Abstract
This cohort study examined the association of Muslim identity and religiosity with mental health issues before and after the 2016 election.
Population health is affected by sociopolitical events, particularly in groups specifically threatened by those events.1 The 2016 presidential election was associated with a rise in anti-Muslim rhetoric, policy, and hate crimes.2,3 Donald Trump called for banning Muslim immigrants.3 We assessed whether the election was associated with changes in the mental health of Muslim college students, an underresearched population potentially facing mental health inequities.4 We further tested whether Muslim individuals who are religious, who are often targets for greater levels of discrimination,5 were most strongly affected.
Methods
Survey data were from a random sample of students 18 years or older from 90 colleges and universities participating in the Healthy Minds Study6 in the 14 months before or after the election. This survey was approved by institutional review boards at all schools, and all participants provided written informed consent.
We assessed a binary outcome defined by exceeding cutoff scores for clinically significant depression, anxiety, or eating disorders (≥15 on the Patient Health Questionnaire–9 or the Generalized Anxiety Disorder–7 or ≥3 on the Sick, Control, One, Fat, Food [SCOFF] questionnaire). Key independent variables were the timing of survey completion (after vs on or before November 8, 2016), Muslim identity (students selecting “Muslim” when providing religious affiliation, with multiple selections possible), and religiosity (students who indicated religion as “important” or “very important” vs “somewhat important,” “neutral,” “not important,” or “very unimportant”).
We adjusted for differences between students who responded and those who did not respond using sample probability weights (inverse of response probability) based on institutional data on sex, race/ethnicity, academic level, and grade point average. We tested for changes in the proportion of Muslim students reporting clinically significant mental health symptoms surrounding the election beyond changes experienced by non-Muslim individuals, using a difference-in-difference logistic regression. We also tested for unique associations for Muslim individuals who were religious. We adjusted for school and self-reported student characteristics known to be associated with mental health (Table). These analyses were conducted between November 2019 and June 2020. Analyses used a 2-sided P < .05 as a threshold for statistical significance and were performed in Stata version 15.1 (StataCorp).
Table. Sample Characteristics Before and After the 2016 Election (N = 75 578)a,b.
| Characteristic | Students, No. (%) | P value | |||
|---|---|---|---|---|---|
| Preelectionc | Postelectiond | ||||
| Muslim | Non-Muslim | Muslim | Non-Muslim | ||
| Total | 684 (1.90) | 34 596 (98.10) | 908 (2.42) | 39 390 (97.58) | NA |
| Positive screening results | |||||
| Depression | 80 (12.64) | 3758 (11.34) | 189 (20.41) | 6442 (16.29) | .46 |
| Anxiety | 53 (7.62) | 3104 (8.82) | 152 (14.83) | 5141 (12.09) | .18 |
| Eating disorder | 72 (9.10) | 3255 (8.67) | 141 (14.39) | 4262 (9.69) | .04 |
| Any | 151 (21.56) | 7230 (20.89) | 308 (33.91) | 10 607 (26.18) | .05 |
| Student characteristics | |||||
| Religious | 451 (63.97) | 11 611 (34.44) | 681 (74.89) | 14 435 (37.13) | .09 |
| Sex/gender | |||||
| Female | 379 (44.78) | 22 829 (55.38) | 569 (49.64) | 27 174 (57.86) | .58 |
| Transgender, genderqueer, or another gender identity | 6 (1.05) | 488 (1.39) | 14 (1.17) | 1013 (2.63) | .36 |
| Sexual minority group member | 54 (7.51) | 4717 (13.69) | 123 (11.52) | 7696 (19.05) | .75 |
| Race/ethnicity | |||||
| White | 100 (11.82) | 23 474 (65.43) | 80 (8.00) | 27 112 (65.87) | .08 |
| Middle Eastern or Arab | 174 (25.62) | 168 (0.39) | 370 (37.39) | 259 (0.51) | .39 |
| Asian | 204 (27.97) | 4485 (11.92) | 227 (25.61) | 3935 (9.89) | .74 |
| African American/Black | 57 (9.65) | 1348 (5.18) | 104 (14.84) | 2024 (7.04) | .65 |
| Hispanic/Latino/Latina | 5 (0.88) | 2376 (8.44) | 0 | 2009 (5.38) | NA |
| Multiraciale | 94 (14.83) | 2367 (7.00) | 75 (8.10) | 3622 (9.17) | <.001 |
| Otherf | 49 (9.23) | 528 (1.65) | 52 (6.06) | 807 (2.14) | .01 |
| Aged 22 y or older | 386 (55.80) | 14 777 (39.18) | 469 (53.08) | 17 392 (42.48) | .15 |
| First-generation college studentg | 163 (24.69) | 8171 (26.01) | 414 (47.17) | 14 541 (37.66) | .01 |
| International studenth | 325 (43.60) | 3472 (9.02) | 308 (32.57) | 2364 (6.10) | .89 |
| School characteristics | |||||
| Small (<5000 students) | 79 (20.22) | 5700 (27.53) | 218 (27.5) | 13 346 (43.69) | .36 |
| Medium (5000-9999 students) | 26 (3.90) | 2930 (8.69) | 190 (19.73) | 5608 (14.66) | .09 |
| Large (≥10 000 students) | 579 (75.88) | 25 966 (63.78) | 500 (52.77) | 20 436 (41.64) | .73 |
| Public | 340 (53.79) | 18 916 (53.48) | 618 (65.48) | 24 243 (53.99) | .22 |
| Carnegie classification | |||||
| Associate’s | 13 (3.48) | 644 (3.13) | 43 (3.53) | 3481 (8.97) | .03 |
| Baccalaureate | 33 (8.14) | 2325 (12.21) | 78 (13.75) | 6852 (27.14) | .41 |
| Master’s | 41 (9.13) | 5102 (21.15) | 271 (33.83) | 10 163 (27.60) | .008 |
| Doctoral | 556 (71.72) | 24 860 (57.46) | 477 (44.96) | 17 623 (32.71) | .77 |
| Special focusi | 41 (7.53) | 1665 (6.05) | 39 (3.93) | 1271 (3.59) | .82 |
| Low graduation rate | 312 (42.25) | 16 421 (44.90) | 613 (73.47) | 25 492 (67.78) | .33 |
Abbreviation: NA, not applicable.
All numbers are unweighted, while percentages are weighted to be representative of the population of students at each school. Significance was tested via an interaction term between the postelection and Muslim variables from a logistic regression model, such that the P values represent a test of the difference in changes from before to after the election, stratified by religion.
A total of 3031 students with missing data on the religious affiliation item (3.86% of the sample) were excluded from the study.
This group included 35 280 students on 29 campuses.
This group included 40 298 students on 61 campuses.
Students who selected 2 or more race/ethnicities were classified as multiracial by the researchers.
Students who selected American Indian or Alaskan Native, Native Hawaiian or Pacific Islander, or self-identify were classified in the other race/ethnicity category because of small group sizes.
Students who indicated that no parent or stepparent had received a bachelor's degree or higher.
Students who indicated they are international students and not US citizens or permanent residents.
Institutions in which a high concentration of degrees is in a single field or set of associated fields (eg, art and design).
Results
The survey response rate was 25%. A total of 75 578 students (56.78% women; 2.24% Muslim) participated. Student and school characteristics are presented for the periods before and after the election for Muslim and non-Muslim participants (Table). Differences between the groups were mostly stable over time. Mental health in Muslim and non-Muslim individuals changed approximately in parallel before the election, with no significant differential change from fall 2015 to spring 2016.
Controlling for changes experienced by non-Muslim participants, the election was associated with a rise of 7.0 (95% CI, 1.0-13.0) percentage points in the proportion of Muslim students experiencing clinically significant mental health symptoms in the 14 months postelection compared with the 14 months prior. Changes from before to after the election were largest for Muslim individuals who were religious, at 10.9 (95% CI, 3.7-18.1) percentage points (vs 8.1 [95% CI, −3.5 to 19.7] percentage points for Muslim individuals who were nonreligious, 3.5 [95% CI, 1.3-5.8] percentage points for non-Muslim individuals who were religious, and 2.8 [95% CI, 1.1-4.6] percentage points for non-Muslim individuals who were nonreligious) (Figure).
Figure. Percentage of Students With Clinically Significant Mental Health Symptoms Before and After the 2016 Election.
A, Results of a logistic regression model with a 2-way interaction between the time frame of survey completion variable and Muslim identity variable. B, Results of a logistic regression model with a 3-way interaction between postelection time frame, Muslim identity, and religiosity variables. Graphs display the percentage of students with clinically significant symptoms before and after the 2016 election of Donald Trump. They depict adjusted probabilities from multivariable logistic regression models (n = 73 664). Models were weighted with nonresponse weights. The 95% CIs were corrected for clustering at the school level using Taylor series linearization. Models were adjusted for student and school characteristics in the Table (including student gender identity, race/ethnicity, first-generation status, student age, international status, and sexual orientation and school size, sector, type and graduation rate). Data collected after November 8, 2016, were classified as postelection (data collected on or before November 8 were classified as preelection). Potential seasonality differences between the postelection and preelection period are not unique to Muslim individuals (per unpublished analyses of Healthy Minds data from the 2018-2019 school year [S. Abelson, MPH; written communication; February 4, 2020]).
Discussion
To our knowledge, this is the first national study of Muslim mental health changes through the 2016 election. Our results indicate the election was associated with declines in mental health among Muslim college students significantly beyond the declines experienced by other students. The largest declines occurred for Muslim individuals who were religious.
Limitations include the survey response rate—although typical for such online surveys and representing the best available data—and a small sample size for Muslim individuals who were nonreligious. We adjusted estimates with nonresponse weights based on known characteristics. Unknown differences between preelection and postelection populations could create bias, but we controlled for student and school characteristics and observed little differential variation in these characteristics from before to after the election. We had limited power to assess variations in outcomes by racial identity within Muslim individuals; this should be a priority for future research.
Conclusions
Our findings highlight links between sociopolitical events and mental health, with potential negative consequences for educational and social outcomes among affected groups. Schools and other communities need to consider these concerns in their efforts to support young adults, and researchers should improve understanding of causal mechanisms and potential prevention and intervention strategies.
References
- 1.Williams DR, Medlock MM. Health effects of dramatic societal events—ramifications of the recent presidential election. N Engl J Med. 2017;376(23):2295-2299. doi: 10.1056/NEJMms1702111 [DOI] [PubMed] [Google Scholar]
- 2.Musu L, Zhang A, Wang K, Zhang J, Oudekerk BA Indicators of school crime and safety: 2018. (NCES 2019-047/NCJ 252571). Published 2019. Accessed August 27, 2020. https://nces.ed.gov/pubs2017/2017064.pdf
- 3.Costello MB. The Trump effect: the impact of the presidential campaign on our nation’s schools. Published 2016. Accessed August 27, 2020. https://www.splcenter.org/sites/default/files/splc_the_trump_effect.pdf
- 4.Samari G. Islamophobia and public health in the United States. Am J Public Health. 2016;106(11):1920-1925. doi: 10.2105/AJPH.2016.303374 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Ikizler AS, Szymanski DM. Discrimination, religious and cultural factors, and Middle Eastern/Arab Americans’ psychological distress. J Clin Psychol. 2018;74(7):1219-1233. doi: 10.1002/jclp.22584 [DOI] [PubMed] [Google Scholar]
- 6.Lipson SK, Kern A, Eisenberg D, Breland-Noble AM. Mental health disparities among college students of color. J Adolesc Health. 2018;63(3):348-356. doi: 10.1016/j.jadohealth.2018.04.014 [DOI] [PubMed] [Google Scholar]