experience occurred” and “not covering up the experience.” In line with the IBQ scoring, scores on the ICQ-Specific were determined by summing the number of items endorsed with “Yes,” with a theoretical score range of 0–18. We will also calculate incidence rates for institutional courage (i.e., the percentage of individuals who have experienced at least one instance of institutional courage).

Perceived gender bias

Perceptions of gender bias in participants’ current workplace were assessed using four items [36]. Participants rated their agreement on a 7-point Likert scale (1 = “Strongly disagree,” 7 = “Strongly agree”) with statements about whether men and women are treated equally in the workplace (e.g., “I think women and men are treated the same way at my current place of employment”). Items were averaged to create a composite score that represents participants’ perceptions of gender bias in their current workplace, with higher scores indicating less perceived gender bias. In both the Kaiser and colleagues [36] and another study [33], these four items demonstrated good internal consistency (α = .88 and α = .95, respectively). In the current study, these items demonstrated good internal consistency (α = .94).

Job satisfaction

The Abridged Job in General Scale (aJIG; [37]) is an 8-item scale that measures global job satisfaction. Participants were instructed to consider their jobs in general and select “Yes,” “No,” or “?” for each item. Items are either single words or brief phrases that might describe participants’ current jobs, including: “Good,” “Undesirable,” and “Makes me content.” A composite score is created by summing each of the eight items. The aJIG is an abridged version of the full-length 18-item Job in General Scale developed by Ironson and colleagues [38]. The aJIG demonstrated good internal consistency across three studies during its initial development [37]. In the current study, this measure demonstrated good internal consistency (α = .91).

Organizational commitment

Organizational commitment was assessed used the Organizational Commitment Questionnaire (OCQ; [39]). The OCQ was used to assess affective organizational commitment using the 6-item affective organizational commitment subscale. Affective organizational commitment refers to one’s affective or emotional attachment to and identification with an organization. An example item is “I really feel as if this organization’s problems are my own.”. The measure is scored by taking the average of the each of the six items, and has demonstrated good internal consistency across previous studies [40] and the current study (α = .94).

One-year leaving intentions

Two items from the Staying or Leaving Index (SLI; [41]) were used to assess participants’ one-year intentions to leave their current place of employment. Participants responded to two questions: “How do you rate your chances of still working at your current place of employment?” and “How would you rate your chances of quitting your current place of employment?” Participants responded using a 7-point Likert scale (1 = “Definitely unlikely,” 7 = “Definitely likely”). The SLI demonstrated good internal consistency in past research [41] and the current study (α = .89).

Work withdrawal behaviors

Work withdrawal behaviors were assessed using 19 items from Hanisch and Hulin’s (WJW; [42]) measure of work withdrawal. This measure assesses work withdrawal behaviors, broadly conceptualized as avoidance of or actual disengagement from day-to-day work activities (e.g., “Wandering around looking busy so you do not have to do your work?”). Participants used an 8-point Likert scale (1 = “Never in the past year,” 8 = “More than once in the past week”) to report how frequently they engage in these behaviors. The WJW has demonstrated good internal consistency in previous studies [42, 43]. In the current study, this measure demonstrated good internal consistency (α = .81).

Trust in management

The Trust in Management Scale (TIM; [44]) is a 12-item measure used to assess employee trust in institutional management. Participants were instructed to consider the managers and executives at their workplace and select “Yes,” “No,” or “?” for each item. A composite score is created by summing each of the 12 items. Items are either single words or brief phrases that might describe management and executives in a given organization. Example word or phrases include: “Qualified,” “Concerned for employees’ welfare,” and “Consistent.” The TIM has demonstrated good internal consistency and validity [44]. In the current study, this measure demonstrated good internal consistency (α = .91).

Depression, anxiety, and somatic symptoms

The Patient Health Questionnaire (PHQ; [45]) is a multi-subscale questionnaire that assesses multiple aspects of psychological and physical health. For this study, the depression subscale (PHQ-9), the anxiety subscale (GAD-7), and the somatic symptom subscale (12 items) were used to assess psychological health (depression symptoms and anxiety symptoms) and physical health (somatic symptoms). Participants report how often they experience each symptom using a 3-point Likert scale (0 = “Not bothered at all,” 3 = “Bothered a lot”) within a specified timeframe (within the last two weeks for depression and anxiety symptoms and within the last four weeks for somatic symptoms). A composite score was computed for each subscale by summing all items within each subscale. The PHQ and its subscale have demonstrated good internal consistency among a number of studies [45–47] and has been previously used in research on institutional betrayal [19]. In the current study, these subscales demonstrated good internal consistency (symptoms: α = .88; anxiety symptoms: α = .91; and somatic symptoms: α = .85; depression).

Lifetime trauma history

The Brief Betrayal Trauma Survey-14 (BBTS; [48]) was used to assess lifetime trauma history. The BBTS is a 28-item measure that assesses traumatic experiences, ranging from those that are low in interpersonal betrayal (e.g., “been in a major earthquake, fire, flood, hurricane, or tornado that resulted in significant loss of personal property, serious injury to yourself or a significant other, the death of a significant other, or the fear of your own death.”) to those high in interpersonal betrayal (e.g., “you were made to have some form of sexual contact, such as touching or penetration, by someone with whom you were very close”). Participants indicate how many times (“Never,” “One or two times,” or “More than that”) they have experienced each of the 14 events both before and after the age of 18. Items were scored by binning each item into one of three categories: low, medium, or high betrayal; this yielded three variables that were then used in the analyses.

Employment information

Participants were asked a series of questions about their current place of employment, including number of employees, employment sector (e.g., public vs. private, non-profit vs. for-profit), employment industry, current occupation (e.g., management vs. office and administrative support), department of their position (e.g., accounting vs. IT), job title (e.g., individual contributor/non-management vs. executive/C-suite), hours worked per week, number of years at current place of employment, number of years in current role, and current yearly salary. Questions about number of employees, employment sector, employment industry, and occupation were based on the Congressional Cooperative Election Survey [49], a large-scale national, stratified survey that has been used to compare MTurk participant demographics to those of participants in national stratified surveys [30]. Considering that it was possible that a portion of participants may work for multiple employers, participants were asked about the number of employers for whom they current work. If they indicated they worked for more than one employer, they were asked to choose one as their primary employer to report on.

Demographic information

Participants answered a number of questions about their demographic information, including age, gender identity (both current and as assigned at birth), sexual orientation, race and ethnicity, highest level of schooling, current household income, and country of residence.

Missing data

Data were examined for missingness before proceeding with analyses. Missing data >5% was found on the following outcomes of interest: PHQ Depression Subscale (16.6%), BBTS (15.3%), and PHQ Anxiety Subscale (12.0%). Missing values were imputed using SPSS Version 25 Multiple Imputation using 10 imputed datasets. Measures of lifetime trauma history (BBTS), institutional betrayal (IBQ), and institutional courage (ICQ-Specific) were not imputed. These measures assess the presence or absence of certain experiences and concerns arose about both the lack of empirical support for imputation of these measures and the accuracy of imputation regarding these types of experiences. The majority of the missingness on the BBTS was related to an error in the Qualtrics survey software. At least one participant noted that there were problems with the display of this measure, as it appeared to be too wide on smaller screens. An examination of the missingness patterns of the BBTS revealed that the majority of missingness occurred in the “age 18 and older” items that were to the rightmost side of the screen that were occluded. Following imputation and computation of the below tests (i.e., linear regression, t-tests, etc.), the micombine function (version 1; part of the mice/miceadds package; [51] in R (version 3.5.3)) was used to pool the estimates of F tests using the D² statistic, which provided the pooled F-statistic and p-value, as well as the D²-adjusted degrees of freedom [51, 52].

Examination of the ICQ-Specific

We subjected the ICQ-Specific to a principal component analysis (PCA) to determine if the ICQ-Specific captured a unidimensional construct that could be defined as institutional courage. This was a similar approach taken by Smith and Freyd [18] using the IBQ, who found that the IBQ is best utilized as a one-dimensional structure. The results of the PCA suggest that the ICQ-Specific is best utilized as a one-dimensional structure. A one-component structure had an eigenvalue of 8.77 that explained 48.71% of the variance.

Sexual harassment and gender differences

Across participants, 39.37% (n = 317) reported at least one instance of workplace sexual harassment. Forty-four percent of women (n = 168) reported at least one instance of sexual harassment, with men at 34.10% (n = 140), transwomen at 80% (n = 4), genderqueer participants at 75% (n = 3), transmen at 50% (n = 1), and non-binary participants at 25% (n = 1). Due to small cell sizes, participants with a gender identity other than “man” or “woman” (n = 9) were excluded from gender analyses. Women were more likely to experience at least one instance of sexual harassment than were men (χ² [1, N = 789] = 8.92, p = .003). Of those participants who did experience sexual harassment, women had higher scores on the SEQ (M = 10.68, SE = 1.14) compared to men (M = 10.35, SE = 1.28), but these differences were not statistically significant, t(284.865) = 0.20, p = .844. The analyses presented in this paper use data from only these 317 participants who reported at least one instance of sexual harassment.

Institutional betrayal prevalence

Six participants (1.9%) did not complete the IBQ following their completion of the SEQ. Of participants who experienced sexual harassment, 54.64% (n = 165) reported at least one experience of institutional betrayal. With respect to gender differences, 53.61% (n = 89) of women and 55.88% (n = 76) of men who reported sexual harassment also reported at least one experience of institutional betrayal; however, these differences were not statistically significant (χ² [1, N = 302] = 0.11, p = .654). Of those participants that did experience institutional betrayal, women experienced significantly more types of institutional betrayal (M = 4.83, SE = 0.37) compared to men (M = 3.70, SE = 0.31; t[161.330] = 2.37, p = .018).

Institutional courage prevalence

Ten participants (3.2%) did not complete the ICQ-Specific following their completion of the SEQ. Of participants who experienced sexual harassment, 76.17% reported at least one experience of institutional courage. With respect to gender differences, 71.34% of women (n = 117) and 82.09% of men (n = 110) who reported sexual harassment also reported at least one experience of institutional courage, and these differences were statistically significant (χ² [1, N = 298] = 4.69, p = .030). Of those participants who did experience institutional courage, women experienced more types of institutional courage (M = 8.01, SE = 0.53) compared to men (M = 6.97, SE = 0.49), but these differences were not statistically significant, t(224.657) = 1.44, p = .151.

Workplace outcomes—H1 & H4

Correlational results—H1.

Bivariate Pearson’s r correlations were computed to examine the initial associations between institutional betrayal, institutional courage, and employee workplace outcomes among the subset of participants who reported sexual harassment (n = 317). Institutional betrayal was negatively correlated with job satisfaction, affective organizational commitment, and perceptions of gender bias (higher scores indicate lower perceptions of gender bias). Institutional betrayal was positively correlated with work withdrawal behaviors and one-year leaving intention (lower scores indicate lower intentions of leaving one’s current employer). Institutional betrayal was not correlated with trust in management (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Means, standard deviations, and bivariate correlations between IBQ, ICQ-Specific and employee workplace outcomes (N = 317).

https://doi.org/10.1371/journal.pone.0278830.t003

Institutional courage was negatively correlated with work withdrawal behaviors and one-year leaving intentions (lower scores indicate lower intentions of leaving one’s current employer). Institutional courage was positively correlated with job satisfaction, trust in management, affective organizational commitment, and perceptions of institutional gender bias (higher scores indicate lower perceptions of gender bias; see Table 3). Institutional betrayal and institutional courage were not significantly correlated with each other.

Regression results—H4.

In order to examine our hypotheses regarding an attenuation effect between institutional courage and institutional betrayal on workplace outcomes among those experiencing sexual harassment, hierarchical linear multiple regression was used to examine the unique role of institutional betrayal, institutional courage, and their interaction on each of the employee workplace outcome variables of interest, controlling for gender. We included gender as a covariate (with cisgender women as the reference group) because of evidence suggesting that gender is linked to higher rates of sexual harassment and perceptions of gender bias. In addition, women reported significantly higher scores on the IBQ in this study. Each employee workplace variable was entered as the dependent variable. Gender and the mean-centered IBQ and ICQ variables were entered in the first step, followed by their interaction term in the second step. Participants who did not complete either the IBQ or ICQ were excluded from regression analyses.

Both institutional betrayal and institutional courage uniquely predicted job satisfaction, with the whole model explaining 24% of the variance in job satisfaction scores. In the second step of the model, the interaction term was significant and improved the model’s predictive power, accounting for an additional 2% of the variance. See Fig 1 for a plot of the interaction and Table 4 for the model statistics. At low levels of institutional betrayal, job satisfaction is similar at all levels of institutional courage. However, at high levels of institutional betrayal, a high level of institutional courage buffers against associations with job satisfaction.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Associations between different levels of institutional courage (at -1SD, the mean, and +1SD) with job satisfaction at different levels of institutional betrayal (at -1SD and +1SD).

https://doi.org/10.1371/journal.pone.0278830.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Hierarchical multiple linear regressions models predicting employee outcomes.

https://doi.org/10.1371/journal.pone.0278830.t004

Both institutional betrayal and institutional courage uniquely predicted affective organizational commitment, with the whole model explaining 22% of the variance in scores. In the second step of the model, the interaction term was significant and improved the model’s predictive power, accounting for an additional 2% of the variance in affective organizational commitment. See Table 4 for model statistics and Fig 2 for a plot of the interaction. At low levels of institutional betrayal, affective organizational commitment is similar at all levels of institutional courage. However, at high levels of institutional betrayal, a high level of institutional courage buffers against associations with affective organizational commitment.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Associations between different levels of institutional courage (at -1SD, the mean, and +1SD) with affective organizational commitment at different levels of institutional betrayal (at -1SD and +1SD).

https://doi.org/10.1371/journal.pone.0278830.g002

Both institutional betrayal and institutional courage uniquely predicted perceived gender bias, with the whole model explaining 20% of the variance in scores. In the second step of the model, the interaction term was significant and improved the model’s predictive power, accounting an additional 3% of the variance in perceived gender bias. See Table 4 for model statistics and Fig 3 for a plot of the interaction. At low levels of institutional betrayal, perceived gender bias is similar at all levels of institutional courage. However, at high levels of institutional betrayal, a high level of institutional courage buffers perceptions of gender bias.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Associations between different levels of institutional courage (at -1SD, the mean, and +1SD) with gender bias at different levels of institutional betrayal (at -1SD and +1SD).

https://doi.org/10.1371/journal.pone.0278830.g003

Both institutional betrayal and institutional courage uniquely predicted one-year leaving intentions, with the whole model explaining 15% of the variance. In the second step of the model, the interaction term was not significant; institutional betrayal and institutional courage did not interact to statistically predict one-year leaving intentions (Table 4).

Both institutional betrayal and institutional courage uniquely predicted work withdrawal behaviors, with the whole model explaining 5% of the variance. In the second step of the model, the interaction term was significant and improved the model’s predictive power, accounting for and additional 2% of the variance in work withdrawal behaviors. See Table 4 for model statistics and Fig 4 for a plot of the interaction. As institutional courage increases, work withdrawal behaviors decrease, but only among employees who experience average or lower levels (-1SD) of institutional betrayal. Those employees who experience higher levels of institutional betrayal (+1SD) appear to have similar levels of work withdrawal behaviors regardless of the level of institutional courage.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Associations of different levels of institutional courage (at -1SD, the mean, and +1SD) with work withdrawal at different levels of institutional betrayal (at -1SD and +1SD).

https://doi.org/10.1371/journal.pone.0278830.g004

Only institutional courage uniquely predicted trust in management, with the whole model explaining 2% of the variance. In the second step of the model, the interaction term was not significant; institutional betrayal and institutional courage did not interact to statistically predict trust in management (Table 4).

Psychological and physical health outcomes—H2, H3, & H5

Correlational results—H2 & H3.

Bivariate Pearson’s r correlations were computed to examine the initial associations between institutional betrayal, institutional courage, and employee psychological and physical health outcomes. Given the previous research linking a history of interpersonal trauma to the outcomes of interest, only individuals who had complete data for the BBTS were included in analyses. Institutional betrayal was significantly and positively correlated with each of the three outcomes of interest, including depression symptoms, anxiety symptoms, and somatic symptoms. Institutional courage was not significantly correlated with any of the psychological or physical health outcomes of interest. Histories of low, medium, and high betrayal trauma were each significantly and positively correlated with the psychological and physical health outcomes of interest. A history of high betrayal trauma had the strongest correlation with the psychological and physical health outcomes of interest (Table 5) and thus was used as a covariate in regression analyses.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Means, standard deviations, and bivariate correlations between IBQ, ICQ-Specific, and employee psychological and physical health outcomes.

https://doi.org/10.1371/journal.pone.0278830.t005

Regression results—H5.

In order to examine our hypotheses regarding an attenuation effect between institutional courage and institutional betrayal on psychological and physical health outcomes among those experiencing workplace sexual harassment, hierarchical linear multiple regression models were computed to examine the influence of institutional betrayal, institutional courage, and their interaction on each of the employee psychological and physical health variables of interest. Gender (with cisgender women as the reference group) and high betrayal trauma history were included as covariates. Each outcome was entered as the dependent variable. The mean-centered IBQ and ICQ-Specific variables were entered in the first step along with gender and the high betrayal trauma variable, followed by the IBQ-by-ICQ-Specific interaction term in the second step. Participants who did not complete either the IBQ or ICQ-Specific were excluded from the regression analyses.

Institutional betrayal, gender, and a history of high betrayal trauma uniquely predicted depression symptoms, with the whole model explaining 26% of the variance. Institutional courage was not a unique predictor of depression symptoms. In the second step of the model, the interaction term was not significant (Table 6).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Hierarchical multiple linear regressions predicting psychological and physical symptoms.

https://doi.org/10.1371/journal.pone.0278830.t006

Institutional betrayal and a history of high betrayal trauma significantly predicted anxiety symptoms, with the whole model explaining 19% of the variance. Institutional courage was not a unique predictor of anxiety symptoms. In the second step of the model, the interaction term was not significant (Table 6).

Institutional betrayal and a history of high betrayal trauma uniquely predicted somatic symptoms, with the whole model explaining 19% of the variance. Institutional courage was not a unique predictor of somatic symptoms in step one. In the second step of the model, the interaction term was significant and improved the model’s predictive power, accounting for and additional 2% of the variance in somatic symptoms. To probe this interaction, we conducted a simple slopes analysis to determine if institutional courage would predict somatic symptoms at different values of institutional betrayal. At +/– 1 standard deviation of institutional betrayal, institutional courage remained a non-significant predictor of somatic symptoms. However, at +/– 1.5 standard deviations of institutional betrayal, institutional courage became a significant predictor of somatic symptoms in the second step of the model. At very low levels of institutional betrayal, institutional courage is associated with decreased somatic symptoms. Conversely, at very high levels of institutional betrayal, institutional courage is associated with increased somatic symptoms (+1.5SD of institutional betrayal, institutional courage is significant, p = .036; at -1.5SD of institutional betrayal, institutional courage is significant, p = .034). See Table 6 for model statistics and Fig 5 for a plot of the interaction.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Associations between different levels of institutional courage (at -1SD, the mean, and +1SD) with somatic symptoms at different levels of institutional betrayal (at -1SD and +1SD).

https://doi.org/10.1371/journal.pone.0278830.g005

Practical and clinical implications

The results presented here suggest that both institutional betrayal and institutional courage are not unusual events in the context of workplace sexual harassment. While it is heartening to see that just over three-quarters of employees who experienced sexual harassment also experienced institutional courage, it is discouraging—although unsurprising—to see that over half of employees experienced institutional betrayal. In fact, these rates of institutional betrayal are higher than those observed in the context of sexual assault in some studies (e.g., [18, 19]). These findings thus have implications for professionals and practitioners in a wide variety of fields, such as management, human resources, and human capital management. Education about the frequency and forms of institutional betrayal and institutional courage is the first step to identifying these phenomena, as well as their effects within institutions at both the macro/structural and micro/individual levels.

As aforementioned, our results suggest multiple domains of impact. In previous research on institutional betrayal, the focus has largely been—and with good reason—on the negative psychological and physical health effects that result from these types of experiences (for a brief review, see [54]). Consider this the “first argument” for reducing institutional betrayal and increasing institutional courage; in other words, because institutional betrayal is associated with psychological and physical harm, it is “worth” reducing institutional betrayal and increasing institutional courage. It is “worth” doing so in the sense that reducing harm is a compassionate, moral act. Much in the way that research on institutional betrayal in the healthcare system has focused on disengagement from healthcare [23, 53], this study similarly presents an additional, second argument both regarding the effects of institutional betrayal, that of negative employee workplace outcomes and institutional health, and for replacing institutional betrayal with institutional courage. Although we did not explore possible downstream economic and institutional effectiveness outcomes (i.e., the degree to which an organization achieves its stated goals, which may include profit benchmarks or production targets) as associated with institutional betrayal, institutional courage, and employee outcomes, it is important to recognize that the employee workplace and employment-related outcomes examined in this study can and do have a substantial impact on organizations ([40], e.g., [55]).

Limitations

The results of this study should be interpreted in light of its limitations. First, we collected data at only one timepoint, which limits any conclusions related to direct causality. This was in line with the aim of the study, which included characterizing institutional courage and institutional betrayal with respect to workplace sexual harassment—the first empirical examination of these phenomena in this context, and the first empirical examination of institutional courage. This design does limit our ability to examine the time-course of events, especially as it is likely that institutional betrayal and institutional courage occur over time and at different points following an incident of sexual harassment. A longitudinal design would be particularly useful for examining both employee workplace outcomes as well as psychological and physical health outcomes over time. Second, we had small cell sizes for non-gender-binary individuals (i.e., those that do not identify as “man” or “woman”). This limits the generalizability of our results to individuals who are gender minorities, a group who has been found to experience very high rates of sexual harassment and other types of victimization [56, 57]. Similarly, future investigations may benefit from oversampling individuals who identify as racial and ethnic minorities, given that racial and ethnic minorities often experience greater rates, and more severe types, of both workplace sexual and general harassment in the workplace [58–60]. Third, because of the exploratory phase of the research, we cast a wide net of variables measured. We acknowledge that this may have inflated Type I error, with some overlap in the outcome variables.

Future directions

There are many avenues for future research on institutional betrayal and institutional courage. First, examining the rates at which institutional courage and institutional betrayal occur following workplace sexual harassment by employment industry and sector would add to the current findings. Additionally, examining institutional courage and institutional betrayal within one organization could be useful for examining multilevel effects, such as individuals nested within teams nested within units. Such models would provide a better understanding of these effects at different levels within organizations, which in turn could further delineate targets for intervention.

Interventions to replace institutional betrayal with institutional courage should start with education—by training employees, supervisors, human resources professionals, executives, and so on—to first identify institutional betrayal and institutional courage when they happen. Institutions can enact policies and procedures designed to reduce institutional betrayal (e.g., allowing victims to have a say in how their case is handled, such as opting for victim-centered policies rather than policies of mandatory reporting/compelled disclosure; [61]), and promote institutional courage (e.g., an incentive structure that rewards reporting sexual harassment). In either of these forms of intervention—and many others exist too—data should be gathered throughout the process of intervention design, implementation, and dissemination.