Ignorance Is Not Bliss: A U.K. Study of Sexual and Domestic Abuse Awareness on Campus, and Correlations With Confidence and Positive Action in a Bystander Program

Abstract

This article reports on a U.K. research study encompassing two surveys which used evidence-based scales of awareness, confidence to intervene, and intervention opportunities and action regarding sexual and domestic abuse on campus. They were sent to all first-year incoming undergraduates (>n = 7,000) at one post-1992 U.K. university and received n = 1,604 responses. The study finds that survey respondents demonstrated low awareness of sexual and domestic abuse as a problem on campus. In the analysis of Survey 2, respondents were divided into three groups, those receiving active intervention, passive intervention, and no intervention from a university social norms marketing campaign challenging abuse on campus. The study drives the field forward by considering how confidence to act mediates the relationship between awareness and positive action. It finds associations between active intervention and raised awareness that is not noted in passive or no intervention. Active intervention potentially brings together the mediating variable of confidence where awareness + confidence = positive action. This article makes recommendations for first-year incoming undergraduates to receive awareness raising information about sexual and domestic abuse, prior to coming to university. Universities may also consider working with schools to counter a lack of awareness, which may emanate from normalization discourses learnt prior to coming to university and perpetuated once there. Managing low awareness of sexual and domestic abuse should be a priority of bystander programs and some form of active intervention is potentially beneficial as early as possible in university student journeys.

Introduction

It is well established in international studies that the ignorance barrier is a strong obstacle to effective bystander intervention against sexual abuse (SA) and domestic abuse (DA) ¹ on campus (Banyard, 2008; Banyard & Moynihan, 2011; Burn, 2009, 2017; Kania & Cale, 2018). For example, Burn (2009) surveyed 378 female and 210 male undergraduate students in a U.S. university and found failure to notice situations as high risk was one of the strongest prevention barriers. Burn (2017) also identified that “due to inadequate knowledge, bystanders may misdiagnose the situation and believe intervention is unnecessary” (p. 1). Recognizing problematic behaviors and identifying those behaviors as intervention appropriate represent the first two stages of Latané and Darley’s (1968, 1970) situational model of helping, which underpins bystander theory. Bystander programs, which derive from this theoretical basis, seek to work with program participants to help them explore potentially problematic behavior, and increase diagnostic abilities, therefore managing the ignorance barrier.

Anderson and Whiston (2005) identify information, empathy, socialization, and risk reduction as four content areas found in successful prevention programs. This article explores the content area of “information,” through considering the awareness that first-year incoming undergraduates have regarding what constitutes SA and DA, and the impact of this upon confidence and positive action as bystanders. This article will accomplish this by reporting on two surveys, which used evidence-based scales of SA and DA to measure awareness, confidence to intervene (referred to as confidence hereafter), and intervention behavior/opportunities (referred to as positive action hereafter) (Prevention Innovations Research Centre, 2015).

To clarify the authors’ use of the term ignorance, it is useful to consider the proverb “ignorance is bliss”. This proverb denotes that when a person does not know about a problem, then they are less likely to be troubled by it or, in this instance, take action. It also denotes that it might be better not to know all the information regarding a situation. Instead, the authors of this article contend that having greater access to information is important, hence the title “ignorance is not bliss.” We do accept that the term ignorance may be perceived to have negative connotations, for example, denoting a general state of ignorance, or an inability to comprehend phenomenon differently in the light of new information. However, it is not used in this article in this way, instead it is used to denote that misinformation about what constitutes SA and DA is related to cultural contexts which normalize particular behaviors. We contend that new information has the potential to enable groups and individuals to comprehend phenomena differently and this article will emphasize the importance of interrupting low awareness regarding what constitutes SA and DA, evidenced as held by some first-year incoming undergraduates.

Literature Review

Prevalence, Normalization, and Awareness of SA and DA on Campus

Long-standing extant literature highlights the prevalence of SA and DA on campus as an international issue (Fisher et al., 2000; Krebs et al., 2007). For example, research in the U.S. estimates that one in five women experiences a completed or attempted sexual assault during their college years (Krebs et al., 2007). Underpinning recent data in the U.S. are Campus Climate Surveys (CCS) measuring prevalence, attitudes, and behavior. Although CCS are not without criticism, for example, lacking consistency and low response rates, they give broad statistical measures indicative of high prevalence rates, statistics not yet available to the same extent in the U.K.

Fenton and Mott (2017), who currently reference the U.K. Nation’s work on prevention and intervention, note that U.K. prevalence data have been less established than the U.S., but prevalence statistics are now beginning to emerge which represent similar patterns as U.S. institutions. U.S. and U.K. universities have similar contexts in that they contain (mostly) young adult populations in concentrated masses, often living away from home for the first time, and who are learning how to negotiate unfamiliar areas where alcohol, drugs, consent, peer pressure, and university environmental social norms impact. U.S. and U.K. universities are mired in similar traditions with similar social norms and stereotypes which can underpin attitudes and behaviors underlying SA and DA prevalence. U.K. university focus on campus SA and DA began to be highlighted more significantly by the National Union of Students (NUS, 2011) who noted that 25% of female students reported having experienced sexual assault. In 2017, Revolt Sexual Assault distributed a survey investigating students’ experience of sexual violence, receiving 4,491 responses from 153 U.K. institutions and found that 48% of female students and recent graduates surveyed and 17% of male students and recent graduates surveyed experienced some form of sexual assault (Revolt Sexual Assault, 2018).

Schwartz et al. (2017) note that, despite high prevalence rates, false beliefs often lead to denial or minimization of the impact of SA. Myths surrounding what constitutes DA may similarly contribute to minimizing or denying that DA has taken place (Bagwell-Gray et al., 2015). In the U.K., a similar normalization and minimization discourse has emerged in research with students, where “35% agree that there is a belief that sexual assault and harassment are ‘not a big deal’” (Revolt Sexual Assault, 2018, p. 5). The U.K. Women and Equalities Committee (2016) Report for House of Commons finds evidence of normalization and minimization discourses and beliefs, emanating from students in schools such that young people are “learning social norms that are carried through to adult life” (p. 15).

Research on Bystander Programs: The Barrier of Ignorance

This article draws on survey research from an institution that has adopted a social norms campaign approach to prevention, part of which included aspects of the intervention initiative (TII) developed by Public Health England (PHE). “The Intervention Initiative (hereafter referred to as TII) (Fenton et al., 2014), became the first evidence-based bystander program for the sector” (Fenton & Mott, 2017, p. 452). The TII was developed from programs prominent within the U.S. (e.g., “Bringing in the Bystander”; Banyard et al., 2004). Many such programs integrate social norms theory (Berkowitz, 2009, 2010) whereby positive bystander intervention, over time, can shift social norms and lead to cultural change.

Research on bystander programs on U.K. campuses is limited, though a body of work is emerging (e.g., Fenton et al., 2016; Fenton & Jones, 2017; Fenton & Mott, 2017, 2018). Bystander programs can have an impact on ignorance, but research is still in its infancy as to what makes these programs effective. Fenton and Mott (2017) clarify that “there is no evidence that knowledge . . . can in and of itself produce behavioural change” but that “knowledge is nonetheless a critical precondition to intervention” (p. 8). Yule and Grych (2017) find a link to confidence in their U.S.-based study, where “individuals who reported more confidence in their ability to intervene in general were more likely to intervene.” (p. 18). They do not consider how low or high awareness mediates this. This study will add to the field of knowledge by considering how awareness (or knowledge) and confidence mediate each other and potentially impact upon positive action.

To consider this further, this review will explore some of the theory underlying bystander and developed from Latané and Darley’s (1968, 1970) situational model of helping. In particular, the first two stages of this model will be considered: noticing an event and identifying it as intervention appropriate. If SA and DA have become normalized, as the research above suggests, then the first two stages of this model are less likely to occur. The concepts of pluralistic ignorance and false consensus can help understand ways in which awareness and confidence impact upon positive action in bystander prevention.

Pluralistic Ignorance and False Consensus

Latané and Darley (1968, 1970) represent the origins of exploring bystander help:

Their research into barriers and facilitators of individual helping led to the articulation of the broader situational model of bystander intervention: (a) notice the event, (b) identify the situation as intervention-appropriate, (c) take intervention responsibility, (d) decide how to help, and (e) act to intervene. (Banyard et al., 2018, p. 3)

Burn (2009) identified five barriers to intervention, among which were failure to notice a situation as needing intervention and failure to identify the situation as high risk and appropriate for intervention, the first two stages of the situational model above. Banyard and Moynihan (2011) also explored barriers to intervention noting ignorance as the most common barrier.

The barrier of ignorance is also noted in U.K. research where Fenton et al. (2016) consider the mutually reinforcing nature of pluralistic ignorance and false consensus. They note pluralistic ignorance as occurring where individuals misunderstand or misperceive others’ desire to intervene, and thus the individual wrongly believes “that their own desire to intervene must be misplaced” (p. 19). Other influencing factors to intervention are also noted, such as “perception of danger.” When problematic behavior is left unchecked in this way, this can, in turn, lead to those engaging in the problematic behavior to believe that this behavior is acceptable and that others behave like this too, leading to a false consensus emerging. Burn (2009) also drew attention to the issue of pluralistic ignorance:

When situations are ambiguous, people rely on others’ reactions to help them decide what to do. Pluralistic ignorance occurs when ignorant, inactive bystanders look to other ignorant, inactive bystanders and consequently all fail to identify the situation as intervention appropriate. (p. 781)

This article contends that if a critical mass of students comes to university with low levels of awareness of what constitutes SA and DA, then a potential impact of this can be the development of inactive bystanders emerging, rather than prosocial bystanders. Therefore, bystander programs crucially need to pay attention to this lack of awareness of SA and DA, as, in such a situation of pluralistic ignorance, step one and step two of Latané and Darley’s (1968, 1970) model are at risk of being missed by many. Potentially fewer and fewer individuals move on to step three of taking intervention responsibility, leading to the mutual reinforcement of pluralistic ignorance and false consensus emerging in whole populations of students. A first stage in bystander intervention needs to be breaking down the ignorance barrier which might be a result of normalization discourses (NUS, 2011; Revolt Sexual Assault, 2018; Schwartz et al., 2017).

Banyard (2008) and McMahon (2010) found that those who participate in programs or courses which discuss the arena of sexual violence are more likely to intervene and less likely to be ignorant to SA and DA behaviors. Burn (2009) further corroborates this, “knowledge reduces this barrier because bystanders are better able to identify when others are at-risk” (p. 781). In terms of confidence or self-efficacy to intervene, there is a wide body of research finding that those who participate in bystander interventions, for example, “Bringing in the bystander” in the U.S. (Banyard et al., 2009, 2007), demonstrate increased confidence and self-efficacy scores as a result of participation.

This study will explore levels of awareness of SA and DA on campus and whether awareness on its own leads to increased positive action. In addition, it will consider whether higher confidence to intervene, or self-efficacy, leads to increased positive action and how the two variables of awareness and confidence might mediate one another. It will also examine whether participating in some form of active bystander intervention strategy may have associations with raised levels of awareness of what constitutes SA and DA, as compared with no intervention (NI) or passive intervention (PI).

Method

Research encompassed two surveys devised through an electronic survey platform, using evidence-based scales measuring aspects of SA and DA awareness, confidence, and action. These were sent via email to all first-year incoming undergraduates (>n = 7,000) at the beginning of Semesters 1 (September 2018) and 2 (January 2019), at one post-1992 U.K. university. The surveys received a total of complete and partial responses of n = 1,604.

Sampling and Bias

The sampling frame was all first-year incoming undergraduate university students to the post-1992 university, registered at the start of the academic year 2018. Surveys 1 and 2 are not test–retest match samples, and populations are treated as separate, though it is possible that some participants may have answered both surveys.

Survey demographics are broadly similar to national U.K. university populations allowing for generalizations beyond this study. Most participants of Surveys 1 and 2 were in the 18 to 21 age group: 75.28% in Survey 1 and 77.8% in Survey 2. This is representative of national U.K. participation statistics with the majority of first-year undergraduates below 21 years old (Universities UK [UUK], 2018).

Gender was broadly representative, with 65.54% of participants in Survey 1 being female and 33.04% male, and 68.3% in Survey 2 being female and 30.9% male. Although national university statistics are more evenly distributed at 57.5% female and 42.5% male in the undergraduate U.K. university population in 2016–2017 (UUK, 2018), the proportion of males to females is similar enough to be broadly representative. It is also reflective of usual patterns of gender bias toward females in answering surveys with regard to sexual violence (Rosenthal & Freyd, 2018).

The survey was broadly representative of U.K. and non-U.K. student participation nationally. Survey 1 participant response was 78.21% U.K. students and 21.79% non-U.K. students; in Survey 2 it was 84.3% U.K. students and 15.7% non-U.K. students. National figures of undergraduate participation in U.K. universities in 2016–2017 were 80.9% U.K. students and 19.1% non-U.K. students (UUK, 2018).

In addition, Rosenthal and Freyd (2018) offer an evidence base from their study that contradicts bias in self-reporting toward those who have a vested interest to do so. Their research of CCS compared a self-selection sample with a sample that minimized self-selection, finding that it was just as plausible that victims of campus violence were at least as likely to be disinterested in participating as interested, therefore concluding that such studies did not seem to be biased either way.

Evidence-Based Measures

Survey 1 (distributed electronically in September 2018) and Survey 2 (distributed electronically in January 2019) both asked the same questions regarding awareness, confidence, and positive action (clarified in the section below).

Readiness to help: Awareness Subscale Short Form—scale used in Surveys 1 and 2

This relates to Prochaska and DiClemente’s (1984) stages for change and awareness. Participant responses ranged from a Likert-type scale of 1 (strongly disagree) to 5 (strongly agree). The full measure is located in Banyard et al. (2014) (Prevention Innovations Research Centre, 2015). This scale asked students whether or not they thought SA and DA are a problem on campus, did they need to think about SA and DA on campus, and could they do much about SA and DA on campus. There were a total of six questions. A mean score, after appropriate reversals, gives a single measure of awareness with increasingly greater scores representing increasingly greater awareness. Cronbach’s alpha for this scale in the sample was calculated to be .600.

Efficacy: Sexual Abuse Bystander Confidence Short Form—scale used in Surveys 1 and 2

Taken from the Bystander Efficacy Scale (Banyard, 2008), this scale assesses a person’s confidence in performing bystander behaviors. Participant responses can range on a sliding scale from 0 (can’t do) to 100 (very certain) (Prevention Innovations Research Centre, 2015). Examples of questions were as follows: get help and resources for a friend who tells me they have been assaulted or raped, and do something to help a very drunk person who is being brought upstairs to a bedroom by a group of people at a party. There were a total of five questions. A mean score gives a single measure of confidence in dealing with SA with increasingly greater scores representing increasingly greater confidence. Cronbach’s alpha for this scale in the sample was calculated to be .838.

Efficacy: Domestic Abuse Bystander Confidence Short Form—scale used in Surveys 1 and 2

As above, this scale assesses a person’s confidence in performing bystander behaviors. Participant responses can range on a sliding scale from 0 (can’t do) to 100 (very certain) (Prevention Innovations Research Centre, 2015). Examples of questions were as follows: talk to a friend who I suspect is in an abusive relationship, and speak up to someone who is making excuses for using physical force in a relationship. There were a total of four questions. A mean score gives a single measure of confidence in dealing with DA with increasingly greater scores representing increasingly greater confidence. Cronbach’s alpha for this scale in the sample was calculated to be .824.

Bystander behavior: Opportunity—scale used in Surveys 1 and 2

The fourth scale asks participants whether or not they had engaged in different types of bystander behaviors. Response choices were to tick N (no), Y (yes), or No Opportunity. No Opportunity was included to denote that the situation had not arisen (Prevention Innovations Research Centre, 2015). Examples of questions asked were as follows: I talked with a friend about sexual and/or domestic violence/abuse, or I saw a friend taking a very drunk person up to their room, and I said something and asked what the friend was doing. There were a total of 18 questions, allowing a count from 0 to 18 for each response type.

Types of Intervention: Questions Asked in Survey 2

Survey 2 (distributed electronically in January 2019) additionally asked survey respondents to identify if they had experienced any bystander program, activities, or information within the university, related to its first-year roll-out of a social norms campaign.

The University Social Norms Marketing Campaign underpinning this study consisted of events determined as passive intervention (PI) or active intervention (AI). PI denotes interventions where recipients receive information, usually in one-way transmission. AI denotes interventions where recipients actively engage with information, usually in two-way communication. Survey 2 respondents were divided into three groups based on their answers to these questions and according to interventions they noted as taking part in when responding to Survey 2. The three groups are PI, AI, and additionally no intervention or (NI). Those that noted taking part in both AI and PI were assigned to the AI group.

PI consisted of:

1. The University’s development of a series of short animated films exploring unwanted touching, consent, inappropriate use of social media, DA, and initiation and humiliation ceremonies. These films were marketed around the university on electronic screens and on the student intranet.

2. A short optional induction event delivered by professional services staff, lasting approximately 20 minutes, introducing these films, and giving information to students, including report and support services, and statistics on prevalence data in U.K. universities. This induction took place in the first few weeks of Year 1, was delivered to large lecture-style groups of students, and reached approximately n = 5,000 first-year incoming undergraduates.

3. Optional attendance at a one-off panel discussion group against violence and abuse (number of attendees not recorded).

4. Optional attendance at any of a series of short discussion groups throughout the year, gathering student views on the campaign (number of attendees not recorded).

AI consisted of:

1. Optional participation in a 2-hr bystander initiative (Bovill et al., 2018), developed from the original 8-hr TII (Fenton et al., 2014). Including exploration and activities around prevalence, impact, and consequences of SA and DA for “victims” and “perpetrators,” consent, and discussion of types of bystander responses. This was optionally attended by approximately n = 130 first-year incoming undergraduates.

2. Optional participation in street harassment workshops lasting up to 1 hr and delivered by an outside agency on university premises (number of attendees not recorded).

3. Optional participation in consent action workshops lasing up to 1 hr and delivered by an outside agency on university premises (number of attendees not recorded).

The NI category is self-explanatory in that survey participants did not declare on the survey having taken part in any of the above interventions.

Ethics

The University Research Ethics Committee (UREC) granted ethical consent with conditions applied, some of which are discussed below. British Educational Research Association’s (BERA, 2018) ethical standards underpinned the research. A consent and information form detailing participant rights and details of the study and study objectives was included at the beginning of each survey. Consent was assumed if respondents ticked agree to the questions “Have you read and do you understand the above information?” and “Do you agree to take part in the research?” Respondents who ticked below 18 years of age were taken directly to the end of the survey as it was assured during the ethical application process that only those aged 18 or more would take part. The survey invited participants to create a unique identity code to facilitate data withdrawal up to one month after completion of each survey. Demographic questions included gender, sexuality, student status, language, religion/belief, relationship status, and living arrangements. Surveys provided contact details of the principal investigator, and support services within and outside of the university.

An issue that required consideration was that of incentive to take part. Krebs et al. (2016) have identified that a level of incentive of US$25 is effective in maximizing completion rates. This benefit effect outweighs perceptions of a compromise to anonymity. Further justification for the use of small incentives is made on the grounds that any minor losses to reliability or validity through potential “persuasion” to take part, rather than freely given participation, are outweighed by the higher return rates which strengthen reliability and validity. While remaining aware of ethical debates regarding use of incentives (BERA, 2018), on balance, it was considered that this process is likely to give more benefits than costs to the research in terms of a wider response rate. This argument was presented within the ethical application and incentive granted approval as follows. For each of the questionnaires, participants were invited to opt in to a randomized draw to receive a £25 Amazon voucher. For each survey, there was the opportunity of four £25 Amazon vouchers to be won. This was made clear in the introductory email to the incoming first-year undergraduates and in the consent sections of the questionnaires. Implications of this decision are further considered in the “Limitations” section.

Analysis

All survey responses were screened for validity and examined for the extent and pattern of any missing data. In Surveys 1 and 2, a total of n = 1,604 logged onto the survey platform across the two surveys to provide some demographic data. Both surveys examined a subsample of n = 560 (Survey 1) and n = 658 (Survey 2) totaling n = 1,218 (total respondents) who provided responses relating to awareness of SA and DA as a problem on campus. Reasons for the gap between those logging onto the survey and completion are considered in the “Limitations” section.

Surveys 1 and 2 were descriptively analyzed to provide a baseline of overall awareness of SA and DA as a problem on campus. Survey 2 was analyzed in more detail using inferential statistics. For this analysis, a subsample of Survey 2 (n = 626) provided some responses relating to the scales of measurement. The analysis is based on those providing some scale data with prior reasoned discrete nonnormal outcome data with limited range. For these reasons, the nonparametric Kruskal–Wallis test was used for an omnibus assessment of between-group differences followed by a post hoc application of the Mann–Whitney Wilcoxon statistic for pairwise comparisons. Across groups, the correlations between measures were assessed and the Baron and Kenny mediation model was used to examine whether measures of confidence mediated the relationship between awareness and positive action (Baron & Kenny, 1986).

A missing data analysis was performed on those not providing any responses to awareness of SA and DA as a problem on campus. This subgroup did not have any discernible differences in demographic characteristics compared with those who did provide data on awareness of SA and DA on campus. Multiple imputation (20 imputed data sets) using MICE (multiple imputation by chained equations) and the resulting analysis did not produce differing statistical conclusions from those obtained from the subsample of Survey 2 (n = 626) who provided some responses relating to the scales of measurement. For brevity of exposition, we therefore only report the Survey 2 data on the sample of n = 626.

Findings

Surveys 1 and 2: Awareness Regarding SA and DA as a Problem on Campus From First-Year Incoming Undergraduates

This section descriptively explores first-year incoming undergraduates’ awareness of SA and DA as a problem on campus at the beginning of their first semester of undergraduate studies (September 2018) in Survey 1 and then at the beginning of their second semester of first-year undergraduate studies (January 2019) without accounting for whether they received a form of intervention (identified above). Overall awareness of SA and DA for a population sample of incoming first-year undergraduate students is important to establish as it allows for a snapshot view which may be applicable to wider, similar populations (implications of sampling and bias upon representation have been considered above and will be further considered in the “Limitations” section). Establishing overall awareness in the total sample population also allows later analysis to check if there is a change or not in overall awareness, according to whether Survey 2 respondents took part in any intervention.

Table 1 summarizes Survey 1 (distributed electronically in September 2018). In response to SA, 30% of survey respondents strongly disagree or disagree that SA is a problem on campus. A total of 50.7% of survey respondents neither agree nor disagree that SA is a problem on campus. And 19.2% of survey respondents agree or strongly agree that SA is a problem on campus.

Table 1.

Survey 1: Awareness Regarding SA and DA as a Problem on Campus (Percentage and Number of Respondents Displayed in Each Category).

Read Each of the Following Statements and Indicate How Much You Agree or Disagree	1. Strongly Disagree	2. Disagree	3. Neither Agree nor Disagree	4. Agree	5. Strongly Agree	Total Respondents
SA is a problem on campus	15.7% (88)	14.3% (80)	50.7% (284)	14.6% (82)	4.6% (26)	560
DA is a problem on campus	18.0% (101)	15.9% (89)	52.9% (296)	9.64% (54)	3.6% (20)	560

Note. SA = sexual abuse; DA = domestic abuse.

In response to DA, 33.9% of survey respondents strongly disagree or disagree that DA is a problem on campus. In total, 52.9% of survey respondents neither agree nor disagree that DA is a problem on campus. And 13.2% of survey respondents agree or strongly agree that DA is a problem on campus.

Overall, these percentages demonstrate a low awareness of SA and DA as a problem on campus in Survey 1 (September 2018) alongside a high percentage of students in the unsure position.

Table 2 summarizes Survey 2 (distributed electronically in January 2019). In Survey 2 in response to SA, 40.2% of survey respondents strongly disagree or disagree that SA is a problem on campus. This represents a 10.2 percentage point increase in this category from Survey 1. A total of 43.2% of survey respondents neither agree nor disagree that SA is a problem on campus. This represents a 7.5 percentage point decrease in this unsure position from Survey 1. And 16.6% of survey respondents agree or strongly agree that SA is a problem on campus. This represents a 2.6 percentage point decrease in this category from Survey 1.

Table 2.

Survey 2: Awareness Regarding SA and DA as a Problem on Campus (Percentage and Number of Respondents Displayed in Each Category).

Read Each of the Following Statements and Indicate How Much You Agree or Disagree	1. Strongly Disagree	2. Disagree	3. Neither Agree nor Disagree	4. Agree	5. Strongly Agree	Total Respondents
SA is a problem on campus	17.2% (113)	23.0% (152)	43.2% (284)	14.0% (92)	2.6% (17)	658
DA is a problem on campus	25.5% (168)	25.8% (170)	40.1% (264)	7.1% (47)	1.4% (9)	658

Note. SA = sexual abuse; DA = domestic abuse.

In Survey 2 in response to DA, 51.3% of survey respondents strongly disagree or disagree that DA is a problem on campus. This represents a 17.4 percentage point increase in this category from Survey 1. A total of 40.1% of survey respondents neither agree nor disagree that DA is a problem on campus. This represents a 12.7 percentage point decrease in this unsure position from Survey 1. And 8.5% of survey respondents agree or strongly agree that DA is a problem on campus. This represents a 4.7 percentage point decrease in this category from Survey 1.

Overall, these percentage point changes from Survey 1 (September 2018) to Survey 2 (January 2019) demonstrate a lowering of overall awareness of SA and DA as a problem on campus in Survey 2 (January 2019) as compared with Survey 1 (September 2018). They also demonstrate a percentage point shift away from the unsure position toward strongly disagree or disagree that SA and DA are a problem on campus. The next section considers associations with NI, PI, and AI on awareness of SA and DA as a problem on campus.

Survey 2: Incoming First-Year Undergraduates Awareness, Confidence, Opportunities, and Actions Across the NI, PI, and AI Groups

This section explores the statistical null hypotheses of no difference between intervention groups on awareness, confidence, and positive action, and the corresponding statistical alternative hypotheses for between-group differences on awareness, confidence, and positive action.

Analysis using the Kruskal–Wallis test indicates that the distribution of awareness responses differ between at least two of the three groups (H = 10.223, df = 2, p = .006). A post hoc analysis using the Mann–Whitney Wilcoxon test indicates no significant difference between the PI and NI groups (p = .643). However, analysis using the Mann–Whitney Wilcoxon test indicates a statistically significant increased awareness in the AI group compared with NI (p = .003) and a statistically significant increased awareness when compared with the PI group (p = .003).

Analysis using the Kruskal–Wallis test indicates that the distribution of confidence in dealing with SA does not significantly differ between groups (H = 0.686, df = 2, p = .710). A post hoc analysis indicates no significant difference between the PI and NI groups (p = .533), no significant difference between the AI and NI groups (p = .675), and no significant difference between the two intervention groups (p = .464.). Similarly, analysis using the Kruskal–Wallis test indicates that the distribution of confidence in dealing with DA does not significantly differ between groups (H = 0.736, df = 2, p = .692). A post hoc analysis using the Mann–Whitney Wilcoxon test indicates no significant difference between the PI and NI groups (p = .770), no significant difference between the AI and NI groups (p = .482), and no significant difference between the two intervention groups (p = .385).

The median number of opportunities to intervene in the AI group is 4 compared with a median of 3 in the other two groups. However, the number of opportunities to intervene does not significantly differ between groups (H = 1.373, df = 2, p = .503). In contrast, the median number of reported positive actions in the AI group is 3 compared with a median of 2 in the other two groups. This difference is statistically significant (H = 7.204, df = 2, p = .027). A post hoc analysis using the Mann–Whitney Wilcoxon test indicates no significant difference between the PI and NI groups (p = .471) on actions. However, the reported number of positive actions is significantly higher in the AI group compared with NI (p = .007) and significantly higher than that in the PI group (p = .048).

Across the sample, awareness, confidence in dealing with SA, confidence in dealing with DA, and positive actions are mutually and positively correlated. Specifically, awareness is positively correlated with the number of positive actions (r = .122, df = 623, p = .002), with confidence in dealing with SA (r = .227, df = 618, p < .001), and with confidence in dealing with DA (r = .165, df = 609, p < .001). Confidence in dealing with DA and confidence in dealing with SA were strongly correlated (r = .681, df = 609, p < .001). In a multiple regression, both awareness and confidence are jointly and individually significantly related to the number of positive actions, and confidence in dealing with DA partially mediates the relationship between awareness and the number of positive actions (Sobel statistic Z = 2.16, p = .030). Similarly, confidence in dealing with SA partially mediates the relationship between awareness and the number of positive actions (Sobel statistic Z = 2.01, p = .044).

Discussion

The findings from this study add to the growing body of international research in this area. It also adds to the U.K. perspective, which has made fewer contributions so far. It presents a large-scale U.K. quantitative study on SA and DA prevention having (n = 1,604) partial and complete responses across Surveys 1 and 2. It surveys awareness of SA and DA as a problem on campus, of first-year incoming undergraduates, and builds on the U.K. work of Fenton and Jones (2017) and Fenton and Mott (2017, 2018).

This study finds, from the descriptive statistics presented in Tables 1 and 2, that overall percentages demonstrate that first-year incoming undergraduates have low awareness that SA and DA are a problem on campus when they begin their first year and semester of study in September 2018. In the beginning of Semester 2 (January 2019), this low awareness remains. In addition, in Semester 2, there is a percentage shift toward strongly disagree and disagree that SA and DA are a problem on campus, away from the unsure position of neither agreeing nor disagreeing that SA and DA are a problem on campus. Thus, as a whole population, not divided into the AI, PI, or NI group, the overall effect between Surveys 1 and 2 has been a lowering of awareness that SA and DA are a problem on campus, for first-year incoming undergraduate students.

In Survey 2, when we factor in those that have experienced AI and compare with those that have experienced PI or NI (as demonstrated in Table 3), there is evidence that awareness of SA and DA as a problem on campus is higher in the AI group but not in the PI or NI group. This suggests that AI is necessary to at least maintain levels of awareness that SA and DA are a problem on campus between students arriving at the beginning of Semester 1 in September 2018 and the beginning of their second semester in January 2019. It also evidences that we can accept part of the alternative hypothesis posed that there is a difference between groups and awareness.

Table 3.

Survey 2: Awareness, Confidence, Number of Opportunities, and Number of Actions by Group.

Group	N	M	Median	SD	Minimum	Maximum
Awareness
No intervention	369	3.190	3.167	0.6341	1.00	4.67
Passive	181	3.178	3.167	0.6577	1.33	5.00
Active	75	3.416	3.500	0.5414	1.67	4.33
Confidence in dealing with sexual abuse
No intervention	365	74.90	77.25	17.737	13.25	100
Passive	180	73.58	75.62	19.299	8.25	100
Active	75	75.10	79.25	19.809	2.25	100
Confidence in dealing with domestic abuse
No intervention	365	72.38	73.25	18.414	17.25	100
Passive	180	72.88	74.00	18.179	9.50	100
Active	75	70.89	71.87	18.554	1.00	100
Number of opportunities
No intervention	370	5.143	3	5.589	0	14
Passive	181	5.099	3	5.358	0	14
Active	75	5.653	4	5.557	0	14
Number of actions
No intervention	370	2.473	2	2.296	0	14
Passive	181	2.729	2	2.643	0	17
Active	75	3.573	3	3.251	0	17

This indicates that in some way AI may potentially challenge normalization discourses which are learnt prior to coming to university and perpetuated once there (Bagwell-Gray et al., 2015; Burn, 2017; Schwartz et al., 2017). Yule and Grych (2017) conducted surveys with 281 U.S. university students measuring bystander situations and efficacy for intervening, similar to this study design. They found that the most prevalent barrier to intervening in their study was a lack of responsibility, at the third level of Latané and Darley’s (1968, 1970) situational model of helping. This study emphasizes first-year incoming undergraduates’ lack of awareness about what constitutes SA and DA, potentially impacting at the first and second levels of the model (noticing and identifying situations as intervention appropriate). Therefore, this article makes important contributions to the field, as if students struggle with the first levels of intervention, this likely impacts on all the following levels.

The inferential statistical analysis carried out on Survey 2 and presented in Table 3 demonstrates that awareness on its own does not lead to increased positive action and this is in line with the U.K. work of Fenton and Mott (2017) on the impact of knowledge and bystander behavior. This study also notes that higher confidence (self-efficacy) to intervene does not on its own lead to increased positive action, thus adding to the U.S. work of Yule and Grych (2017). This study finds that it is the capacity for these two variables together which potentially leads to more positive action. It finds that AI potentially brings together the mediating variable of confidence where awareness + confidence = positive action. Therefore, this study moves the field forward by exploring the interplay of awareness and confidence on positive action. Already confident survey respondents who have raised awareness of SA and DA as a problem on campus are more likely to record positive action responses. In addition, AI is associated with raised awareness. Therefore, within the context of already confident survey respondents, this study accepts the alternative hypotheses that there is a difference between groups on awareness, confidence, and positive action.

The findings from this study both confirm and contradict other studies. The systematic review of Jouriles et al. (2018) found that students who participate in bystander programs compared with those that did not “have more pro-social attitudes and beliefs about sexual violence” and “engage in more bystander behaviour” (p. 463). McMahon et al. (2018) back this up in a longitudinal study of college students who received three active bystander interventions. One year after the intervention, 498 college students gave qualitative responses and “demonstrated a change in rape supportive attitudes” (McMahon et al., 2018, p. 13). In addition, they found that respondents demonstrated increased knowledge and awareness of circumstances in which SA might occur, paying more attention in social situations. These findings are consistent with the findings of this study. Jouriles et al. (2018) also found that although longer bystander programs were likely to lead to greater change, there were notable benefits to be found in shorter programs. This too is consistent with the study findings which implemented 1- or 2-hr active interventions, rather than the 8-hr intervention advocated by Fenton et al. (2014). It is important to note the potential of shorter intervention programs alongside longer intervention programs as university timetables for academic and nonacademic provision are always under pressure. In addition, it might be easier to persuade larger numbers of students to attend shorter sessions of intervention which require a smaller investment of their own limited resources.

In contradiction to the findings of this study, Jouriles et al. (2018) found that “regardless of method, positive effects for bystander programs emerged” (p. 463). PI in this study did not demonstrate raised awareness. This is important to note as it indicates that universities consider building in aspects of AI into bystander programs. This study suggests that PI may not be an optimal way to approach this, even though it might reach greater numbers of students and be cheaper to administer and may fulfill some of the recommendations emerging from guidance (UUK, 2016, 2018).

Limitations

As this study did not conduct pretest and posttest analysis, it is possible that the individuals who chose to engage in AI were those who were already higher in awareness resulting from a self-selection phenomenon, and this needs further research with pretest and posttest analysis in place. As such, this article is claiming associations between AI and raised awareness, rather than impact.

Missing data are considered and accounted for in the “Analysis” section; however, the gap between those who started surveys and the analysis (n) needs further clarity and there are three potential reasons why this may have occurred:

• Survey fatigue: Surveys were deliberately constructed to last 10 minutes or less in response time. However, first-year incoming undergraduates are deluged with information in the first months of university participation, so this may have been a factor.

• Triggering: The sensitive nature of the survey content may have resulted in survey respondents feeling unable to finish the survey as the questions became more sensitive. There is some evidence of a tail-off of completion toward the end of the surveys.

• Filling out demographic information to be entered in the prize draw: There is some evidence of a tail-off of completion once demographics were entered. However, as discussed in the “Ethics” section, this was a compromise considered worth taking to enhance the overall response rate. As discussed below (Saleh & Bista, 2017), response rate could be considered good and thus ameliorate this limitation somewhat.

Although these findings may represent important contributions to knowledge in a U.K. context and may have implications for other U.K. universities and to an international audience, replication studies would provide greater reliability. However, as discussed in the “Sampling and Bias” section, the survey demographics of survey respondents were broadly similar to national U.K. university populations in terms of age, gender, and U.K. and non-U.K. student populations and so generalization beyond this study is possible. In addition, Saleh and Bista (2017) note that it is difficult to get a good response rate at the start of term. Thus, the 11.2% response rate to Survey 1 and the 11.8% response rate to Survey 2 could be considered a good response rate, further supporting generalization beyond the immediate study.

The research has not conducted follow-up surveys so it is unknown whether the raised awareness and positive action from AI in this study lasts beyond the date of the survey; Jouriles et al. (2018) draw attention to this flaw in research. Many studies do not conduct follow-up research and those that do suggest that program effects diminish over time and this may be the case with this study. Follow-up surveys as with the work of McMahon et al. (2018) would strengthen the findings of this study. Fenton and Mott (2018) also demonstrate that concurrent campaigns have greater effect. Furthermore, although this study contradicts the findings of Jouriles et al. (2018) in that it finds that AI is potentially more beneficial than PI, this research does not tell us what aspects of AI are useful. For example, Kleinsasser et al. (2015) found that online intervention (often considered a more passive form of intervention through one-way transmission of information) is effective. However, their study considered a 20-minute online intervention where students were required to stop, think, and input ideas, which suggests an active element. Therefore, more research is needed to understand what elements of AI are effective and in what modes it can be delivered.

Finally, as with Jouriles et al. (2018) who find that most analyses of bystander find only small effects, the effects of this study were similarly small, for example, the median number of opportunities to intervene in the AI group was 4 compared with a median of 3 in the other two groups. The median number of reported positive actions in the AI group is 3 compared with a median of 2 in the other two groups. However, as Jouriles et al. (2018) note,

It is important to appreciate the power of small effects, especially when a program is disseminated to an entire campus. To offer an illustration, if a bystander program led to the performance of only one additional helpful bystander action a month for 20% of the students who received the program, at a college with 20,000 students this translates to approximately 12,000 additional helpful bystander behaviors over a 3-month period. Such an increase could make a huge difference in campus victimization rates and prove to be a valuable strategy for college-campus violence prevention toolkits (p. 463).

Recommendations

It seems particularly important that information reaches students prior to starting university, raising awareness of SA and DA as a problem on campus, before entry. Therefore, universities might consider the information that students receive in induction materials and welcome packs from offer onward. This is a high-risk strategy as universities vie for students and seek to strengthen and keep their reputations; however, some university surveys are already noting areas such as safety and ranked on these measures. For example, the Times Higher Education Experience Survey 2018 asked 20,000 students “the extent to which they agreed that their university offered ‘good security’ giving universities a ‘good security score’” (Times Higher Education, 2018, p. 1) noting that

Feeling safe while studying is a top consideration for many prospective students when they start thinking about where to go to university. Enjoying a sense of security, comfort and happiness during your time at college can make the difference between staying on and getting a good degree or deciding to look elsewhere or even drop out altogether.

Universities could market themselves as having a proactive response to SA and DA on campus, promoting this as a responsible safety measure, which may make particular universities a more attractive option to students and their parents/primary carers.

A further powerful measure and one that perhaps has less reputational damage attached to it is for universities to consider their potential role in working with schools so that students might come to university with greater awareness about SA and DA on campus, strategies for protection, and a mind-set that challenges cultural acceptance and normalization of abusive practices. The Women and Equalities Committee’s (2016) Report for House of Commons states that to tackle the problem of sexual violence “work should start much earlier, in schools” (p. 3). Beginning with effective education that tackles low awareness of what constitutes SA and DA in schools may mean that greater numbers of first-year incoming undergraduates arrive as new university students who are more prepared to prevent and intervene in situations of high-risk sexual behavior. Noticing and identifying are the first steps in Latané and Darley’s (1968, 1970) situational model of bystander help. Raising first-year incoming undergraduates’ awareness of what constitutes SA and DA is therefore critical, and this article finds AI to be a potentially more effective way of achieving this. The first few days, weeks, and months of student life and the information that students receive prior to coming to university may shape their acceptance of and resistance to SA and DA on campus throughout their whole student journey. It might also shape the cultural norms of university regarding SA and DA on campus challenging pluralistic ignorance and false consensus.

Author Biographies

Helen Bovill BA, PhD, is Associate Head of Department Research and Scholarship in the Department of Education and Childhood at the University of the West of England, Bristol. She worked with colleagues across the university, including students and the Student Union to develop a social norms marketing campaign against harassment and abuse—SpeakUp. She co-convenes the Department research group BRIDGE.

Paul White BSc, MSc, PhD, is an associate professor and director of the Applied Statistics Group at the University of the West of England, Bristol. He has provided ethical, methodological, and statistical support in more than 100 studies including areas in statistical methodology, psychology, health, and the bio-, social, and medical sciences.