Is Graduate School Worth It? Harassment and Graduate-Student Satisfaction in Political Science

ABSTRACT This article investigates the dynamics of discrimination in political science PhD programs with a survey of current political science graduate students in the top 50 departments. The study focuses on mentorship, funding, sexual harassment, racism, homophobia, and labor exploitation: 20% of respondents reported labor exploitation, 19% experienced racial discrimination, 9% reported sexual harassment, and 6% experienced homophobia. Discrimination is uneven across individuals; some groups of graduate students experience widespread discrimination, especially racial discrimination, whereas other groups are largely unaware of these issues. We conducted a survey experiment to gauge the impact of misconduct on formal reporting mechanisms and find that hearing about racial discrimination has a chilling effect on reporting. We find that experiencing discrimination harms how satisfied students are in their program. We find that factors linked to student vulnerability, such as international status and funding, are significantly associated with harassment and that reporting discrimination predicts more discrimination.

I n response to scandals around harassment and exploitation of graduate students, political scientists have decried the frequency of these issues (Hardt et al. 2019;Mershon and Walsh 2016;Tolleson-Rinehart and Carroll 2006). Studies have highlighted the harm that discrimination causes graduate students (Almasri, Read, and Vanderweert 2022;Gillooly, Hardt, and Smith 2021;Hyder et al. 2022;Mahmoudi 2021). Institutions have responded to scandals and graduatestudent demands with programs to curtail discrimination in political science and to improve support in PhD programs.
We examined the dynamics of discrimination and student satisfaction in PhD programs with a survey of current political science graduate students in the United States. This article highlights the main determinants of graduate-student satisfaction and discusses the results of a survey experiment to gauge the impact of misconduct on formal reporting mechanisms. We found that experiencing discrimination harms student satisfaction with their program and that variables linked to student vulnerability were significantly associated with harassment in our sample. Moreover, reporting discrimination predicts more discrimination.
Our findings have important implications for the discipline. The survey experiment suggests that people who hear about discrimination actually may be less likely to report misconduct than people who do not. This finding implies that discrimination in departments may have a negative impact on willingness to report misconduct.
academic labor system (Bousquet 2008;Kezar, DePaola, and Scott 2019). Schwartz (2014, 507) pinpointed "self-oriented behavior" on the part of tenured faculty, as well as a lack of inter-rank solidarity between the tenured and nontenured, as the main reason for overall worsening labor conditions. Bousquet (2008) argued that university administrators facilitated the casualization of labor through their policies toward graduate students. As such, graduate students have become "byproducts" of the academic system's labor extraction (Bousquet 2008, 21).
Research identifies sexual harassment and sexism as pervasive problems in the academy (Hardt et al. 2019;Monroe et al. 2014;Shames and Wise 2017;Tolleson-Rinehart and Carroll 2006). Women are underrepresented and underpaid in academia (Monroe et al. 2014). Political science has recognized issues of "gender disproportionality" as well as "inhospitable institutional climates" (Shames and Wise 2017, 811, 813). Moreover, in a recent study of Title IX complaints, Cipriano et al. (2022, 343) found that "nearly all participants experienced severe, education-limiting consequences" of harassment.
People of color within political science face bullying and harassment, which interacts with sexism. Sediqe and Nelson (2022) noted that women of color are disproportionately underrepresented in higher education and burdened with service to resolve issues of representation and equity. The discipline has been decidedly unwelcoming to people of color of all genders, especially those who work with marginalized groups. Black, Indigenous, and people of color (BIPOC) academics have written of the "presumed incompetence" that impacts how their research is viewed (Gutiérrez y Muhs et al. 2012). This bias impedes BIPOC scholars' path toward common benchmarks of success (McClain et al. 2016), and these challenges are compounded for graduate students.
Little research exists on homophobia and transphobia in the academy. McNaron (1997) documented 304 LGBTQþ faculty members' work with students, colleagues, and administrators, noting both hostility and enthusiasm. As a Black lesbian in academia, Glover (2017) explored isolation and harassment; our respondents echoed many of her experiences. To our knowledge, our article is one of the first to measure and address homophobia and transphobia in political science.

METHODOLOGY AND DATA
We conducted an online survey in February 2022 with current graduate students across the top 50 political science programs, according to US News & World Report. The survey was pretested by two graduate-student focus groups. The full questionnaire is in online appendix 1. We compiled an email list of all current graduate students in the top 50 programs (i.e., 3,349 students) and sent them the Qualtrics survey. We received 271 responses and 205 complete responses, encompassing 49 of the top 50 programs, 1 public and private schools in all regions of the country, multiple genders, eight ethnic groups, students aged 22 to 48, LGBTQþ students, international students, firstgeneration students, students with children, and all subfields (El Kurd and Hummel 2023).
The number of respondents and the sampling frame are comparable to other recent studies of political science graduate students, although our 8% response rate is lower. Almasri, Read, and Vandeweerdt (2022) surveyed students in the top 10 programs and had a 47% response rate with 308 responses. Gillooly, Hardt, and Smith (2021) surveyed students in the top 50 programs and also recorded 308 responses.
Our survey's demographics largely resemble those reported by previous studies. Table 1 also compares our sample to data from the American Political Science Association (APSA), which surveys members who have completed their PhD-thus, we expected the APSA sample to vary systematically from our sample. 2 Inferences from our sample are limited by the low response rate. We do not know if our sample is truly representative of current political science graduate students because a census does not exist. However, when we compared recent samples of political science graduate students in table 1, our sample is similar on some dimensions, and it includes more cisgender women, more people of color, and younger political scientists than in other samples. We also asked additional demographic questions that other studies have not: for example, in our sample, 5% of respondents are trans or nonbinary, 31% identify as LGBTQþ, and 45% belong to a union.
We suggest that our sample includes useful variation across graduate students. We do not believe that the sample's descriptive statistics hold for all graduate students, which is why we report them without confidence intervals. We posit that the survey findings represent important trends in political science for several reasons. Our findings on mentorship, harassment, and vulnerability reflect existing findings in political science (Hesli, Fink, Duffy 2003;Hofstra et al. 2022). Additionally, our survey demographics are similar to other samples of political science graduate students. The survey captures demographic, experiential, and preference variation, even if we were unable to compare to true population parameters. If we had oversampled vulnerable populations or harassment experiences, our descriptive statistics would differ from population parameters; nevertheless, the trends that we discuss would exist in the broader political science community. Finally, any type of discrimination is an issue that the field should take seriously.

RESULTS
The survey asked 63 questions about funding, working conditions, sexual harassment, homophobia, racism, mentorship, and satisfaction. Funding across the sample was adequate: 73% of students stated that their department "always" or "usually" provides enough funding for them to pay their bills in an average month-90% have at least five years of guaranteed funding and 58% have guaranteed summer funding. Those students who do not receive adequate funding struggle financially: 31% reported that they worked outside of their department to make ends meet and only 29% stated that funding decisions were made fairly or transparently.
When asked about labor practices, 20% of students reported experiencing labor exploitation. In open-ended responses, they repeatedly described working uncompensated hours as teaching assistants (TAs). Some students were promised coauthorship by professors or other students only to see their work published without attribution. Table 2 presents descriptive statistics on the sampled students who reported labor exploitation, sexual or racial harassment, or homophobia overall and by demographic categories. Table 2 demonstrates that harassment experiences are common across the sample, with one in three students reporting at least one form of harassment. Some demographic groups experienced more harassment than others. Cis white men reported less harassment whereas trans people reported concerning levels of harassment of all types. Women of color reported high levels of labor exploitation and racism; Asian women in particular reported high levels of labor exploitation, sexual harassment, and racism.
We asked direct questions about sexual harassment and misconduct. A concerning 9% of our respondents reported having experienced sexual harassment or misconduct by faculty members or other students. In open-ended answers, these experiences included inappropriate comments, unwanted advances, and actions bordering on assault.
We conducted a priming experiment to probe reporting dynamics. Almost 80% of students stated that they would "likely" or "very likely" report misconduct by a faculty member. This experiment was designed to explore whether a climate of misconduct has a chilling effect on future reporting. Research demonstrates that harassment can have a psychological toll on victims and also has bystander effects within the institution (Bowes-Sperry and O'Leary-Kelly 2005; Buchanan and Fitzgerald 2008;Fox and Stallworth 2009;Johnson, Kirk, and Keplinger 2016;Nickerson et al. 2014;Schneider, Swan, and Fitzgerald 1997). Both victims and bystanders might develop a fear of retaliation.
In the experiment, we randomly assigned respondents to three articles, all published in The New York Times for consistency. The first article primed respondents on sexual harassment by using the case of sexual misconduct by Jorge Dominguez; the second article primed respondents on racial discrimination by using the case of Lorgia Garcia Pena's tenure denial; and the third article acted as a neutral prime by providing information about the shifting ages of graduate students (see online appendix 1 for these primes). Following exposure to the prime, we asked respondents how likely they were to report discrimination or harassment in their department. We had no prior research to suggest one form of harassment would have a more substantial impact than another.
Our results showed that those who received the racialdiscrimination prime were 7% less likely to report misconduct than those who received the control (p<0.001; see the full analysis in online appendix 3). We interpreted these results to suggest that hearing about discrimination could have a slight chilling effect on reporting because reporting might provoke retaliation or the systemic nature of discrimination makes reporting seem futile. We do not have data to distinguish between these mechanisms. We expect that this effect is amplified in the context of a student's own department and environment.
We asked whether respondents believed that their department would address sexual harassment if it were reported. Only 28% of respondents answered yes. These results were gendered: 40% of cis men stated that their department would act and 21% of cis women and trans people stated the same. Results were divided by ethnicity: white men trusted their department more than other groups (i.e., 54% believed that their department would act if they filed a complaint). This corroborates previous research at the faculty level of gendered differences in perceptions of sexism (Hill and Hurley 2022). We also asked about experiences of homophobia: 21% of students reported hearing of incidents of homophobia but only approximately 6% of the sample and 13% of the students who identify as LGBTQþ reported experiencing it themselves.
We asked comparable questions regarding racial discrimination. A staggering 59% of respondents reported having witnessed such discrimination; 19% had experienced it themselves, including 45% of students of color. In their answers to open-ended questions, respondents outlined discrimination ranging from micro-aggressions to exclusionary practices that harmed their career. In response to the question, "If you file a racial harassment complaint, will your department take action," only 29% T a bl e 2 answered yes-and they primarily (70%) were white. In the openended responses, students stated that they would not report discrimination because they did not trust faculty or they feared retaliation.

Harassment experiences by demographic group
Mentorship is a crucial determinant of academic inequality (Hesli, Fink, and Duffy 2003;Hofstra et al. 2022). Surveyed students ranked how supportive their advisor was on a scale of 1 to 10, with 10 being the most supportive. The average score was 7.98, with a standard deviation of 2.32. That variation is not highly associated with gender identity or ethnicity. In general, most students were satisfied with their advisor and mentorship.
Finally, we asked about students' overall satisfaction with their program on a scale of "burn it down" (1) to "everyone should go here" (10). The average was 6.4 on this question, with a standard deviation of 2.32. Harassment and exploitation appear to be correlated with program satisfaction. These trends can be further probed with correlation coefficients (table 3) and regression analyses (table 4). The correlation matrix in table 3 demonstrates that students who reported any form of exploitation or harassment reported less satisfaction in their program; of concern is that students who reported one form of harassment or exploitation were more likely to report another.
When labor exploitation and sexual or racial discrimination were the dependent variables in a logistic regression, we found that other forms of discrimination and variables representing a student's vulnerability were associated with a given form of discrimination (see the tables in online appendix 3). The variables highly associated with labor exploitation were a student's funding package and racial discrimination. For racial discrimination, the significant variables were racial demographic categories and experiencing labor exploitation. Sexual harassment was significantly associated with gender identity, labor exploitation, and being an international student (statistical analyses are in online appendix 3). We interpreted these results to suggest that discrimination and student vulnerability predict other forms of discrimination.
In our main regression analysis in table 4, the satisfaction score was the dependent variable. Independent variables included an overall funding, research funding, satisfaction with advisor, gender, race, status in the program (i.e., student versus candidate), labor exploitation, sexual harassment, and racial discrimination. For alternative model specifications that included more demographic variables, see online appendix 3. The model specification with harassment variables (Model 2) accounts for much of the variation in satisfaction, with an R-squared of 0.52.

DISCUSSION AND CONCLUSION
Our results suggest that some political science students experience productive and safe learning environments but that these positive training experiences do not extend to all students. Women, trans people, and students of color are more likely to report that they have been targets of harassment and exploitation. Furthermore, variables that could indicate student vulnerability were significantly associated with harassment. Our survey experiment suggests that people who hear about discrimination may be less likely to report misconduct than people who do not, suggesting that a department culture in which discrimination is rampant could have a negative impact on the willingness of graduate students to report. Conversely, respondents who had not heard of discrimination in their department were more likely to state that they would report misconduct and that their department would take reports seriously. The unevenness of discrimination and trust in institutions suggests that people who do not experience discrimination may dismiss such reports because they perceive discrimination to be rare and reporting to be easy. Finally, we found that advising, funding, and harassment largely explain why some political scientists in our sample reported better or worse experiences in graduate school. Based on answers to open-ended questions in the survey, we suggest that all political science PhD departments adopt the following recommendations to curtail discrimination. To address labor exploitation, departments should communicate and enforce clear guidelines for TA and research assistant (RA) positions and enforce limits on work hours. Departments should advocate for adequate health insurance and mental health coverage for students and enforce maternity-leave policies. Departments can ease some financial burdens on graduate students by paying upfront for costs such as conference registration and airfare, with no reimbursement delays. Finally, departments should encourage unionization efforts because unions can address labor disputes and cost-of-living issues better than many departments.
To address sexual and racial harassment, departments should conduct regular climate studies and make those data available to students. Departments should communicate where to report misconduct and establish multiple points of contact for student support. University-mandated trainings for faculty members are not enough; departments should communicate their standards for sexual harassment and incorporate discussions of appearance politics. Departments also should create student leadership positions and allow students to have representation on appropriate committees. Departments also could focus on transparently communicating funding decisions and establish clear guidelines for comprehensive exams.
Graduate students should consider that in the survey's openended responses, those with union representation stated that they were more comfortable reporting misconduct and felt more protected. Additionally, many respondents noted in their open-ended answers that they did or would not report misconduct because they believed that their department would not act. Unionization could address some of these concerns.
Researchers note that many policies exist to rectify discrimination; that they are not widely adopted suggests "not a lack of knowledge so much as apathy, prejudice, gender stereotypes, and cultural cues that end by depriving society of some of its best talent and energy" (Monroe et al. 2014, 418). Training programs can create and maintain healthy environments with the application of many of these recommendations.

ACKNOWLEDGMENTS
We thank Hamad Al-Humaidan for exceptional research assistance. Carew Boulding, Jami Nelson-Nuñez, Ximena Velasco Guachalla, our focus-group participants, the editors, and three anonymous reviewers provided extremely helpful feedback and suggestions. We appreciate the support of APSA, which awarded us a Centennial Grant that funded research assistance on this project.

DATA AVAILABILITY STATEMENT
Research documentation and data that support the findings of this study are openly available at the PS: Political Science & Politics Harvard Dataverse at https://dataverse.harvard.edu/dataset. xhtml?persistentId=doi:10.7910/DVN/KN28MM. The unevenness of discrimination and trust in institutions suggests that people who do not experience discrimination may dismiss such reports because they perceive discrimination to be rare and reporting to be easy.
…students who reported any form of exploitation or harassment reported less satisfaction in their program; of concern is that students who reported one form of harassment or exploitation were more likely to report another.