National Academies of Sciences, Engineering, and Medicine
The Committee on the Impacts of Sexual Harassment in Academia
500 Fifth St. NW
Washington, DC 20001
Dr. Kevin M. Swartout, Georgia State University
Chair, Administrator-Researcher Campus Climate Collaborative
INTRODUCTION TO THE PROJECT
Per the consultant agreement, Dr. Swartout obtained data related to sexual harassment from the University of Texas System (UT), which collected student data using the ARC3 Campus Climate Survey. He then conducted a series of analyzes focused on understanding the effects of sexual harassment experienced by students majoring in areas related to science, engineering, and medicine. Results from an additional ARC3 survey implantation across the Pennsylvania State System of Higher Education (Penn State)—presented at the October 2017 Working Group meeting—are included in the report at key points for comparison purposes. Dr. Swartout did not have access to the raw Penn State data; therefore, all statistical analyses described in this report were conducted using only the UT climate data.
Dr. Swartout was well positioned to carry out these proposed tasks. He currently chairs the ARC3 group, which is a collaborative of sexual violence researchers and student affairs professionals who came together to respond to calls issued by the White House Task Force to Protect Students from Sexual Assault, particularly the need to identify the scope of sexual misconduct on college campuses. As chair, Dr. Swartout had led efforts to develop, test, and disseminate the no-cost campus climate survey of sexual misconduct used to collect the data that he proposes to analyze. Representatives of approximately 400 institutions of higher education have requested the ARC3 survey since September 2015, and more than 150 U.S. institutions have used the survey to collected campus climate data from their student populations.
The UT and Penn State campus climate data are well suited to help address the working group’s research questions. The UT climate survey included 13 state institutions of higher education across Texas. More information on the UT System campus climate survey and results can be found at https://www.utsystem.edu/sites/clase. The Penn State climate survey data includes data from the University Park Campus and the College of Medicine at the Hershey campus. More information on the Penn State System campus climate survey and results is at https://studentaffairs.psu.edu/assessment/smcs/.
The National Science Foundation’s definition of STEM fields was used for the purposes of this project. This definition includes fields of medicine, engineering, and the natural, computational and social sciences (e.g., psychology and anthropology). Additionally, the Working Group elected to include the field of public health as a STEM science. STEM students were further broken down into students of the sciences (e.g., biology, computer science, psychology), engineering (e.g., electrical, mechanical, petroleum), and medicine (i.e., M.D. students) for more fine-grained analysis.
Faculty/Staff Sexual Harassment
Overall, 3,831 students (20.0 percent) reported experiencing sexual harassment perpetrated by a faculty or staff member; 3,343 (17.4 percent) reported experiencing sexist hostility, 1,411 (7.7 percent) reported crude behavior, 595 (3.1 percent) reported unwanted sexual attention, and 240 (1.3 percent) reported sexual coercion. Table D-1 depicts the overall faculty/staff sexual harassment rates by student gender identity. Of note, incidence of sexual harassment by faculty or staff significantly differed as a function of gender, with high incidence rates among women and those who endorsed a gender other than male or female relative to the overall sample. This pattern held for three of the four subtypes of faculty/staff harassment: sexist hostility (chi-square = 248.29, p < .001), crude behavior (chi-square = 126.95, p < .001), and unwanted sexual attention (chi-square = 21.41, p < .001), but not sexual coercion.
Table D-2 depicts the overall faculty/staff sexual harassment incidence by student status (i.e., undergraduate student, graduate student, or medical student). Incidence of sexual harassment by faculty or staff significantly differed as a func
|Student Gender||Faculty/Staff Sexual Harassment|
|Female||9,548 (78.0%)||2,697 (22.0%)*|
|Male||5,685 (84.7%)*||1,025 (15.3%)|
|Another Gender||124 (53.7%)||107 (46.3%)*|
(chi-square = 225.35, p < .001; *standardized residual > 2.0)
|Student Status||Faculty/Staff Sexual Harassment|
|Undergraduate||10,520 (80.6%)||2,537 (19.4%)|
|Graduate (Non-Med.)||4,347 (80.0%)||1,088 (20.0%)|
|Medical Student||351 (63.2%)||204 (36.8%)*|
(chi-square = 80.16, p < .001; *standardized residual > 2.0)
tion of student status, with high incidence rates among medical students relative to the overall sample. This pattern held for sexist hostility (chi-square = 98.21, p < .001) and crude behavior (chi-square = 33.32, p < .001), but not unwanted sexual attention or sexual coercion.
Taken together, these findings indicate that gender identity and student status are both relevant factors in faculty/staff perpetrated sexual harassment incidence. Because female students were at greater risk for experiencing harassment, additional analyses focused on the female subsample to generate more specific implications for those students at greatest risk. Corresponding figures depicting rates of sexual harassment reported by the male subsample are presented for comparison purposes. Although the subsample that endorsed a gender other than male or female were also at increased risk for experiencing faculty/staff harassment, that subsample was too small for more fine-grained analysis.
Figure D-1 depicts the percentages of female students of each major who experienced different forms of sexual harassment by faculty or staff in the UT sample. Results of a binary logistic regression suggest that female medical students were 220 percent more likely than non-STEM majors to experience sexual harassment by faculty or staff (OR = 3.20, p < .001), and female engineering students were 34 percent more likely than non-STEM majors to experience sexual harassment by faculty or staff (OR = 1.34, p = .002).
This trend held for sexist hostility by faculty and staff: Female medical students were 235 percent more likely than non-STEM majors to experience sexist hostility by faculty or staff (OR = 3.35, p < .001). Female engineering students were 36 percent more likely than non-STEM majors to experience sexist hostility by faculty or staff (OR = 1.36, p = .002).
This trend partially held for crude behavior: female medical students were 149 percent more likely than non-STEM majors to experience crude harassment by faculty or staff (OR = 2.49, p < .001), but female engineering students were not significantly more likely to experience crude behavior.
Finally, there were no statistically significant differences in female students’ likelihood of experiencing unwanted sexual attention or sexual coercion as a function of their academic major. Figure D-2 depicts similar rates reported by women in the Penn State sample, and Figures 3 and 4 depict sexual harassment rates reported by men in the respective samples.
Outcomes of Faculty/Staff Harassment
Health and Safety. Female medical students who experienced any sexual harassment by faculty or staff, compared with those who had not, reported significantly worse physical (t = 2.88, p = .004) and mental health outcomes (t = 3.22, p = .001), and they reported feeling less safe on campus (t = 2.35, p = .020).
Female engineering students who experienced any sexual harassment by faculty or staff, compared with those who had not, reported significantly worse physical (t = 2.92, p = .004) and mental health outcomes (t = 2.83, p = .005), but there was not a significant difference in their feelings of safety on campus.
Female science majors who experienced any sexual harassment by faculty or staff, compared with those who had not, reported significantly worse physical (t = 2.92, p < .001) and mental health outcomes (t = 10.77, p < .001), and they reported feeling less safe on campus (t = 3.25, p = .001).
Female non-STEM majors who experienced any sexual harassment by faculty or staff, compared with those who had not, reported significantly worse physical (t = 10.14, p<.001) and mental health outcomes (t = 11.96, p < .001), and they reported feeling less safe on campus (t = 4.97, p < .001).
A series of 4(major) × 2(SH status) analysis of variances supported significant differences in physical health, mental health, and feelings of safety on campus as functions of both academic major status (non-STEM, Science/Technology, Engineering, and Medicine) and faculty/staff sexual harassment experience (Yes vs. No); however, the interactive effect of the two factors was nonsignificant for all outcomes. Figures D-5 through D-7 present means on each outcome for each group.
Academic Disengagement. Female engineering majors who experienced any sexual harassment by faculty or staff missed significantly more classes (t = 2.99, p = .003) and made more excuses to get out of classes (t = 3.78, p < .001) compared with female engineering majors who had not experienced sexual harassment by faculty or staff. These two groups did not significantly differ in how often they reported being late for class or doing poor work. The contrasts are depicted in Figure D-8.
Female medical students who experienced any sexual harassment by faculty or staff reported doing poor work significantly more often than female medical students who had not experienced sexual harassment by faculty or staff (t = 2.34, p = .02). These two groups did not significantly differ in how often they reported missing class, being late for class, or making excuses to get out of class. The contrasts are depicted in Figure D-9.
Female science majors who experienced any sexual harassment by faculty or staff reported missing class (t = 7.26, p < .001), being late for class (t = 9.03, p < .001), making excuses to get out of class (t = 6.20, p < .001),
and doing poor work (t = 7.30, p < .001), significantly more often than female science majors who had not experienced sexual harassment by faculty or staff. The contrasts are depicted in Figure D-10.
Female non-STEM majors who experienced any sexual harassment by faculty or staff reported missing class (t = 8.43, p < .001), being late for class (t = 10.07, p < .001), making excuses to get out of class (t = 8.69, p < .001), and doing poor work (t = 6.29, p < .001), significantly more often than female non-STEM majors who had not experienced sexual harassment by faculty or staff. The contrasts are depicted in Figure D-11.
A series of 4(major) × 2(SH status) analysis of variances support significant differences in reports of missing class, being late for class, making excuses to get out of class, and doing poor work as functions of both academic major status (non-STEM, Science/Technology, Engineering, and Medicine) and faculty/staff sexual harassment experience (Yes vs. No). In addition, the two factors interacted significantly to affect being late for class (F = 3.08, p = .01), but not the other outcomes. The contrasts and graphs presented above suggest the negative effect of faculty/staff sexual harassment on being late to class was larger for science and non-STEM majors than it was for engineering and medical students.
Among female STEM students, white (non-Hispanic) students collectively reported significantly higher incidence of sexual harassment by faculty/staff (chi-square = 24.68, p < .001) than students of another race or ethnicity (Figure D-13).
Among these students, however, there was a significant interaction between experiencing sexual harassment by faculty/staff and race/ethnicity on student perceptions of campus safety (F = 4.42, p < .001). As depicted in Figure D-14, students who experienced sexual harassment by faculty/staff and endorsed a race or ethnicity other than white (non-Hispanic) perceived their campus as less safe than the other female STEM students. There were no other significant interactions between race and sexual harassment experiences on health and safety outcomes.
|White||African American||AIAN||Asian||Biracial||Hispanic||Multiracial||Pacific Islander||Total|
- Overall, 20.0 percent of the students surveyed reported experiencing sexual harassment perpetrated by a faculty or staff member.
- Female students (22.0 percent) and students who endorsed a gender other than male or female (46.3 percent) had significantly higher incidence rates of sexual harassment by faculty/staff, compared with male students (15.3 percent).
- Female medical and engineering students both reported significantly higher incidence of sexual harassment by faculty/staff (medical: 47 percent, engineering: 27 percent), compared with students enrolled in another major (i.e., sciences, non-STEM).
- Female students who experienced sexual harassment, compared with those who had not, generally reported worse physical and mental health outcomes, feeling less safe on campus, and higher levels across various indicators of academic disengagement.
- Among female STEM students, although white (non-Hispanic) students reported greater incidence of sexual harassment by faculty/staff, students of color and white Hispanic students who experienced sexual harassment by faculty/staff generally perceived their campus as less safe than the other female STEM students.
METHODS APPENDIX (UT CLIMATE SURVEY)
Human Subjects Protection
The UT Austin Institutional Review Board (IRB) approved and oversaw this campus climate study (IRB approval No. 2015-09-0110). Other UT institutors also independently reviewed and approved the study procedures. The IRB proposal was submitted by the UT principal investigator and research team and shared with Dr. Swartout and the National Academies staff for review. The IRB proposal included the overall research protocol, amendments to the principal survey instrument, which included variations made on an institution-by-institution basis. All institutions were provided with a copy of the study protocols and IRB approval at the time of the study. Each institution tailored the instrument to their specificities and population (e.g., each institution was able to define their own list of programs of study). No research-related activities involving human subjects took place until the study was fully reviewed and approved by the UT Austin IRB.
Students’ privacy and confidentiality were protected at every step of the data collection and analysis process. Each institution’s registrar office provided a list of official student e-mail addresses. The UT principal investigator and research team used the Qualtrics online survey software platform to conduct the survey and store the sampling frame information. The survey data were initially stored in a separate database within Qualtrics while the survey was active. There was no link between the sampling frames and the survey data. The platform generated a unique URL for each eligible participant that was destroyed upon survey completion. The institutional registrar did not provide the UT research team with any additional identifying information, nor was identifying information collected with the sensitive survey data. Although e-mail addresses were collective to facilitate incentives, they were not linked to the sensitive survey data.
Informed consent information was presented to students on the first page of the survey. It included a written description of the study made available online to participants, external resources for students, and information on incentives, risks, and benefits of survey participation. After reviewing the informed-consent information, participants were able to click “yes” to participate in the survey. Participation was confidential and voluntary, and participants could choose to skip any question in the survey without penalty, discontinue survey participation, or stop and restart at any time.
The UT research team used the e-mail addresses provided by the institutional registrar offices to advertise the study to eligible students across institutions. This e-mail included an individualized hyperlink to the survey website. Additionally, the research team encouraged stakeholder groups at each institution to engage in survey recruitment. Each group was provided with templates for recruitment and promotional e-mails, fliers, and social media posts to help increase awareness of the study. Most institutions sent a recruitment/promotional letter to all students, faculty, and staff to announce the survey and express institutional support. Most institutions promoted the survey via social media (e.g., Facebook and Twitter). Social media posts included a general hyperlink to the survey website.
Individual institutions selected and funded incentives for their student participants. Incentives therefore differed across the UT institutions. Incentives included randomly selected drawings for parking passes, gift cards, athletic tickets, and cash prices. Participants could enter a given drawing by clicking on a link at the end of the survey, which took them to a separate incentives survey. This process separated participants’ sensitive survey data from their identifiable incentive information, which included their names and contact information. Incentive winners were selected by the individual institution stakeholder groups.
The research team successfully recruited 28,270 (12.4 percent) of the 228,710 students actively enrolled in the UT system. Of this, 17,959 (63.6 percent) identified as women, 9,934 (35.2 percent) as men, 230 as another gender identity (< 1.0 percent), and 120 did not respond to the gender identity item (< 1.0 percent). Furthermore, 6.1 percent of the students self-identified as African American, 17.1 percent as Asian, 2.3 percent as biracial, 39.6 percent as Hispanic, 1.1 percent as multiracial, 39.5 percent as white (non-Hispanic), and 4.9 percent as another unspecified race/ethnicity. Undergraduates made up a majority of the sample at 69.8 percent, followed by master’s students at 17.5 percent, doctoral students at 8.0 percent, medical students at 2.0 percent, and students in a number of post-baccalaureate or professional programs accounting for a total of 0.7 percent. For the present analyses, students were categorized into non-STEM (12,788, 45.2 percent), science and technology majors (11,069, 39.2 percent), engineering majors (3,157, 11.2 percent), and medical students (573, 2.0 percent). Of just the subsample of female students, 8,636 (49.4 percent) were non-STEM, 7,603 (43.5 percent) were science and technology majors, 939 were engineering majors (5.4 percent), and 304 (1.7 percent) were medical students. Students who had not yet
declared a major at the time of the study (2.4 percent) were excluded from the present analyses.
Faculty/Staff-Perpetrated Sexual Harassment. The Sexual Harassment by Faculty/Staff module of the ARC3 Campus Climate Survey was adapted from the Department of Defense Sexual Experiences Questionnaire (SEQ-DOD), originally modified from the Sexual Experiences Questionnaire (Fitzgerald et al. 1988; Fitzgerald, Gelfand, and Drasgow 1995). This 16-item questionnaire had strong high internal consistency as part of this implementation of ARC3 Campus Climate Survey (α = .90). The 16 items and 4 subscales are as follows:
- Sexist Hostility/Sexist Gender Harassment (α = .83)
Unwanted and unwelcomed words, actions, symbols, gestures, and behaviors that are based on sex or gender and characteristically repetitive.
- 1.1. Treated you “differently” because of your sex.
- 1.2. Displayed, used, or distributed sexist or suggestive materials.
- 1.3. Made offensive sexist remarks.
- 1.4. Put you down or was condescending to you because of your sex.
- Sexual Hostility/Crude Gender Harassment (α = .83)
Unwanted and unwelcomed words, gestures, and body language of a sexual nature and characteristically repetitive.
- 2.5. Repeatedly told sexual stories or jokes that were offensive to you
- 2.6. Made unwelcomed attempts to draw you into a discussion of sexual matters.
- 2.7. Made offensive remarks about your appearance, body, or sexual activities.
- 2.8. Made gestures or used body language of a sexual nature which embarrassed or offended you.
- Unwanted Sexual Attention (α = .83)
Persistent unwanted, unwelcomed, or violating behaviors and gestures of a sexual nature that caused discomfort.
- 3.9. Made unwanted attempts to establish a romantic sexual relationship with you despite your efforts to discourage it.
- 3.10. Continued to ask you for dates, drinks, dinner, etc., even though you said “No.”
- 3.11. Touched you in a way that made you feel uncomfortable.
- 3.12. Made unwanted attempts to stroke, fondle, or kiss you.
Sexual Coercion (α = .95)
Sexually compelled involuntary actions by an individual without regard for their desire or volition by use of force, threat, or authority.
- 4.13. Made you feel like you were being bribed with a reward to engage in sexual behavior.
- 4.14. Made you feel threatened with some sort of retaliation for not being sexually cooperative.
- 4.15. Treated you badly for refusing to have sex.
- 4.16. Implied better treatment if you were sexually cooperative
Students who attended one of the academic institutions were randomly assigned to one of three survey paths—A, B, and C—to manage the overall level of survey burden on the student population. Path A mainly addressed campus climate and sexual misconduct victimization. Path B included fewer campus-climate questions, but included an economic impact module. Path C focused on a mix of victimization and perpetration questions. Of note for the present analyses, the sexual harassment modules appeared in versions A and B, but not C. All health institution students were given a version of the survey that included both sexual harassment modules.
The UT research team assessed the climate survey data for quality and consistency using a multiple-step approach. First, individual survey responses were inspected and average response times were computed to determine a reasonable minimum threshold for the acceptable time it should take a student to earnestly complete the survey. This in-depth process involved examining the questions missed by students, the relevance of open-ended responses to the topic being assessed, and whether participants had at least attempted all the victimization sections, when applicable. Participants’ right to skip any question per the IRB-approved protocol was considered. Using this process, the UT research team established that 10 minutes was the minimum threshold for an acceptable survey completion. This criterion was therefore set to determine if a response would be retained in the final sample and used for subsequent analyses. In addition, the UT research team evaluated open-ended responses, and excluded responses where there was obvious evidence for survey abuse or participant response error.