In recent months, the US has seen a deluge of high-profile sexual harassment reports stretching from Hollywood movie studios to DC government offices to famous kitchens. In response, leaders, such as Speaker of the House Paul Ryan, have called for mandatory sexual harassment training, stating, “our goal is not only to raise awareness, but also make abundantly clear that harassment in any form has no place in this institution” (Schor, Reference Schor2017). However, as many media outlets have reported, these issues are largely embedded in workplace cultures (Gonzalez, Reference Gonzalez2017) and thus require systemic and pervasive changes if organizational leaders are going to reduce, and ultimately, eliminate sexual harassment. Similarly, the recent high-profile court case People v. Turner, among others, has drawn focus once again to the issue of sexual assault on college campuses. Although universities responded to the Violence Against Women Act (1994) by instituting campus-wide training programs, ranging in focus from self-defense (Gidycz, Rich, Orchowski, King, & Miller, Reference Gidycz, Rich, Orchowski, King and Miller2006) to bystander intervention (Banyard, Plante, & Moynihan, Reference Banyard, Plante and Moynihan2004), sexual harassment and assault still remain a prevalent issue on college campuses today. Further, in a review of 69 studies on the topic, Anderson and Whiston (Reference Anderson and Whiston2005) found that the most effective training was longer, more in-depth, and focused on gender-role socialization, rape myths, and risk-reduction strategies but that very few of the included studies measured how training efforts influence long-term behavioral or cultural change (e.g., Brecklin & Forde, Reference Brecklin and Forde2001).
Personnel psychology literature has long recognized the importance of examining long-term effectiveness and addressing influences outside of the immediate training environment (Georgenson, Reference Georgenson1982; Saks, Reference Saks2002). In the 1980s, organizations encountered the “transfer problem”—the failure to facilitate the application of new knowledge, skills, or attitudes gained in training to the actual work environment (Baldwin & Ford, Reference Baldwin and Ford1988)—as they siphoned money into training efforts while seeing little change in employee performance outside of the training context. Since, scholars have investigated factors influencing transfer including trainee (e.g., Colquitt, LePine, & Noe, Reference Colquitt, LePine and Noe2000), training design (e.g., Machin & Fogarty, Reference Machin and Fogarty2004), and organizational culture (e.g., Mathieu, Tannenbaum, & Salas, Reference Mathieu, Tannenbaum and Salas1992) characteristics. Despite interest in this topic, evaluation of workplace sexual harassment prevention training efforts and efficacy remains elusive outside of the context of college campuses, military contexts, or publicly accessible data (e.g., Antecol & Cobb-Clark, Reference Antecol and Cobb-Clark2003). Further, little work has investigated how transfer may be facilitated and what unique interventions may be applied in the increasingly more common but less traditional organizational structures and designs with regard to sexual harassment (Perry, Kulik, & Field, Reference Perry, Kulik and Field2009).
Thus, although mandating training represents a step toward changing organizational cultures, these efforts will prove ineffective if not carefully designed and executed, with close attention paid to how the organizational culture facilitates transfer to the television studio, the halls of the Senate, a restaurant kitchen, the boss’s office, the military battleground, or the fraternity house. As Ruth Anne Koenick, Rutger’s director of the Office for Violence Prevention and Victim Assistance, stated, “You can’t do a one-time program and expect it to make a difference” (Howard, Reference Howard2015). Indeed, without proper consideration of conditions outside of the training environment, even the best training efforts may fall short of influencing post-training behavior. Sexual harassment and assault trainings are no exception. However, when evidence regarding the long-term effectiveness of these programs has been gathered, results generally suggested a positive attitude change immediately following the intervention that tended to regress over time (Brecklin & Forde, Reference Brecklin and Forde2001). If behavioral change outside of the training session is the ultimate goal, we must consider how factors both within and outside of the training environment influence transfer with regard to sexual harassment and assault prevention.
The current focus on workplace sexual harassment and college campus assault presents an opportunity for industrial and organizational (I-O) psychologists to apply our knowledge of training, organizational culture, and program evaluation to influence how these programs are designed and assessed to ultimately create lasting behavioral and cultural changes at the organizational and societal levels. However, sexual harassment and assault training efforts present unique challenges not akin to more traditional training efforts. First, the nature of these issues does not facilitate the direct application of traditional training principles in a straightforward manner. For example, in order to facilitate long-term change, the training literature emphasizes the importance of creating a work environment that provides opportunities to practice and apply learned skills (Goldstein & Ford, Reference Goldstein and Ford2002). However, if the goal is to eliminate sexual harassment, then creating an environment where employees or students are prompted to practice either responding to sexual advances, reporting these issues, or disengaging their own inclination to sexually harass seems counterintuitive. Thus, finding unique methods for these groups to extend their learning outside of the training becomes important. Second, measuring the effectiveness of these programs is challenging due to the sensitive and socially desirable nature of relevant attitudes. Most critically, there seems to be a misalignment between the constructs being measured and the intended outcomes of training initiatives. Evidence of this has been borne out by the lack of continuity between both knowledge and attitudes (Lonsway & Kothari, Reference Lonsway and Kothari2000), and short-term and long-term attitudes (e.g., Black, Weisz, Coats, & Patterson, Reference Black, Weisz, Coats and Patterson2000; Breitenbecher, Reference Breitenbecher2001). In order to assess what is intended—behavioral intentions vis-à-vis attitudes—researchers, practitioners, and university administrators must revisit and adapt existing measures, develop new scales that better approximate behavior, and reflect on the core message being relayed to the trainees during content delivery (e.g., “we just need to check the box”). This combination of challenging transfer circumstances and lack of precision in measurement presents a gap for I-O psychologists to address if these efforts are to ever prove fruitful.
The present effort proposes a framework for designing sexual harassment prevention programs in organizations that is grounded in the literature on training and development, workplace sexual harassment, and college sexual assault. In presenting recommendations for sexual harassment prevention, we draw heavily from the literature on college campus sexual assault. Although there are limitations to this approach, the sustained interest in, and investigations into, college campus sexual assault prevention provide a repository of empirically tested interventions that helps fill the gaps left by the lack of empirical research on sexual harassment interventions. Further, sexual harassment and assault have been previously described as a continuum by which, if left unchecked, less severe behaviors may become more violent over time (Department of Defense, 2014; Fitzgerald, Reference Fitzgerald1993). Thus, although sexual harassment and assault are not one and the same, the escalating nature suggests that interventions should target the full spectrum of behavior, especially in light of limited time and financial resources dedicated to training initiatives.
Prior to turning to our proposed framework, it is also important to address the common belief that sexual assault is “severe, but rare.” The little research that is available on sexual assault in the workplace challenges this belief. For example, a commissioned RAND report (Morral, Gore, & Schell, Reference Morral, Gore and Schell2015) found that, across four branches of the military, between 4.5% to nearly 8% of female active-duty service members and, on average, 1% of male activity-duty service members reported experiencing some form of sexual assault in the past year. Although the military is a non-traditional work environment, we point to its prevalence rates here because (a) the military is one of the few workplaces that transparently provides incidence data to the public about both sexual harassment and assault that specifically occurs within its organization and (b) compared to the Society for Human Resource Management (SHRM) sexual harassment survey conducted in 2017, the reported prevalence of sexual harassment in more traditional workplaces (36% of respondents) is greater than the average rates reported across surveyed branches of the military (28% of respondents). Although these are clearly different timeframes and populations with distinct norms and policies, this is not an isolated circumstance. Research suggests that employees are experiencing sexual assault and aggressive harassment in some of the largest growing economic sectors (e.g., health care; Rippon, Reference Rippon2000) and some of the least visible work environments (e.g., janitorial staff and shift work; Zhang, Spiller, Finch, & Qin, Reference Zhang, Spiller, Finch and Qin2014). We do not believe it is in the interest of employees, I-O psychologists, HR professionals, or other training personnel to denounce or overlook the need to address sexual assault simply because organizations with more traditional structures or job designs have been less forthcoming with data about prevalence and training outcomes or avoided disclosure entirely via the use of nondisclosure agreements.
Together, this evidence suggests that sexual harassment and sexual assault are issues in organizations of all kinds and need to be addressed. The I-O community is in an especially unique position to add a voice to this discussion by drawing on expertise related to attitude, behavior, and culture change. Thus, we propose a framework for designing sexual harassment training programs that will transfer to the workplace by addressing individual differences, training design, evaluation, and institutional climate. We argue that organizations, traditional and nontraditional, should take preventive actions to eliminate sexual harassment through careful training design and aligning the culture to support the application of training content and transfer of attitudes to the broader organizational environment. To do so, organizations must regularly monitor and evaluate attitudes and cultural artifacts. If true transformation is to occur, we must make real commitments to change, infusing this commitment within formal and informal networks across all organizational levels.
Individual differences
Individual differences such as personality (e.g., Barrick & Mount, Reference Barrick and Mount1991), cognitive ability (e.g., Colquitt et al., Reference Colquitt, LePine and Noe2000; Day, Arthur, & Gettman, Reference Day, Arthur and Gettman2001), and self-efficacy (e.g., Harrison, Rainer, Hochwarter, & Thompson, Reference Harrison, Rainer, Horchwarter and Thompson1997; Morin & Latham, Reference Morin and Latham2000) have been shown to play a key role in training transfer. However, individual differences outside of demographics have typically been ignored in a sexual harassment context (Perry et al., Reference Perry, Kulik and Field2009). Thus, we focus our attention on factors that are important to training transfer yet have received little attention in the sexual harassment and assault literatures. In their review, Burke and Hutchins (Reference Burke and Hutchins2007) found a moderate to strong relationship between training transfer and eight individual-level characteristics: (a) cognitive ability, (b) self-efficacy, (c) pretraining motivation, (d) anxiety/negative affectivity, (e) openness to experience, (f) perceived utility, (g) career planning, and (h) organizational commitment. Of these, self-efficacy, perceived utility, and pretraining motivation are most readily influenced by an organization prior to, and during, sexual harassment and assault education efforts.
Self-efficacy
Self-efficacy refers to one’s own belief that they can perform the task and accomplish a goal (Bandura, Reference Bandura1982). Applied here, self-efficacy manifests as the belief that one can successfully implement strategies and knowledge related to sexual harassment and assault learned in training to real-world situations. Broadly, research on training performance and self-efficacy suggests a positive relationship (e.g., Mathieu, Martineau, & Tannenbaum, Reference Mathieu, Martineau and Tannenbaum1993). Other work in this area has successfully manipulated self-efficacy, suggesting that self-efficacy may be an entry point for influencing training effectiveness (e.g., Vancouver, Gullekson, Morse, & Warren, Reference Vancouver, Gullekson, Morse and Warren2014).
In a review of social cognitive theory (Bandura, Reference Bandura1986, Reference Bandura2001), Stajkovic and Luthans (Reference Stajkovic, Luthans, Porter, Bigley and Steers2002) review four determinants of self-efficacy: (a) mastery experiences, (b) vicarious learning/modeling, (c) verbal persuasion, and (d) physiological and psychological arousal. Applying social cognitive theory to the sexual harassment and assault prevention context may involve providing trainees with challenging activities such as complex cases or roleplays that would allow them to practice their responses and subsequently build self-efficacy. This must be followed, however, by feedback encouraging employees and students to properly attribute their performance. Additionally, studies of sexual assault education efforts have often touted the importance of using peer trainers, but meta-analytic work shows little value in this approach (Anderson & Whiston, Reference Anderson and Whiston2005). We argue that peers do play a critical role but not as the trainers. Providing cases with peers similar to those in training (e.g., from the same university or a similar organization, field, or identity) and real-world examples solicited from students at the university or similar organizations will provide cues that other students or employees may be able to successfully implement similar tactics and actions. Further, it is important to draw on diverse sources across multiple student groups. For example, many university interventions often target Greek Life, but by doing so, they fail to connect with students not involved in a fraternity or sorority.
Perceived utility
Perceived utility refers to the assumed training usefulness for job performance (Chiaburu & Lindsay, Reference Chiaburu and Lindsay2008). Grossman and Salas (Reference Grossman and Salas2011) argue that trainees who view training as useful are more likely to transfer learned skills or knowledge to the job. Similarly, Roszkowski and Soven (Reference Roszkowski and Soven2010) found a strong positive relationship between perceived utility and learning in an undergraduate orientation program. Thus, emphasizing the degree to which sexual harassment and assault training will prove useful should be a feature of these programs. Practical strategies for communicating importance include explicit support from organizational leaders, university faculty, and peers. Formally, positive feedback collected about the training could be disseminated. Similarly, the university or organization could highlight the importance of this training to leaders such as managers and faculty, as well as ask for their support in encouraging employees and students to engage in training. Statistics and trainee feedback regarding the efficacy of training may further communicate the utility of training. Of course, this points to the need for sound measurement, which will be discussed in a later section.
Pretraining motivation
Several studies have established a positive relationship between trainee motivation to learn and actual learning (e.g., Mathieu et al., Reference Mathieu, Tannenbaum and Salas1992). Motivation is a key factor in training as it prompts arousal, persistence, and direction. Arousal and persistence refer to the time and effort an individual invests, whereas direction refers to the behaviors in which an individual invests (Mitchell, Reference Mitchell1982). Thus, motivation impacts the degree to which an individual not only puts time into the training but also effort into learning content (Facteau, Dobbins, Russell, Ladd, & Kudisch, Reference Facteau, Dobbins, Russell, Ladd and Kudisch1995). Switzer, Nagy, and Mullins (Reference Switzer, Nagy and Mullins2005) found that training reputation, or the degree to which training is perceived as high quality, influences training transfer through pretraining motivation. Thus, increasing quality perceptions regarding sexual harassment and assault education efforts may increase their effectiveness. Note, increasing perceptions of quality should be accomplished by actual increases in the quality of a program in addition to managing perceptions. Four key ways to signal the quality of a program are by using high quality trainers, aligning systems to support the training purpose, making the training interactive and practical, and providing appropriate resources (Hill, Lomas, & MacGregor, Reference Hill, Lomas and MacGregor2003). Of particular importance is the systems alignment piece. This suggests that one must have a clear understanding of the reporting mechanisms, support, and cultural cues that signal to organizational members, employees, and students the importance placed on the training within the organization.
Needs analysis
A needs analysis gathers information on four key areas: (a) current and potential problems, (b) management support, (c) baseline evaluation data, and (d) costs and benefits (Brown, Reference Brown2002). More broadly, a needs analysis consists of three steps: (a) organizational analysis, (b) task analysis, and (c) person analysis (Arthur, Bennett, Edens, & Bell, Reference Arthur, Bennett, Edens and Bell2003). This information is subsequently used to inform training design and communication (Goldstein & Ford, Reference Goldstein and Ford2002).
In a sexual harassment context, a needs analysis may provide information regarding both reactive and proactive initiatives. For instance, an organizational analysis that reveals pervasive sexual harassment issues across an organization may prompt a reactive response aimed at reducing or eliminating existing behavior as well as healing the organization. In contrast, a needs analysis may also reveal situations ripe for sexual harassment to emerge. For instance, a person analysis may identify pockets of negative attitudes that could lead to sexual harassment. Similarly, an organizational analysis may identify cultural artifacts that are conducive to sexual harassment. In both cases, information gathered should then be used to design training sand culture change initiatives to limit the likelihood of sexual harassment issues arising. Although reactive responses are important, we argue that needs analysis should be conducted routinely to prevent sexual harassment issues.
A needs analysis may reveal other organization- or person-centered issues that must be paid special attention in a sexual harassment context. For instance, if an organization identifies a single department with particularly pervasive sexual harassment issues, then it may be necessary to include additional solutions for this group or to monitor this group more closely than others. Further, identified problems may challenge popular conceptions of how a training should be delivered and to whom. For instance, popular notions suggest that men in positions of power may be especially likely to harass female subordinates. However, some research suggests that women in power positions are more likely to report sexual harassment compared to subordinates (McLaughlin, Uggen, & Blackstone, Reference McLaughlin, Uggen and Blackstone2012). Thus, collecting qualitative or quantitative data regarding attitudes and instances of sexual harassment in a specific organization may challenge popular beliefs and reveal unique considerations that, if appropriately incorporated into the training design, may improve training success. In response to findings in a needs analysis, some would suggest designing specific trainings tailored to particular groups. For instance, special interventions are often developed specifically for women such as self-defense training (e.g., Sochting, Fairbrother, & Koch, Reference Sochting, Fairbrother and Koch2004), and recent work suggests that typical interventions may embolden men who pose a high risk of committing sexual assault prior to training (Malamuth, Huppin, & Linz, Reference Malamuth, Huppin and Linz2018).
Although it may be logical to design specific interventions for subgroups of organizational members based off results obtained in this process, it is important to bear in mind the implications of such an approach. For instance, when discussing threats to internal validity, Shadish, Cook, and Campbell (Reference Shadish, Cook and Campbell2002) discuss compensatory rivalry and resentful demoralization. Although in an internal validity sense, this refers to a control group’s response to feelings of inequity, it could also be applied to those who are selected for training of a sensitive nature such as sexual harassment training. For example, if a needs analysis identifies individuals who are most likely to sexually harass others and provides training only to those members, resentment regarding why one was selected may fester. Resentment may give way to hardened attitudes and unwillingness to listen or consider the information presented. Further, those who are not selected for the training may respond differently to those who are and generate undesired effects.
A similar issue has been discussed in the diversity training literature. Specifically, the diversity literature hotly debates the effectiveness of group heterogeneity within a training session, with some arguing that heterogeneity increases the quality of discussion and others arguing that heterogeneity may reinforce existing beliefs (Roberson, Kulik, & Pepper, Reference Roberson, Kulik and Pepper2003). In the sexual harassment context, it is then important to consider who is selected for training and how these individuals are combined in the training setting. Although some research has been advanced in this regard in the sexual harassment literature, decisions regarding if and how to assign organizational members based on gender, level, attitudes, and previous experiences or accusations should be made carefully, with attention paid to the specific behavioral and attitudinal problems in an organization. In order to make these decisions, however, one has to be knowledgeable of the attitudes and issues present, and where they have potential to become problematic.
Additionally, organizations may benefit from including a climate analysis, which provides information bearing on the shared beliefs within an organization (Tracey, Tannenbaum, & Kavanagh, Reference Tracey, Tannenbaum and Kavanagh1995). In this instance, a climate analysis may provide information regarding perceived support for transfer, attitudes of leadership, and beliefs regarding policies. This information provides a baseline from which leaders may, in turn, make improvements to better facilitate transfer.
Relatedly, a needs analysis may shine a light on moderators of training transfer that should be planned for in advance. For instance, managerial attitudes regarding sexual harassment and related initiatives may inhibit or facilitate transfer. A carefully conducted needs analysis will identifying these attitudes in advance, allowing for a more comprehensive post-training plan to ensure that the intervention is supported or, if needed, develop follow-up training for specific topics or concerns. Thus, a needs analysis is critical to successful sexual harassment prevention programs and should serve as a foundation for subsequent design, content, and post-training decisions to help facilitate transfer and ultimately prevent sexual harassment.
Training design
In addition to individual characteristics, several choices must be made in the design process including what content should be taught, how the content will be delivered, and how to incorporate practice opportunities (Goldstein & Ford, Reference Goldstein and Ford2002). Further, each choice made in this process will not only influence the degree to which trainees acquire new knowledge, skills, or attitudes, but also the extent to which they will apply what they have learned to their lives. In their review of training transfer, Burke and Hutchins (Reference Burke and Hutchins2007) found five variables with a strong or moderate relationship with training transfer. These include (a) learning goals, (b) content relevance, (c) practice and feedback, (d) behavioral modeling, and (e) error-based examples. Although some of these design characteristics have been discussed in the sexual harassment literature (Perry et al., Reference Perry, Kulik and Field2009), it is important to emphasize their unique role in sexual harassment efforts and impact on changing cultures.
Learning goals
Learning goals describe the behavior that the training is attempting to influence (Goldstein & Ford, Reference Goldstein and Ford2002; Mager, Reference Mager1962). We assume here that the goal of most sexual harassment and assault trainings is to reduce or, ideally, eliminate instances of sexual harassment or assault. However, this goal does not clearly outline desired individual posttraining behaviors. The literature suggests that setting both short- and long-term behavioral goals is associated with increased transfer (Brown, Reference Brown2005; Kontoghiorghes, Reference Kontoghiorghes2001). Thus, sexual harassment and assault training efforts may benefit from specific goals that align with desired future behavior. For example, Paul Ryan’s statement that the training’s goal is to increase awareness should then be attached to a behavior such as being able to better identify and stop instances of sexual harassment or assault. Or, perhaps a more appropriate shift would be to increase the specificity of the goal to address awareness of what behaviors constitute sexual harassment and assault, strategies for curbing these behaviors before they are executed, and consequences for engaging in the various forms of sexual misconduct.
Content relevance
When establishing goals and designing training activities, the real-world context in which the training will be applied should be taken into account (Bates, Reference Bates, Maycunich Gilley, Callahan and Bierema2003). Regarding the university setting, students should see the connection between the training content and their campus life (Burke & Hutchins, Reference Burke and Hutchins2007). As an example, Hanson and Gidycz (Reference Hanson and Gidycz1993) describe an acquaintance rape prevention program in which female students watched a video highlighting situational variables prior to an acquaintance rape. This video was followed by questions from the instructor and a discussion of these situational variables. Further, trainers should make the connection between content, practice activities, and potential post-training experiences explicit—making it clear how the content relates to them.
Practice and feedback
The relationship among practice, accurate feedback, and training transfer is well-established (e.g., Holladay & Quinnones, Reference Holladay and Quinones2003; Van den Bossche, Segers, & Jansenm Reference Van den Bossche, Segers and Jansen2010). At first thought, practice in a sexual harassment and assault context may seem inappropriate. However, practice may be provided through cases and roleplays. Cases provide trainees with realistic scenarios and ask them to address which actions should occur next. This approach brings to the fore differing perspectives that prompt discussion among the trainees. Roleplays could be implemented similarly and provide a more realistic practice opportunity for students to practice new strategies. For instance, Heppner, Humphrey, Hillenbrand-Gunn, and DeBord (Reference Van den Bossche, Segers and Jansen1995) had facilitators improvise a situation that ends in rape. Students provide suggestions for improving decisions and actions taken in the scenario in order to avoid the rape. The facilitators then act out this rewritten script to show how these actions affect the outcome.
It is noteworthy to mention the skill required to facilitate trainings with practice opportunities: providing accurate, valuable feedback. Goldstein and Ford (Reference Goldstein and Ford2002) note that perceptions of feedback accuracy influence the effect of feedback, and the accuracy of feedback depends, in part, on the credibility of the source. Thus, those conducting these trainings should be seasoned professionals. This aligns with the meta-analysis conducted by Anderson and Whiston (Reference Anderson and Whiston2005), which found that trainings conducted by professionals were more effective than those conducted by peers. It cannot be stressed enough that facilitating these trainings is a difficult undertaking. If seasoned professionals are not available, it is imperative that trainers themselves receive ample training, support, practice, and feedback prior to leading the training. The importance of attuned feedback in this context also highlights the need for in-person training sessions. Although online training may be more cost effective and easier to administer, the opportunity to provide specific feedback is limited. Additionally, it limits the type of practice opportunities that may be presented and, thus, may limit the acquisition of relevant knowledge and skills in this context.
Behavioral modeling and error-based examples
Behavioral modeling takes root in social learning theory (Bandura, Reference Bandura1977), which argues that four processes are important to learning: attention, retention, reproduction, and motivation. In a meta-analysis of behavioral modeling training, Taylor, Russ-Eft, and Chan (Reference Taylor, Russ-Eft and Chan2005) found that transfer was at its highest when trainees were presented with both positive and negative examples, developed their own scenarios, and set goals. Additionally, the meta-analysis found that transfer was highest when supervisors also received training and a post-training reward, and a punishment system was established. Providing trainees with rule codes, or information presented as a general rule rather than describing a behavior (Decker, Reference Decker1980), was also related to transfer.
In the context of sexual harassment and assault training, behavioral modeling may be accomplished through a variety of mechanisms. For instance, cases can provide trainees with behavioral examples to observe and assess. Alternatively, these efforts could provide richer sources of behavioral modeling through roleplays, videos, or skits. Regardless of mechanism, Taylor et al. (Reference Taylor, Russ-Eft and Chan2005) suggest that the chosen delivery method should incorporate examples of both desired and undesired behavior. Error-based examples provide trainees with examples of the consequences of poor decisions, behaviors, or a misapplication of strategies. Past research (e.g., Hesketh, Reference Hesketh2000; Smith-Jentsch, Jentsch, Payne, & Salas, Reference Smith-Jentsch, Jentsch, Payne and Salas1996) provides evidence for the efficacy of this approach for increasing training transfer. An assessment of these models—encouraging trainees to analyze the good and bad choices made and strategies applied—should follow their introduction. Additionally, these efforts could ask trainees to develop, analyze, and discuss their own skits, cases, or roleplays displaying the rule codes taught during the training.
Measurement and evaluation decisions
Accurate evaluation is critical to understanding which methods of training are most effective for preventing sexual harassment and assault (Arthur et al., Reference Arthur, Bennett, Edens and Bell2003; Phillips & Phillips, Reference Phillips and Phillips2016). Given the scarcity and restricted breadth of workplace sexual harassment training research, in the following sections we rely on the abundance of training research conducted in concert with college campuses to gain a better understanding of how best to approach sexual harassment and assault prevention training efforts, methodologically and psychometrically speaking.
Researchers have tended to rely on knowledge measures and attitudinal surveys to assess training effectiveness (Morrison, Hardison, Mathew, & O’Neil, Reference Morrison, Hardison, Mathew and O’Neil2004), and, in addition, there is a great deal of variety in assessment types (e.g., survey, interview, focus group) and research designs (e.g., post-test only, pretest/post-test, long-term evaluation). Despite the variety of outcome criteria and evaluation methodologies in the literature, there is a dearth of evidence regarding which methodologies result in the greatest training effectiveness and, ultimately, transfer outside of the training context to reduce sexually violent and inappropriate behavior (for exception see Morrison et al., Reference Morrison, Hardison, Mathew and O’Neil2004; Yeater & O’Donohue, Reference Yeater and O’Donohue1999). To address this need, a critical review of measurement and design frequently utilized in sexual assault prevention efforts is provided in the following sections, followed by recommendations for improvement in future training endeavors.
Outcome measures
In the past, researchers have assessed a variety of intended training outcomes, including knowledge measures that focus on evaluating explicit knowledge such as definitions and statistics about rape and sexual assault (Breitenbecher, Reference Breitenbecher2001; Breitenbecher & Scarce, Reference Breitenbecher and Scarce1999; Orchowski, Gidycz, & Raffle, Reference Orchowski, Gidycz and Raffle2008; RAVE, 1977; Weisz & Black, Reference Weisz and Black2001), and consent and coercion (Gibson & Humphrey, Reference Gibson and Humphrey1993), as well as researcher-developed training-specific knowledge assessments (Banyard, Eckstein, & Moynihan, Reference Banyard, Eckstein and Moynihan2010; Lonsway & Kothari, Reference Lonsway and Kothari2000); attitudinal measures about rape mythology (Ashton, Reference Ashton1982; Burt, Reference Burt1980; Humphrey, Reference Humphrey1996; McMahon & Farmer, Reference McMahon and Farmer2011; Payne, Lonsway, & Fitzgerald, Reference Payne, Lonsway and Fitzgerald1999), bystander intervention (Ahrens, Rich, & Ullman, Reference Ahrens, Rich and Ullman2011; Banyard, Moynihan, & Crossman, Reference Banyard, Moynihan and Crossman2009; Banyard, Plante, & Moynihan, Reference Banyard, Plante and Moynihan2005; Banyard et al., Reference Banyard, Eckstein and Moynihan2010; Coker et al., Reference Coker, Cook-Craig, Williams, Fisher, Clear, Garcia and Hegge2011; McMahon, Reference McMahon2010), and acquaintance and date rape (Hanson & Gidycz, Reference Hanson and Gidycz1993; Holcomb, Holcomb, Sondag, & Williams, Reference Holcomb, Holcomb, Sondag and Williams1991; Holcomb, Sarvela, Sondag, & Holcomb, Reference Holcomb, Sarvela, Sondag and Holcomb1993; Humphrey, Reference Humphrey1996; Lanier, Elliott, Martin, Kapadia, Reference Lanier, Elliott, Martin and Kapadia1998); individual self-reported behavior (Hanson & Gidycz, Reference Hanson and Gidycz1993; Koss & Gidycz, Reference Koss and Gidycz1985; Orchowski et al., Reference Orchowski, Gidycz and Raffle2008; Stephens & George, Reference Stephens and George2004); and organizational incidence reporting (McMahon, Reference McMahon2008; Meilman & Haygood-Jackson, Reference Meilman and Haygood-Jackson1996).
Sexual assault prevention training that has relied on increases of, or changes in, explicit knowledge to assess effectiveness has shown mixed results, and reported positive trends are often conditional on other factors. For instance, some gains in knowledge appear to be a function of moderating factors (e.g., trainee gender; Weisz & Black, Reference Weisz and Black2001). Changes in attitudes also appear to be dependent on other factors, including trainee gender (Heppner, et al., Reference Heppner, Humphrey, Hillenbrand-Gunn and DeBord1995; Kress, et al., Reference Kress, Shepherd, Anderson, Petuch, Nolan and Thiemeke2006) and time of measurement (Foubert & Marriott, Reference Foubert and Marriott1997; Heppner et al., Reference Heppner, Humphrey, Hillenbrand-Gunn and DeBord1995; Lonsway & Kothari, Reference Lonsway and Kothari2000), and, in addition, such attitudinal changes appear to be short lived. Relationships between scores on knowledge measures and other outcome measures appear to be tenuous as well. Despite long-term sexual assault knowledge retention, Lonsway and Kothari (Reference Lonsway and Kothari2000) and Gibson and Humphrey (Reference Gibson and Humphrey1993) found that training participants experienced no long-term attitude change. Likewise, Heppner, Humphrey, Hillenbrand-Gunn, and DeBord (Reference Heppner, Humphrey, Hillenbrand-Gunn and DeBord1995) measured five time periods, including four post-training waves extending 5 months, and found that attitudinal scores rebound after the 1-month mark. Indeed, Heppner, Neville, Smith, Kivlighan, and Gershuny (Reference Heppner, Neville, Smith, Kivlighan and Gershuny1999) proposed three trainee response profiles on “rejection of rape” attitudes: improvers who improve attitudes after training in both short- and long-term measurement, deteriorators whose attitudes gradually decline over time, and rebounders who show short-term but not long-term progress. Overall, there is a clear pattern that emerges when considering long-term changes in attitudes following sexual assault prevention training. In short, results suggest that although trainees are learning new terms, they are not changing their minds or their behavior, and, more troublingly, in some instances trainees are worse off than having had no training at all.
Evaluation methodology
In terms of design, training evaluation systems generally adhere to one of three types: post-training evaluation only, pre- and post-training evaluation, and multiwave training evaluation. Each type of evaluation system answers a specific question and, in some cases, does so at the expense of understanding long-term trends important for assessing training transfer. Post-training evaluations only, for instance, are typically used for convenience and to evaluate differences between training conditions and control conditions (Coker et al., Reference Coker, Cook-Craig, Williams, Fisher, Clear, Garcia and Hegge2011). Such evaluation systems may work well in early phases of training development in order to assess which methods of training work best for the stated goals of training (Barone, Wolgemuth, & Linder, Reference Barone, Wolgemuth and Linder2007; Choate, Reference Choate2003). However, post-training evaluation forfeits the ability to assess any change in knowledge, attitudes, or behavior as a result of training. To evaluate attitude change, pre- and post-training evaluations are required. Given that studies of sexual assault prevention training have been mixed in terms of short-term attitude change, there is still a clear need to establish which content and delivery systems could result in meaningful change at all.
Perhaps the most notable finding from this body of work is that training efforts seem to immediately alter attitudes, but there is less evidence that attitudes are maintained or continue to improve in the long term (Black et al., Reference Black, Weisz, Coats and Patterson2000; Foubert & Marriott, Reference Foubert and Marriott1997; Heppner et al., Reference Heppner, Humphrey, Hillenbrand-Gunn and DeBord1995, Reference Heppner, Neville, Smith, Kivlighan and Gershuny1999; Lonsway & Kothari, Reference Lonsway and Kothari2000). This finding is especially pronounced as time between training and post-training evaluation increases (Breitenbecher, Reference Breitenbecher2001) despite evidence of long-term knowledge retention (Breitenbecher & Scarce, Reference Breitenbecher and Scarce1999). After reviewing available evaluations of sexual assault prevention training programs, it is clear that those concerned with developing impactful training should focus on long-term, multiwave assessment. To anyone vaguely familiar with research or training methodology, this suggestion may appear arbitrary and obvious; however, it is noted here not as a formality but rather as a criticality. Multiwave evaluation systems are the only means through which long-term attitude change, behavioral change, and trainee ability to transfer training content to novel circumstances can be assessed (Orchowski et al., Reference Orchowski, Gidycz and Raffle2008).
Despite this, as of yet, exceptionally few researchers have opted to utilize long-term measurement, and even fewer organizations have disclosed year-over-year training effectiveness data or reports despite (a) the majority of organizations surveyed by SHRM offering sexual harassment prevention training (SHRM, 2018) and (b) various legal mandates to provide sexual harassment prevention training to employees broadly in California, Connecticut, Maine, and Michigan and to employees of state agencies in Florida, Illinois, Iowa, Nevada, New Jersey, North Carolina, Pennsylvania, Tennessee, Texas, Utah, and Washington. Even with this privately held abundance of data, public sharing of data remains scarce. It is possible given the secrecy surrounding data and organizational information (Costas & Grey, Reference Costas and Grey2014) that one reason of many for failure to disclose data is that the data may show that current training programs or methodology are simply not working. Alternately, it is possible that organizations fear harm to their reputations or recruiting efforts should the data they disclose mimic the results found on college campuses, especially if there is little improvement year over year. Regardless of the motivation to withhold information, without data it is impossible to pinpoint common problems or their solutions. By drawing attention expressly to the necessity of long-term data collection here, we hope to spur the use and disclosure of multiwave assessments in organizations. Doing so will aid in piecing together an understanding of what is working (and what is not) while bringing the full scope of this issue into public focus.
Measurement recommendations
As this review highlights, the evaluation of sexual assault prevention training continues to be hampered by several measurement and methodology concerns. Although withholding training effectiveness information is a critical issue, for this section, we focus our attention on a widespread issue uncovered through sexual assault prevention training research on college campuses: the misalignment between the constructs being measured and the intended outcomes of training. Discrepancies between both knowledge and attitudes (Lonsway & Kothari, Reference Lonsway and Kothari2000) and short-term versus long-term attitudes (Black et al., Reference Black, Weisz, Coats and Patterson2000; Breitenbecher, Reference Breitenbecher2001; Foubert & Marriott, Reference Foubert and Marriott1997; Heppner et al., Reference Heppner, Humphrey, Hillenbrand-Gunn and DeBord1995, Reference Heppner, Neville, Smith, Kivlighan and Gershuny1999; Lonsway & Kothari, Reference Lonsway and Kothari2000) suggest this is a problem regardless of the outcome of interest.
To avoid the concerns associated with self-reported attitudes and behavior, it may be fruitful to employ well-developed situational judgment tests (SJTs), which provide more insight into true attitudes and behavior (Chan & Schmitt, Reference Chan and Schmitt2002). An SJT variant already in use, the Comprehension of Consent/Coercion Measure (Gibson & Humphrey, Reference Gibson and Humphrey1993), presents respondents with complex vignettes describing sexual interactions with elements of coercion and pressure. This measure, which parallels case-based learning, requires respondents to not only demonstrate that they understand the difference between consent and coercion but also why a particular scenario is coercive. This type of low-fidelity simulation may mimic problem-solving processes involved in knowledge application that have been linked to long-term transfer (Halpern & Hakel, Reference Halpern and Hakel2002) and increase the likelihood that trainees will retrieve and apply learned knowledge in novel circumstances. Therefore, to measure employee knowledge acquisition and application, we suggest adapting this measure using organizational scenarios (e.g., unequal power dynamics and unwanted romantic advances or overhearing inappropriate comments about a coworker).
When topics are sensitive in nature, it is simply more difficult to accurately measure attitudes (Taylor, Reference Taylor1961). Many attitudinal measures contain overt items (e.g., “Rape happens when a guy’s sex drive goes out of control”; Payne et al., Reference Payne, Lonsway and Fitzgerald1999), and, as a result, respondents implicitly understand they should answer a particular way on these items, despite, perhaps, their own contradictory opinion. Furthermore, few studies (e.g., Schewe & O’Donohue, Reference Schewe and O’Donohue1996) account for socially undesirable responding before drawing conclusions about training effectiveness. To address this issue, sexual assault prevention training initiatives may draw inspiration from other sensitive topic areas, including sexism, racism, and alcoholism. Researchers within these topic areas have developed subtle measures of attitudes, including incorporating modern contexts (Swim, Aikin, Hall, & Hunter, Reference Swim, Aikin, Hall and Hunter1995) and considering latent response time (Lindgren et al., Reference Lindgren, Neighbors, Teachman, Wiers, Westgate and Greenwald2013). Some scales, like the Symbolic Racism Scale (Henry & Sears, Reference Henry and Sears2002), do not provide clear right or wrong response options and instead include some items that take the form of discussion questions (e.g., “How much discrimination against blacks do you feel there is in the US today?”). Previous work has noted that open-ended questions have the potential to more fully assess attitudes (Esses & Maio, Reference Esses and Maio2002; Haddock & Zanna, Reference Haddock and Zanna1998). With respect to attitudes about sexual harassment, questions asking if trainees believe sexual harassment is a problem at their workplace could prove useful for both understanding training outcomes and future training design.
Some researchers have also worked to incorporate the full dimensionality of attitudes into measurement instruments, which is also an important challenge for measuring sexual assault and harassment outcomes. The Ambivalent Sexism Inventory, for example, includes items that speak to sexist attitudes beyond typical manifestations (e.g., “Women should be cherished and protected by men”; Glick & Fiske, Reference Glick and Fiske1996). Finally, although not appropriate in every context, latent response time has shown promise in measuring some socially undesirable identities and predicting related behaviors. For instance, the Drinking Identity Implicit Association Test (Lindgren et al., Reference Lindgren, Neighbors, Teachman, Wiers, Westgate and Greenwald2013) measures the extent to which respondents associate the terms “drink” with “me.” This pairing resulted in a strong, consistent predictor of alcohol consumption problems as well as alcohol cravings. Given these results, it may be worthwhile to develop and validate an implicit association test focused on sexual harassment and assault attitudes.
Research investigating the impact of cognitive biases on ethical decision making and behavior also informs means through which sensitive personal information can be gathered via proxy measures. In short, individuals are more likely to respond in ways that downplay their own negative, questionable, or socially undesirable attitudes and behavior when asked to self-report information (i.e., self-enhancement effects; Brown, Reference Brown1986; Manley, Russell, & Buckley, Reference Manley, Russell and Buckley2001). However, when predicting the attitudes and behaviors of others, individuals are likely to engage in another type of cognitive bias: false consensus (Dawes & Mulford, Reference Dawes and Mulford1996). False consensus effects occur when individuals perceive their own attitudes and behaviors are representative of the broader majority, especially in the absence of information that would lead one to draw conclusions otherwise—conditions that can easily manifest when data and information secrecy are the norm. Considering the issues associated with self-reporting sensitive information, asking trainees to predict how others would behave in certain workplace scenarios may provide a more accurate estimate of what the trainees themselves think about sexual harassment or how they might behave. In fact, Balcetis and Dunning’s (Reference Balcetis and Dunning2013) research supports that conclusion, finding that individuals are better at predicting behavior when asked about what others might do rather than relying on enhanced or guarded judgments about their own behavior.
Perhaps the most challenging element of measurement in this space is not the measurement instrument at all but rather the need to measure attitudes and behavior in organizational cultures that explicitly or implicitly promote silence. SHRM’s (2018) sexual harassment survey finds that of nonmanagerial employees who anonymously disclosed that they experienced sexual harassment over the last 12 months, 76% did not report the incident in their workplace. When asked why this information was not officially reported, victims pinpointed fear of retaliation against them, downplayed the severity of the incident, and thought little or no recourse was available to them. Given the deluge of reports about inappropriate comments and behavior of organizational leaders and powerful figures, sometimes extending back decades, and whisper networks and secret lists shared informally to aid in avoiding certain individuals due to potential backlash, it would appear that such concerns were founded. However, in light of attitude change in broader society about sexual harassment and related issues (Saad, Reference Saad2017) following the increase in sexual harassment and assault reporting, we believe that now is an opportune time to re-examine and reconsider measurement and climate issues at every organization, especially in light of another finding in SHRM’s sexual harassment survey: nearly 60% of HR professionals believe that unreported sexual harassment happens a “small extent” of the time.
With this in mind, we urge organizational leaders and HR professionals to evaluate their own organizational attitudes, cultures, and policies to elucidate what is truly prioritized in their organizations and how such priorities may be affecting their training efforts. For instance, in Faragher v. City of Boca Raton and Burlington Industries, Inc. v. Ellerth, the Supreme Court dictated that training regarding sexual harassment policies must be offered in order for organizations to avoid liability issues in sexual harassment and hostile work environment cases. If avoiding legal liability is the guiding principle behind prevention training, then training content, evaluation methods, and disclosure of training outcomes will reflect a “check the box” mentality. Namely, that is does not matter if training transfers to the workplace and is effective in the long run; it only matters that some form of training occurred at all. We explore the critical role of transfer climate further in the following sections.
Transfer climate
The environment outside of the direct training context plays a significant role in facilitating training transfer (e.g., Goldstein & Ford, Reference Goldstein and Ford2002; Tracey, Tannenbaum, & Kavanagh, Reference Tracey, Tannenbaum and Kavanagh1995). Thus, to truly encourage transfer, a more systems-based approach aligning multiple environmental elements to support the training effort must be taken. In their review of organizational training efforts and transfer, Burke and Hutchins (Reference Burke and Hutchins2007) found four environmental factors to have a moderate to strong relationship with training transfer: (a) climate, (b) leader support, (c) peer support, and (d) opportunity to perform.
Climate
Climate encompasses the shared perceptions of the organizational setting, including the organization’s policies, practices, and procedures (Schneider & Reichers, Reference Schneider and Reichers1983). Further, climate includes employees’ expectations regarding rewards and punishment for particular behaviors. When an organization possesses a positive climate, it provides opportunities to practice the newly learned skills, feedback on related performance, and cues or goals to use the newly learned behaviors (e.g., Ford, Quinones, Sego, & Sorra, Reference Ford, Quinones, Sego and Sorra1992; Tracey et al., Reference Tracey, Tannenbaum and Kavanagh1995). The notion of practicing skills with combatting or eliminating sexual harassment may sound counterintuitive. Indeed, creating an environment where organizational members are cued to use the new behaviors directly contradicts the typical goal of sexual harassment training. However, practice could be incorporated by introducing follow-up or coaching sessions. This underscores the importance of preparing organizational leaders to facilitate the application of skills that may be accomplished by providing worksheets, discussion points, or specific training geared toward implementation.
Further, how an organization responds to instances of sexual harassment may provide feedback to employees. For instance, Fitzgerald, Drasgow, Hulin, Gelfand, and Magley (Reference Fitzgerald, Drasgow, Hulin, Gelfand and Magley1997) found cultures that were lax in punishing sexual harassment resulted in greater sexual harassment. Thus, appropriately punishing perpetrators, following through with investigations, and praising those who come forward may signal desired post-training behavior. Sweeping instances of sexual harassment under the rug undermines training efforts by indicating that this climate supports perpetrators and places a higher value on outcomes such as profit and performance. In turn, this response may allow sexually harassing behaviors to flourish. If organizations truly want to make a difference, they must ensure that the climate signals that the newly learned knowledge, skills, and behaviors are desired, promoted, and rewarded.
Leader support
In a modeling effort to examine the social context, individual factors, processes, and transfer, Chiaburu, van Dam, and Hutchins (Reference Chiaburu, van Dam and Hutchins2010) found significant support for the role of the supervisor in training transfer (see also Blume, Ford, Baldwin, & Huang, Reference Blume, Ford, Baldwin and Huang2010). Specifically, they found that supervisory support influenced self-efficacy, learning goal orientation, and motivation to transfer. Supervisory support, in the organizational sense, might involve managers providing support, guidance, or access to needed resources related to sexual harassment (Scaduto, Lindsey, & Chiaburu, Reference Scaduto, Lindsay and Chiaburu2008). Further, research in a military setting suggests that leaders play a particularly important role in subordinate behavior and organizational culture with regard to sexual harassment and assault (Sadler, Booth, Cook, & Doebbeling, Reference Sadler, Booth, Cook and Doebbeling2003; Schaubroeck, et al., Reference Schaubroeck, Hannah, Avolio, Kozlowski, Lord, Treviño and Peng2012). Specifically, leaders directly influence subordinate behavior by role modeling acceptable actions (e.g., DoD, 2014) and signaling acceptable and unacceptable behaviors based on what they tolerate and punish (e.g., Sadler et al., Reference Sadler, Booth, Cook and Doebbeling2003). Additionally, leader behaviors shape an organization’s culture that, in turn, influences subordinate behavior (e.g., Schaubroeck, et al., Reference Schaubroeck, Hannah, Avolio, Kozlowski, Lord, Treviño and Peng2012). Thus, research in a military setting suggests that leaders play a key role in the sexual harassment and assault behaviors of their subordinates. Along these lines, Sadler, Lindsay, Hunter, and Day (Reference Sadler, Lindsay, Hunter and Day2018) discuss transformational, transactional, and laissez-faire leadership in relation to sexual harassment and assault in the military, arguing that transactional leadership may lead to compliance but that creating a culture that resists harassment and assault requires transformational leadership. Leaders, then, must reflect on their behaviors and practices, model appropriate behavior and standards, and respond to issues of harassment and assault in a timely manner that aligns with desired behavior in order to encourage training transfer and reduce instances of sexual harassment and assault.
In a university setting, such responsibility may fall, in part, on faculty, staff, and administration. Recent Title IX legislation extends the law such that faculty are responsible for reporting any known sexual violence experienced by a student (Engle, Reference Engle2015). This change in legislation places faculty members in a leadership role that extends beyond required reporting and into providing social support via on-campus resources and contacts for students struggling with their own experiences or how to process training content. In less traditional or central organizations such as Hollywood movie studios or the US government, it becomes important to establish allies, or men and women in powerful positions to emphasize and support the reduction of sexual harassment. The Time’s Up movement in Hollywood provides a good model for others to follow in this regard.
Peer support
Previous research suggests the importance of peer support, or the extent to which peers support the application of new skills on the job (Holton, Bates, Seyler, & Carvalho, Reference Holton, Bates, Seyler and Carvalho1997) in facilitating training transfer (e.g., Facteau et al., Reference Facteau, Dobbins, Russell, Ladd and Kudisch1995). In fact, Martin (Reference Martin2010) and Hawley and Barnard (Reference Hawley and Barnard2005) found that peer support could compensate for a negative transfer climate. In an organizational setting, peer support may be expressed through regular meetings and discussions regarding the training content (e.g., Martin, Reference Martin2010). In a university setting, student advocates may be encouraged to discuss the training content outside of the formal educational setting. Placing these advocates in dorms, clubs, and Greek life may further perpetuate these discussions and provide more opportunities to interact with fellow students. Similarly, student volunteers may be requested and may pledge to be student advocates for sexual assault prevention. It is noteworthy to again mention the importance of providing the necessary resources and training for those in advocate roles. For example, student advocates should receive training on how to facilitate discussion and how to help other students practice their newly learned strategies.
Opportunity to perform
Related to the idea of practice opportunities is the finding that trainees must have the opportunity to perform their newly learned skill or knowledge on the job in order for it transfer (e.g., Clarke, Reference Clarke2002; Lim & Morris, Reference Lim and Morris2006). This is important particularly in the instance of sexual harassment and assault training. In this instance, providing trainees with a direct opportunity to perform is dangerous, may be traumatizing, and ultimately, directly contradicts the purpose of the training effort. However, follow-up or refresher sessions soon after the training may help trainees continue to practice without placing them in danger. Similarly, providing guided discussions or activities for leaders to engage in with their subordinators may continue the conversation and provide trainees with the opportunity to practice in their day-to-day work environment.
Conclusion
The present effort provides several key suggestions regarding needs analysis, training design, evaluation, and post-training support for those designing and implementing sexual harassment and assault training efforts. As a whole, we suggest a proactive training approach that includes a strong focus on climate elements and mindfully adapts individual assessment to account for measurement of socially undesirable attitudes and behavior. To develop training that transfers, we call upon organizational leaders to critically reflect and probe discrepancies between their values and what is rewarded, prioritized, and valued in practice. To hold organizations accountable and to promote a broader understanding of when training works (and when it does not), we also call on organizations to be more transparent about their training content, methodology, and effectiveness. Although recent events have brought much needed attention to how widespread sexual harassment is at work, there remains much work to be done in developing and supporting training programs that result in meaningful change. As such, we hope the present effort brings a discussion of how I-O psychology can contribute to the current need for well-designed and effective sexual harassment and assault training efforts.
Author ORCIDs
Kelsey Medeiros 0000-0002-7388-3970
