The Sexual Health Initiative to Foster Transformation (SHIFT) was funded by Columbia University and formally launched in early 2015, under the joint leadership of Jennifer Hirsch and Claude Ann Mellins. SHIFT’s broad goals were “to advance the science of sexual assault prevention and to contribute to building a healthier and safer undergraduate community.”1 Columbia administrators and staff provided institutional input through the research process. However, the SHIFT faculty investigators retained complete scientific independence in terms of study oversight, data access, and decisions about what and when we published. Unlike many faculty who conduct research on sexual assault at their own institution, the SHIFT team benefited from enormous institutional support, including help securing an exemption from mandated reporting. Federal rules require designated university employees to report all instances of gender-based misconduct, defined as sexual harassment, sexual assault, sexual exploitation, stalking, domestic violence, and dating violence. At Columbia, all faculty and staff are designated mandated reporters; without an exemption, the SHIFT team would have been required to make a formal report to the university for each instance of sexual assault we learned about. But as most students chose not to report experiences of assault, this would have rendered SHIFT unable talk to most students about their experiences. Our SHIFT-related mandated reporting exemption enabled us to guarantee confidentiality to students with whom we interacted in our capacity as SHIFT researchers (although not in roles of teaching or administration).
In addition to the ethnography, SHIFT’s overall research design involved two forms of quantitative research: a 60-day daily diary study with 427 Columbia undergraduates and a population-based survey, sent out to a representative sample of 2,500 Columbia and Barnard undergraduates.2 The project drew on the tenets of community-based participatory research, including stakeholder engagement both to improve the quality of our research and to help us develop effective recommendations for innovative, institutionally appropriate, evidence-based strategies to reduce sexual violence and promote sexual health at Columbia.3 Concretely, this meant working with an Institutional Advisory Board, consisting of a group of administrators from Columbia and Barnard responsible for undergraduate safety and well-being; an Undergraduate Advisory Board (UAB); and a Faculty Advisory Board. The UAB was comprised of approximately eighteen undergraduates from Columbia and Barnard. They represented a range of different student interest groups, from the head of the leading student organization protesting how the University handled sexual assault cases, to captains of athletic teams, members of student government, and those who, by their own account, were relatively disengaged from campus life. They also represented a range of sexual and racial identities, as well as socioeconomic backgrounds and national origins. We met with this diverse group of undergraduates once a week, on Monday mornings, from 8 to 10 a.m., over the course of two years. Members were paid for the time so they didn’t have to choose between participating in our group and working a campus job. The UAB also helped us develop strategies to recruit students for the different components of the research, and provided insight into life at Columbia and Barnard. They contributed in important ways to our campus presence as a student-friendly organization that could be trusted. The UAB helped create a campus-wide buzz about the project with activities such as posting flyers, sponsoring study breaks, advertising on Facebook, and sending creative emails. As our UAB was composed of current students, we set clear guidelines about their work with us. Each member signed collectively developed and mutually agreed-upon rules that outlined the confidentiality of information shared in our meetings as well as respect for group members’ opinions. UAB members were excluded as research subjects and were never presented with any data that would allow them to identify an interviewee or someone we encountered during our fieldwork.
Jennifer and Shamus first met in November 2014 to discuss working together on SHIFT. Nine months of pre-fieldwork preparation were spent constructing interview protocols, participant observation guides, and the wide range of research procedures required to do this study. We also hired our staff during this time. Interviews began in late summer 2015; ethnographic observations began shortly thereafter. Research continued through January 2017, spanning one and a half academic years. The team completed 151 in-depth student interviews, 25 student follow-up interviews, 18 key informant interviews, 17 focus groups (of about 10 students each), and about 600 hours of community and participant observations. All of the research procedures, including steps taken to protect the data itself, were reviewed and approved by Columbia’s Institutional Review Board (IRB). The three tables in Appendix B contain information on the characteristics of our interview sample and our focus groups, and the spaces and length of time where we made observations. We began our analysis while still collecting data. We drafted our first papers in 2016. We began writing this book in January 2018 and completed it in the summer of 2019.
For the ethnography we hired a diverse staff with master’s- and doctoral-level training in both interviewing and participant observation. We provided this staff of five people with additional training in study goals and methods, procedures for the protection of human subjects, and our shared emergency protocol in case a research participant became distressed. In addition to the research expertise of Hirsch and Khan, the ethnographic team collectively had done research on queer communities of color (anthropology), masculinity, power, and relationship violence (anthropology), and trauma and sexual abuse (social work), and had worked on a series of public health research projects on sexual activity within communities of color. Like everyone involved with the SHIFT project, team members adhered to SHIFT’s publication policy, which specified the process for proposing and doing the analysis for scientific publications drawing on the SHIFT data. As we prepare this book for publication, seven papers have been published or are scheduled for publication with a SHIFT ethnographic team member as first author, and several others are in development or under review. From the beginning of the research, Jennifer and Shamus made clear to the SHIFT research team that they intended to write a book drawing on the ethnography.
Students were recruited for interviews in six ways: (1) by being contacted after they wrote directly to SHIFT requesting to be interviewed, in response to an email sent to all students about the SHIFT survey; (2) through our Undergraduate Advisory Board (UAB), which connected us to student interest groups we wanted to reach, and where we made in-person presentations about our project; (3) using a snowball sample where some interviewees recommended to their friends that they contact the research team; (4) by targeted recruitment of survivors of sexual assault through email that asked students to contact us if they had “a story to tell”; (5) through an on-campus presence in the form of articles in the student paper, flyers, signs, and electronic bulletin boards; and (6) through our participant observation, during which we asked people if they were interested in being interviewed. About half of the students who participated in in-depth interviews were recruited by research staff in this sixth way: through participant observation.
A word of explanation is necessary about that first group of students. The email sent out in the winter of 2016, announcing the upcoming survey, went directly to the entire undergraduate student body at Columbia from SHIFT’s (no longer operative) general account, shift@columbia.edu, and was sent to Barnard undergraduates as well. We were somewhat taken aback to receive responses from many individual students, saying that they wanted to participate in a confidential interview because they had a story to tell. We hired another interviewer, Megan Kordenbruck, who had a background in trauma-informed interviewing, to help us respond to these interviews; Megan followed up with each one, and then they were interviewed by a team member of their choice. We were struck by how eager many students were to tell their stories as a way of shaping campus responses to sexual assault. Because of how we picked students to interview, there are likely biases within our sample. We did not interview any students who described intentionally acting in a way that they understood in that same moment as sexual assault; we suspect that those who deliberately act to harm their peers were unlikely to choose to sit down with us for an interview. Students who were assaulted and who were unhappy with their experiences reporting to the university were probably more likely to seek us out than those who were happy with their experiences.
The team tracked various axes of diversity to ensure that the many voices and perspectives of the student population were represented. Research participants, who were screened to ensure that they were currently enrolled undergraduates at either Columbia or Barnard, also provided demographic information. Students received $35 for participating in the initial in-depth interview; for those for whom subsequent interviews were required to capture the breadth of their experiences, $40 was given for each follow-up interview. Interviews averaged approximately two hours. The SHIFT website featured a photo and brief bio of every member of the ethnographic team, and participants could select who they wanted to interview them. Our team of seven people represented a range of gender identities, sexual orientations, religious backgrounds, racial identities, and personal experiences. Students tended to pick the person who first contacted them as their interviewer. One student expressed regrets about what they had said in their interview; we deleted this interview and did not use it in our analysis.
Our in-depth interviews were all done in a private office on campus. During these interviews we gathered information on the participant’s life before college, including their relationship with their family; their experiences with substance use; their life as an undergraduate; current and past sexual and intimate relationships, with an emphasis on experiences during college; how consent worked within their sexual interactions; nonconsensual sexual experiences; and how the organizational rules and structures at Columbia or Barnard shaped their lives. Our interview started by asking students about how they ended up at Columbia, and continued by asking about their experiences with their family, with substance use, and sex were before coming to college. We then asked about college experiences, before asking subjects to describe a typical sexual experience, a good sexual experience, and a bad one. We never used the word “consent” in these first conversations; instead we simply asked students to describe what happened. Often they would say things like, “Well, you know . . .” so as to avoid more explicit descriptions of how they ended up naked with another person. We would reply with some version of “I’m sorry, but I don’t know. I’d like you to tell me, to the degree that you’re comfortable.” After we heard these descriptions of what happened, we went back over the sequence of events to talk about consent. Students are well versed in the necessity for consent, and had it been in the forefront of their mind when they first described their sexual experiences, they would have likely framed all their encounters relative to consent. After they described the three kinds of encounters we asked them about, we asked them about assault. We didn’t ask questions about stalking, revenge porn, sexual harassment, or nonsexual forms of intimate partner violence. We intentionally chose to dig deep on a narrower set of topics focusing on assault rather than to provide a shallower understanding of a wider range of experiences. Several of the students we interviewed had not had many sexual experiences; some had had none at all. These interviews were briefer, but we nonetheless asked questions about their sexual projects. About a quarter of the way through our research we revised the interview questions in light of what was and was not working within the interviews. The interviews were professionally transcribed by a company that had a written agreement with Columbia University to adhere to standards for data protection. For quotes drawn from the interviews, we have removed “ums” and “likes,” but otherwise have not altered direct quotations in any way.
The focus groups, all moderated by Shamus, touched on topics similar to those of the in-depth interviews, but emphasized how groups of students think about those issues, and general patterns of behavior and shared ideas. Focus groups lasted approximately two hours, and participants received $30. Groups averaged about ten participants each. Students either volunteered to participate by responding to posters placed around campus or to a campus-wide email, or they were directly recruited because they represented or participated in a group of interest. There was little overlap between focus group participants and interview subjects. In addition to groups that included any member of the student body, we constituted groups that were solely comprised of students of a single sex, international students, first-year students, athletes, LGBTQ students, religiously engaged students, racial minority students, and students who were the first in their family to attend college. Our aim was to capture both the general conversation on campus, and those of specific groups or kinds of campus experiences that were of interest to us. Table 2 in Appendix B provides a more detailed account of the focus groups’ composition.
Jennifer and the research staff conducted key informant interviews, composed to form two groups. First were administrators who oversaw specific policies and procedures like housing, alcohol, or orientation that were relevant to our study. The second group consisted of a small number of students who held student leadership positions or served as resident advisors. The goals of those interviews were to understand what administrative actors know or perceive regarding social, academic, and interpersonal student experiences; how administrative actors perceive and describe sexual assault; and what the institutional terrain is, including relevant policies related to alcohol and substance use, and to sexual assault. Students received $35 for the key informant interviews; administrators were not compensated.
For just over one and a half academic years, from September 2015 through January 2017, the team engaged in participant and community observation at various locations both on and off campus. On-campus spaces included Columbia and Barnard dorms, academic buildings, campus community space, athletics spaces, fraternity-sorority life spaces, and dining halls. Off-campus spaces, which were visited by invitation from a student, included bars, restaurants, coffee shops, and public and community spaces around New York City.
Jennifer and Shamus’s participant observations were limited to large public events; the experiences and activities that were in some way the most essential to understanding students’ lived experience were off-limits to us. Instead, we had to set up structures to transfer both the field notes and the “headnotes” from our research staff to us. We required weekly submission of detailed field notes for all observations and brief write-ups of each week’s in-depth interviews. These field notes were submitted each Sunday. Jennifer and Shamus reviewed the notes and met for two hours on Mondays or Tuesdays. On Thursdays we held two-hour meetings with the entire ethnographic team, who also read each other’s notes and interview summaries. These weekly meetings provided an opportunity to ask questions, discuss emerging themes, push the staff to follow specific lines of inquiry, and keep tabs on which groups were being included in the research and which were still not adequately covered. It also allowed all members of the research team to discuss their various impressions of the previous week and the project more generally.
From fall 2015 through spring 2017 the authors of this book met for a minimum of six hours a week, and often longer: two with each other, to discuss the field notes and interview summaries; two with the UAB to discuss the design and findings; and two with the ethnographic team as a whole. We also met with the quantitative team regularly. The quantitative team was led by Claude A. Mellins, an expert in adolescent and young adult development, mental health, substance use, and trauma, and included Louisa Gilbert, a professor of social work with expertise in gender-based violence, substance use, and HIV; John Santelli, a pediatrician and expert on adolescents, sexual health, and sexual education (and Jennifer’s husband); Melanie Wall, a biostatistician with expertise in public health, epidemiology, and quantitative analysis; Kate Walsh, a psychologist whose research has focused on trauma exposure, particularly sexual assault; and Patrick Wilson, a psychologist with experience studying health disparities, HIV, and the health outcomes of more marginalized populations. We held occasional all-team meetings to share findings with the quantitative and ethnographic teams.
Throughout the book, all of the fieldwork activity is presented in the first-person plural; that is, we talk about students talking “to us” or write about things that “we” saw. In participant observation in public places, interviews with students that Jennifer or Shamus personally conducted, all of the key informant interviews with university administrators, and all of the focus groups, this “we” refers to Jennifer or Shamus, who collected the data directly. In all other instances, “we” refers to the ethnographic research team staff members, with the write-ups based on interview transcripts, interview summaries, field notes, and other documents produced in the course of data condensation and analysis.
An essential element of the participant observation was the clearly articulated protocol for identifying ourselves as researchers when interacting with students. Team members always identified themselves as researchers, told students what SHIFT was, advised them that interacting with the researcher was optional, and offered contact information for the lead researchers should the student have questions or concerns. SHIFT researchers were never “under cover.” We publicized our project with flyers, emails, and several pieces in the student newspaper, the Columbia Spectator. Team members always secured permission from hosts to enter any private student space such as a dorm room or a suite or basement party, and provided information about who students could contact if they had further questions. Jennifer and Shamus did not do any participant observation with students in student-controlled private social spaces such as dorms, student apartments off campus, and fraternity houses. We are both faculty at Columbia University; it seemed unlikely that we would be able to gather reliable data, and we were conscious of student concerns about having their professors hanging out at parties with them. All of the participant observation with students in student-controlled spaces was conducted by the research staff we hired, who were closer in age to the students and thereby less likely to stand out in such spaces.
Given our topic, we developed a process to identify students we encountered who were in crisis and refer them to services in the hope of getting them the help they needed. This “emergency protocol,” drawing on Claude Mellins’s research with vulnerable adolescents and young adults, provided our staff with guidelines to assess and appropriately refer students in crisis. As a further safeguard, one of SHIFT’s two clinicians—Mellins, who is a clinical psychologist, or John Santelli, who is an adolescent medicine physician—was always on call in case a team member had questions. Over the course of our research, we retrained our team multiple times in our crisis assessment and procedures. Mellins also supported the entire team in managing our own emotional responses to the difficult stories that emerged during the course of our research. At one point or another every member of our team experienced some degree of “vicarious traumatization”—taking on some of the emotional burdens experienced by our interview subjects. Mellins helped all of us, including the authors, manage these experiences.
During our research, our team sometimes observed illegal activities, from cocaine use (relatively rare) to underage drinking (fairly common). We developed a protocol to articulate our obligations for intervention: simply put, if someone seemed in acute danger of harming either themselves or others, we had an obligation to intervene. Otherwise, we simply observed. We made clear to our research staff that even if it meant threatening our capacity to do research, we had an obligation to intervene if, say, we saw a man carrying a woman who was all-but-passed-out from a party to some other location. This never happened. But it was necessary, for example, to help get a young woman an ambulance to the emergency room because she was so exceptionally drunk at a downtown club (away from campus). She was also underage—but our concern was her severe incapacitation, not her age. Our researchers never broke the law, nor facilitated any breaking of the law. Our role was not to police behavior, but to observe it. Through our observations, we hoped, we would generate insights that could make experiences of sexual assault less likely. Had we intervened at every illegal act, our observations would have been impossible. We were always transparent with our subjects about what we were doing, and raised with the IRB any ethical concerns that arose during the course of our research; we were always in compliance with what we saw as our ethical obligations, and with the requirements of the IRB.
We started every ethnographic team meeting with a mental health check-in. This book contains a lot of painful stories; hearing them directly was emotionally difficult, and talking to one another helped with this burden. Acknowledging and providing each other with support was essential to the team’s capacity to do our work. The mental health check-in was also the time to share feelings that research staff suppressed as we strove to maintain a relatively neutral listening stance when students shared stories that generated strong feelings. Taking into account how each member of our staff deliberately crafted a social self, we asked the team members involved in participant observation to write up lengthy notes about who these social selves had been and how they had experienced inhabiting them. These created a written record of how each researcher made deliberate choices regarding self-presentation, and gave us a lens through which to read and consider the interviews that each member of the ethnography team produced. Such reflections were essential to our data analysis.
The data from this research consisted of interview transcripts, focus group transcripts, research notes about each interview and focus group (describing the mood and other details that could not be captured in a transcript), field notes, and descriptions of each researcher’s “social self.” The interview transcripts totaled nearly 20,000 pages; the other materials added thousands more. We analyzed our data in two ways. Two team members coded all information for eleven major themes: socializing, partner selection, relationships, sexual projects, stories of sexual assault, consent, telling someone about sexual assault, mental health experiences, alcohol and substance use, sexual experience (not assault), and other notes. We also constructed many types of secondary documents: for example, extracting all the stories of assault from interviews and analyzing what happened before, during, and after the event. These secondary documents were essential, given the mass of data we had; they also helped us see themes we hadn’t coded for. While many qualitative researchers today use coding and analysis software, we do not. Instead, we pored over the transcripts and field notes, writing about them, having conversations about them, and eventually, coming to some kind of understanding. There were major themes we paid attention to because the broad literature guided us to. But there were also themes that emerged over the almost five years of working together, talking about it with one another and our team, and writing up our results. We did not include quotes from every single student we interviewed or interacted with in the book. But the experiences of every single participant shaped the analysis presented here; we have spent hours and hours poring over their transcripts, reading the field notes, reviewing the write-ups produced immediately after each interview.
To protect the identity of participants, we took a number of steps beyond the obvious one of not using their real names, including changing critical identifying details such as hometown, family characteristics, physical descriptions, and extracurricular activities. We have gone to particularly great lengths to mask group identities as well. We tried, in selecting alternate descriptions of students’ group participation, to pick groups that paralleled the social prestige and meaning of the original group. Anything named as a team is not that team and may not have been an athletic organization at all. Yet we would not transform a member of the football team into a member of the chess club. Football players may be members of the chess club, but the reader’s interpretation of such a replacement would likely generate an inaccurate impression, and thus when we made changes, we sought to preserve key elements of student organizations’ social meaning. Some organizations we made up; we know that there is not an epicurean club.
At the same time, some information we do not change. All the stories of assault remain strictly true to the experiences of students as they described them. Demographic and personal information might be purposefully inaccurate, but the stories hew extremely close to how the events were recounted to us. And if information is analytically important (for example, gender or race), we haven’t changed it. The experience of Black men having unique experiences navigating consent are all from Black men. We have not created any “synthetic characters”—combining stories from different people into a single person. However, we have, in a small number of instances, made a single person into different people for purposes of telling different stories unrelated to one another and, in our evaluation, unimportant to one another. Doing so reduced the likelihood that research subjects could be identified by their friends and peers. We believe that most people will be unable to identify themselves, though given the specifics of some of the stories, and the fact that they know they spoke with us, some subjects may suspect that a person in the book is drawn from their story.
In some cases, students who participated in the research struggled to grasp the separation between our roles as professor or class teacher and as researcher. We stressed that all research would remain confidential and that data would be deidentified, coded with research IDs and pseudonyms, not actual names. However, students often missed this distinction. For example, a student in Shamus’s class told him that her assignment would be late because of reasons discussed in her recent in-depth interview, which had been conducted by another team member. Shamus had to reiterate that he did not know which of his students participated in the research, and that her name was not associated with her stories. Similarly, we repeatedly had to clarify both with research participants and with the UAB that our team members were not “spies”; we always identified ourselves as researchers when doing observations in both public and private spaces.
Alexander Wamboldt, one of our ethnographic team members, reviewed this manuscript with two tasks: (1) to weigh in on the changes we had made to protect students from identification by their peers, and (2) to verify our representation of every single story, making sure our accounts accurately represented the material in our transcripts and field notes. This second task allowed us to authenticate every account provided in this book. We are deeply indebted to Alex for undertaking this task. Any errors are, of course, our own.