Exposure to online harassment and disordered eating in transgender and gender-diverse young adults

Abstract

Objective:

To estimate the association between online harassment and same-day disordered eating among a sample of US-based transgender and/or gender diverse young adults.

Method:

Participants (n = 150) were United States residents aged 18–25 years, English-speaking, and identified as transgender, nonbinary, and/or another gender different than that assigned at birth. We collected data between September 2022 and August 2023 using a 10-day daily diary design. Multilevel models were employed to estimate the association between any/no online harassment in the past-24 hours and odds of same-day disordered eating behavior (overeating, binge eating, weight-controlling behavior, coping-related eating, and body dissatisfaction).

Results:

The sample included 1,192 diaries (mean per participant = 7.95, range = 1, 9). Online harassment was associated with greater odds of same-day overeating (adjusted odds ratio, aOR = 3.38, 95% confidence interval, CI = 1.02, 9.17), binge eating (aOR = 15.09, 95% CI = 2.73, 83.43), and coping-related eating behavior (aOR = 4.07, 95% CI = 1.51, 10.93).

Discussion:

Experiencing online harassment was associated with higher odds of same-day disordered eating behavior in a sample of transgender and/or gender diverse young adults. Clinical and public health interventions must focus on improving safety in online environments and mitigating potential harms arising from negative digital interactions.

Introduction

An estimated 398,900 young adults ages 18–24 years in the United States (US) identify as transgender and/or gender diverse (TGD).¹ TGD people include those with a gender identity different than societal expectations based on the sex they were assigned at birth. The umbrella term TGD includes trans, genderqueer, nonbinary, and gender-nonconforming people. Importantly, TGD people are at higher risk than their cisgender counterparts for several mental health concerns, including depression, anxiety, suicidality,² and eating disorders.³ TGD young adults specifically are at elevated risk of disordered eating and body dissatisfaction.^4,5 This elevated risk of disordered eating and body dissatisfaction is of particular concern because these symptoms predict diagnosable eating disorder development,⁶ a source of increased mortality and reduced quality of life.⁷ Furthermore, disordered eating and body dissatisfaction symptoms themselves represent substantial psychiatric morbidity of critical public health concern and economic cost.^8,9 While access to gender-affirming care can reduce the risk of eating disorders and body dissatisfaction,¹⁰ and thus is necessary, evidence suggests that access to gender-affirming care alone is not sufficient to fully mitigate this heightened risk of disordered eating and body dissatisfaction.¹¹ Moreover, understanding avenues to reduce the risk of disordered eating is of heightened importance in the current US political environment, where anti-TGD policy attacks, including those on gender-affirming care, have risen dramatically.¹²

Several factors contribute to disparities in disordered eating between TGD and cisgender young adults, including their heightened risk of experiencing interpersonal discrimination. As outlined by minority stress theory, these mental health disparities are often driven by interpersonal discrimination.^13,14 Empirical studies have broadly supported the minority stress model in this context, documenting relationships between harassment, discrimination, abuse, and disordered eating behavior.^15,16 One manifestation of discrimination targeting this population is online harassment, defined as physical threats, stalking, harassment, purposeful embarrassment, and/or sexual harassment in digital environments.¹⁷ However, it is unknown whether there is an association between online harassment and disordered eating in TGD youth.

This gap is of critical importance because TGD people experience more online harassment and abuse than their cisgender peers,¹⁸ with recent evidence demonstrating escalating risk following the last election cycles.^19,20 Furthermore, online harassment and abuse can lead to internalized stigma¹⁹ and externalizing problems²¹ in TGD people, highlighting the potential mental health effects of online interactions in this population. Capturing day-to-day encounters with online harassment and their impact on mental health in TGD young adults is essential to understanding drivers of inequities in eating disorders risk, as well as optimizing online safety interventions. This work can then inform optimal policy solutions to support TGD young people’s online safety and overall well-being. Therefore, in this study, we used a daily diary study design to measure the association between exposure to online harassment, disordered eating behaviors, and body dissatisfaction in a sample of US-based TGD young adults.

Method

Data

Participants were recruited for the Interactive Media and Gender-related Stressors and Strengths (IMAGES) Study, a 10-day daily diary study of TGD young adults across the US, conducted September 2022 to August 2023. We used multiple online recruitment strategies, including social media (e.g., Instagram) advertisements and direct digital outreach to lesbian, gay, bisexual, transgender/gender-diverse, and queer/questioning (LGBTQ+) community organizations in every US state. To be eligible for the study, participants were US residents ages 18–25 years, identified as transgender, nonbinary, and/or of another gender identity different than that assigned at birth, could complete the surveys in English, and had internet access. To ensure participant eligibility, eliminate any potentially fraudulent responses, and enhance data quality, we implemented a multi-step screening process, including an online and phone screening with trained research assistants who cross-validated eligibility and explained daily diary procedures. A flow chart of participants in the screening process is provided in Figure 1. To minimize the potential that participants faked eligibility for the study, the recruitment email, flyer, and eligibility screener did not contain the inclusion criteria for participation. All participants completed an informed consent process; this study was approved by the Boston University Medical Campus Institutional Review Board.

Figure 1.

Flow chart of recruited and retained participants in the IMAGES daily diary study 2022–2023

Note: Not all who who did not continue between the eligible for phone screening and contacted for phone screening questions were ineligible. Due to the limited budget, not all eligible participants could be contacted. The final sample was determined based on strategic sampling to ensure a diverse sample across gender and race/ethnicity.

Participants who agreed to participate completed a baseline questionnaire on demographic, social media use, and mental health-related variables. The daily diary study then continued for nine additional days, where each evening, at a time of their choosing, the participant received their daily dairy survey to complete. Participants were compensated $5 for each completed survey and a bonus of $10 for completing all days. There were 152 participants in the daily diary study. The analytic sample included those with at least one daily diary with complete data on online harassment after baseline (n = 150; 2 participants excluded due to missing data on online harassment). The total analytic sample included 1,192 daily diary entries that were at least partially complete without missing data on online harassment exposure. Among the 150 participants in the analytic sample, 88 (59%) completed all nine diaries with data on online harassment, 112 completed at least eight diaries (75%) with data on online harassment, and 128 (85%) completed at least seven diaries with data on online harassment. The distribution of participants in the final analytic sample by recruitment source is provided in Supplemental Table 1. Participants in the analytic sample resided in 36 states throughout the US, and the distribution of participants by state is available in Supplemental Figure 1.

Measurement

Our exposure of interest was online harassment in the past 24 hours. We measured online harassment with a single daily question adapted from the Pew Research Center,¹⁷ asking participants whether they experienced any of the following online: physical threats, stalking, harassment, purposeful embarrassment, and sexual harassment. Responses were yes/no.

There were five outcomes of interest: overeating, binge eating, weight-controlling behavior, coping-related eating behavior, and weight/shape-related body dissatisfaction. Overeating was measured as experiencing an eating binge with or without loss of control. Binge eating was adapted from a measure validated against interviews in prior cohort studies of youth,²² including both overeating and loss of control, and was self-reported as any or none in the past 24 hours.²² Weight-controlling behavior was adapted from the Youth Risk Behavior Surveys and defined as self-report of past-24 hour use of one or more of the following behaviors “to lose weight or keep from gaining weight”: fasting for at least one day, inducing vomiting, taking laxatives, and taking diet pills without a doctor’s advice.²³ We measured coping-related eating behavior using a measure from a daily diary study of sexual minority women,²⁴ which asked how much the participant had eaten due to feeling stressed, sad, or upset, where responses were on a slider scale from “not at all” (0) to “very much” (6). We dichotomized coping-related eating behavior as any (1 or more) versus none (0) due to the relative rarity of higher numbers on the scale. Finally, we measured weight/shape dissatisfaction with the question: “Overall, how satisfied were you with your body shape and weight in the past 24 hours?” Responses ranged from 0 (completely dissatisfied) to 10 (completely satisfied). Higher body dissatisfaction was defined as a score of 3 or lower on the scale (versus lower body dissatisfaction, four or higher).²⁴

Covariates included age, gender, race/ethnicity, baseline negative and positive affect, and average daily negative affect. Gender was categorized as nonbinary/genderqueer, trans man, and trans woman. We categorized race/ethnicity as Asian/Asian American, Black/African American, Latine/Hispanic, multiple races/ethnicities, and White. Baseline positive and negative affect levels were measured with selected items from the Positive and Negative Affect Schedule (PANAS), the modified version of which has two subscales that each range from 5 to 25 (to produce a total summed range from 10–50).²⁵ For each subscale, higher scores indicate a greater frequency of negative emotions. In addition to the baseline modified PANAS score, we assessed daily negative affect, which was measured as the mean daily score for each individual of the five items of the modified negative affect component of the PANAS (the positive items were not included on the daily surveys; thus, the total range of negative affect in the daily diary measures was 5–25). The original PANAS has been validated in both adults²⁵ and adolescents.²⁶

Statistical Analysis

We performed descriptive analysis to characterize the analytic sample by demographic, exposure, and outcome variables. Then, we used multilevel logistic models with Laplace estimation to calculate adjusted odds ratios (aOR) and 95% confidence intervals (CI) for the cross-sectional associations between within-subjects measures of concurrent (same-day) past-24-hour online harassment exposure and disordered eating outcomes while adjusting for between-subjects covariates (age, gender, race/ethnicity, and affect). We selected the past-24-hour time frame based on prior daily diary research showing associations between social stressors and eating behaviors in samples of sexual minority cisgender women.^27–29 In initial models of the four disordered eating outcomes, we included fixed effects for online harassment, age, race/ethnicity, and gender and a random intercept for participants. In the main models, we added fixed effects for within-subjects mean daily negative affect score to assess the degree to which any associations were accounted for by negative affect. To evaluate whether or not an individual’s baseline affect may have been a driver of associations between harassment and eating behaviors over the week, we then performed sensitivity analyses replacing fixed effects for mean daily negative affect score in main models with fixed effects for baseline positive and negative affect score, measured separately. Fit statistics from each step of the model building process are provided in Supplemental Table 2. We performed analyses in R version 4.4.1 and SAS software version 9.4 M8.

Results

Descriptive statistics

Characteristics of the participants in the analytic sample are provided in Table 1. There were 150 participants who contributed a total of 1,192 daily diaries. The mean age of participants was 21.3 years (SD = 2.2). More participants identified as nonbinary or another gender identity (n = 74, 49%) than man/trans man (n = 46, 31%) or woman/trans woman (n = 30, 20%). Participants were distributed across racial/ethnic groups, with the largest being White (n = 51, 34%) and multiple races/ethnicities (n = 41, 27%). During the nine days of daily diary data collection, 31 participants (21%) experienced some form of online harassment, and the mean number of days of online harassment across the full sample was 0.29 (SD = 0.67). Among disordered eating and body dissatisfaction outcomes, coping-related eating behavior was reported among the majority of participants (n = 115, 77%), followed by weight/shape dissatisfaction (n = 81, 54%), overeating (n = 41, 27%), weight-controlling behavior (n = 33, 22%), and binge eating (n = 23, 15%).

Table 1.

Distribution of key covariates, exposures, and outcomes among U.S. transgender and nonbinary young adults in the IMAGES daily diary study analytic sample, 2022–23 (N = 150)

	N	%
Age, m (SD)	21.3	2.2
Gender identity
Woman / Trans woman	30	20%
Man / Trans man	46	31%
Nonbinary or another gender identity(ies)	74	49%
Race/ethnicity
Asian / Asian American	18	12%
Black / African American	24	16%
Latine / Hispanic	16	11%
Multiple races/ethnicities	41	27%
White	51	34%
Baseline modified PANAS, m (SD)
Baseline positive affect subscale	15.0	3.8
Baseline negative affect subscale	13.6	3.9
Person-average daily negative affect subscale, m (SD)	10.4	3.9
Experienced online harassment during daily diary study	31	21%
Mean total days of online harassment during daily diary study, m (SD)	0.29	0.67
Outcomes
Reported overeating during daily diary study	41	27%
Reported binge eating during daily diary study	23	15%
Reported weight-controlling behavior during daily diary study	33	22%
Reported coping-related eating behavior during daily diary study	115	77%
Reported weight/shape dissatisfaction during daily diary study	81	54%

M = mean; SD = standard deviation; PANAS = Positive and Negative Affect Schedule

Note that selected items from the PANAS negative affect subscale ranges from 5–25

Associations between online harassment and disordered eating outcomes

The results of multilevel models are presented in Table 2. In models adjusted for participant age, gender, race, and mean daily negative affect, experiencing online harassment was cross-sectionally associated with greater odds of same-day overeating (aOR = 3.38, 95% CI = 1.02, 9.17), binge eating (aOR = 15.08, 95% CI = 2.73, 83.43), coping-related eating behavior (aOR = 4.07, 95% CI = 1.51, 10.93). We did not find evidence of an association between online harassment and weight/shape dissatisfaction (aOR = 2.02, 95% CI = 0.64, 6.36). Similarly, there was no evidence of an association between online harassment and weight-controlling behavior (aOR = 0.81, 95% CI = 0.18, 3.67). In sensitivity analyses adjusting for baseline positive and negative affect rather than mean daily negative affect (Table 3), the direction and magnitude of results and the degree of statistical precision remained similar to the main reported results.

Table 2.

Concurrent associations between experience of online harassment and disordered eating and body satisfaction outcomes among U.S. transgender and nonbinary young adults in the IMAGES daily diary study analytic sample, 2022–23 (N = 150)

	Model 1a	Model 2b
Outcome	OR (95% CI)	OR (95% CI)
Overeating	3.38 (1.12, 10.20)	3.06 (1.02, 9.17)
Binge eating	19.77 (3.49, 112.14)	15.08 (2.73, 83.43)
Weight-controlling behavior	0.86 (0.18, 4.18)	0.81 (0.18, 3.67)
Coping-related eating behaviord	4.18 (1.56, 11.20)	4.07 (1.51, 10.93)
Weight/shape dissatisfactiond	2.25 (0.70, 7.25)	2.02 (0.64, 6.36)

^aModel 1 is controlled for participant age (continuous), gender, and race

^bModel 2 is controlled for participant age (continuous), gender, race, and mean daily negative affect

^dDue to missing data for one individual in each of coping-related eating behavior and weight/shape dissatisfaction outcomes, n = 149

Table 3.

Sensitivity analysis examining concurrent associations between experiences of online harassment and disordered eating and body satisfaction outcomes among U.S. transgender and nonbinary young adults in the IMAGES daily diary study analytic sample, 2022–23 (N = 150)

Outcome	OR (95% CI)
Overeating	3.12 (1.05, 9.32)
Binge eating	16.98 (3.06, 94.23)
Weight-controlling behavior	0.84 (0.18, 3.88)
Coping-related eating behaviora	4.00 (1.50, 10.67)
Weight/shape dissatisfactiona	1.96 (0.62, 6.20)

Models are controlled for participant age (continuous), gender, race, and baseline positive and negative affect

^aDue to missing data for one individual in each of coping-related eating behavior and weight/shape dissatisfaction outcomes, n = 149

Discussion

In this study of TGD young adults, experiences of online harassment were prevalent and associated with higher odds of several disordered eating behaviors. Even within the 9-day window of the daily diary study, over one in five participants reported experiencing online harassment, emphasizing the frequency of these adverse digital encounters in daily life. Reporting online harassment was associated with a higher likelihood of overeating, binge eating, and coping-related eating behavior on that day, highlighting the relationship between online harassment and disordered eating behaviors in this marginalized population. Together, these results point to potential links between online harassment and disordered eating behaviors, which can take a substantial toll on psychiatric morbidity and quality of life.⁸ Given documented shortcomings in platforms’ enforcement of online harassment protections,³⁰ our findings underscore the need for public health policy designed to increase online safety for this population to reduce exposure to this potential preventable contributor to adverse mental health.

While no studies to our knowledge have been published on the relationship between digitally based (online) harassment and disordered eating behavior or body dissatisfaction in TGD young adults, our results on overeating, binge eating, and coping-related eating are consistent with theoretical and empirical work on other forms of harassment exposure and mental health outcomes in TGD populations. Specifically, these findings accord with minority stress theory, which posits that disparities in mental health outcomes disproportionately burdening sexual and gender minority populations are driven by experiences of sexual and gender minority-related interpersonal discrimination.¹³ Consistent with this theory, empirical research has supported a relationship between experiences of non-online harassment and body dissatisfaction¹¹ and disordered eating (e.g., binge eating)¹⁶ in TGD people, including TGD young adults. It is essential, then, that we understand how these findings extend to social interactions in the online environment, where many TGD adolescents and adults report experiencing harassment and abuse.^18,21 Research on online harassment and broader mental health outcomes among TGD adults point to the potential effects these negative interactions have on psychological distress.³¹ Our work expands upon these findings to emphasize the particular role of online harassment and abuse in disordered eating for TGD young people.

Notably, over three in four participants reported at least one disordered eating behavior over the 9 day period. Using a range of measures (most commonly the 5-item Sick, Control, One, Fat, Food [SCOFF] questionnaire), prior studies have observed prevalence estimates of 5–20% for eating disorders symptoms among the general population of adolescents and young adults.^32,33 A recent study of US adolescents found much higher prevalence estimates for individual behaviors in the past year (e.g., 28–42% of the sample engaged in binge eating in the past year).³⁴ We posit that the observed prevalence of some disordered eating behaviors in this study is linked to experiences of online harassment. However, it is clear that other factors also contribute to disordered eating in this population. For instance, in contrast to our findings on overeating, binge eating, and coping-related eating, we did not find an association between online harassment and body dissatisfaction or weight-controlling behavior. Further, the point estimate for the association between online harassment and weight-controlling behavior was protective (i.e., below zero). While not statistically significant, such a point estimate suggests lower likelihood of engaging in weight-controlling behaviors on days when online harassment occurred. Additionally, even for behaviors clearly associated with online harassment in our sample, the prevalence of disordered eating in the 9-day daily diary period far exceeded the prevalence of online harassment. Other drivers of disordered eating in this population which may impact our sample include perfectionism, anxiety symptomology, poor self-image, and other (non-online harassment) forms of interpersonal discrimination.¹¹ Our findings nonetheless contribute to the literature by emphasizing the potential ways in which online harassment may fit into this constellation of risk factors for TGD young adult disordered eating.

Importantly, this study focused on online social interactions, which have been the center of recent public health conversations, including a 2023 Surgeon General’s report.³⁵ This report drew attention to the critical need to better understand the consequences of online interactions, especially in stigmatized populations such as TGD young people, as these online experiences can both provide vital social support and resources to stigmatized communities and also put them at risk of harassment and abuse.³⁵ Notably, young adults^36,37 and sexual and gender minorities^18,37 are at higher risk of experiencing online harassment or cyberbullying than their peers, making this issue particularly relevant in the TGD young adult population. In particular, compared to the 21% of our sample who experienced online harassment in just a 9-day period, the Youth Risk Behavior Survey of the general US adolescent population estimates between a 16–17% prevalence of cyberbullying in the past year.³⁸ This risk is of rising concern given the current US policy environment with proliferating anti-TGD legislative efforts, which may foster perpetration of harassment towards TGD people.¹²

Strengths and limitations

This study has several strengths. We used a robust daily diary design with repeated measures on participants, resulting in almost 1,200 diaries in the analytic sample. This level of detail on the day-to-day experiences of TGD young adults is beneficial to implementing effective public health policy in this population. Additionally, we examined disordered eating behaviors, which are less often studied but are of particular public health relevance due to their high burden on psychiatric morbidity, mortality, and quality of life.⁷

This study also has several limitations. First, because we measured online harassment reporting and disordered eating behaviors on the same day, we cannot rule out the possibility that results were driven by reverse causation. Disordered eating behaviors could lead to increased time in online spaces, thereby increasing exposure to potential online abuse. To address this limitation, we performed lagged analyses to examine the effect of online abuse on next-day disordered eating. Given that reduced sample size led to reduced statistical precision in these lagged analyses, future studies with larger sample sizes should explore how the effect of online abuse exposure varies on disordered eating outcomes with differing lag times. Researchers may also consider utilizing strategies such as ecological momentary assessment(s), with its more frequent, random moments of data collection, to improve temporality. Furthermore, because we did not collect data on participant history of exposure and outcome at baseline using the same measures as the daily diary study, we were unable to control for those variables in the analysis. Including these baseline measures would strengthen the potential causal inferences from this study. Future work should collect comprehensive data on both online harassment and disordered eating at baseline using the same measures included in the daily diary study for potential refinement of analyses. Additionally, the self-reported online harassment exposure and disordered eating outcomes may be affected by recall bias as participants may not accurately remember or report such experiences, leading to potential likely non-differential misclassification of these variables and increasing imprecision of estimates. Moreover, a single-item online harassment measure is unlikely to comprehensively capture the range of experiences participants may have with online harassment. Further, the online harassment measure did not specify whether participants attributed the harassment to their gender identity, and therefore our findings likely reflect both gender-based and other forms of online harassment. The development of brief, valid, and reliable scales of recent (e.g., daily) exposure to online harassment is essential to further research in this area. Additionally, there is potential for selection bias, because participants self-select to enroll in the study, and recruitment was conducted online, including with the help of social media content creators involved in TGD community advocacy. As a result, the sample may be more likely to be engaged in online advocacy themselves and thus may have higher risk on average of online harassment leading to bias in estimates. Finally, we used odds ratios in our results, even though we do not meet the rare disease assumption for several of our outcomes. While prevalence ratios would be preferred in this setting, we faced significant model convergence issues in doing so. Therefore, while we present odds ratios here, it is important to note that these estimates may not approximate the true prevalence ratio.

The need to address the potential harms of online harassment to the mental health and well-being of TGD young people is more urgent than ever. This is particularly true in the current US political environment, given record-breaking numbers of legislative and other policy-based attacks on gender-affirming care. Gender-affirming care is of notable importance because it can reduce the heightened risk of eating disorders in TGD young adults.^11,39 Further, recent shifts in internal policy-making by social media platforms have eliminated content moderation and features designed to enhance users’ online safety.⁴⁰ Researchers must conduct studies to identify effective interventions to improve safety for TGD youth specifically, which have not yet been investigated. Innovative research design may aid in evaluating these interventions’ success, including more objective assessments of exposure to online harassment. Specifically, big data and machine learning methods may be useful to analyze social media content and changes in response to safety interventions. Additionally, advocates have called on social media companies to be accountable for developing and enforcing anti-harassment policies and incorporating design features that ensure a safe environment on their platforms.⁴¹ However, interventions seeking to enhance online safety should also account for the real benefits of social and informational support online⁴² and protect access to such vital resources for TGD young adults. Researchers and policymakers may consider how emerging platforms with more robust anti-harassment policies could serve as safer digital environments for TGD young adults.

Experiencing online harassment was prevalent and associated with disordered eating behaviors in this daily diary study of TGD young adults. Further work is necessary to create safer online environments for TGD young adults and young people of all genders.

Public significance statement:

Transgender and gender-diverse young adults are at higher risk for disordered eating and eating disorders compared to their cisgender peers. Online harassment is prevalent in this population and may contribute to their elevated risk. Public health and policy interventions are necessary to increase online safety and social support for this marginalized population.

Data, materials, and code availability statement:

Due to the sensitive nature of these data, they cannot be made publicly available. The research team may be able to make specific analyses available upon reasonable request. Code to generate these analyses is available upon request.