Beyond disability: how being twice-marginalized shapes the experiences of STEM undergraduates with disabilities Authors: Riley B. Pizza, Danielle Pais, Logan E. Gin, Katelyn M. Cooper, Sara E. Brownell, Emma Goodwin Published: 2026 ABSTRACT ABSTRACTThe experiences of marginalized students in higher education are often examined through a single-identity lens, overlooking how other identities can jointly shape their college experiences. In particular, little is known about how holding an additional marginalized identity affects the experiences of students with disabilities in STEM. We surveyed 947 undergraduates (379 with disabilities) at 85 institutions across the United States to (i) identify marginalized identities that are highly represented among students with disabilities and (ii) examine whether holding an additional marginalized identity corresponds to distinct experiences among students with disabilities. We found that LGBTQ+, nonbinary, and white students were overrepresented among students with disabilities relative to students without disabilities, with 50% of students with disabilities identifying as LGBTQ+. LGBTQ+ students with disabilities reported significantly more negative academic and social experiences than non-LGBTQ+ students with disabilities. These findings highlight some of the heterogeneity that exists in the experiences of students with disabilities, underscore the need to address the challenges faced by LGBTQ+ students with disabilities, and suggest that some patterns previously attributed to disability alone could be partly explained by the experiences of LGBTQ+ students. INTRODUCTION Although research into the experiences of students with disabilities has expanded in the last decade (1, 2), most studies focus solely on disability status (3) and do not consider how students’ educational experiences could be jointly shaped by an additional marginalized identity that a student holds (4). As such, existing research may be overlooking important variation in the experiences of students with disabilities. Beginning to understand this variation is relevant for supporting accessible educational pathways into STEM (science, technology, engineering, and math) degrees and strengthening the future STEM workforce (5, 6). Thus, research is needed to understand how multiple forms of marginalization could influence the educational experiences of students with disabilities. In the context of this research, we define “disability” as an impairment that affects a student’s education, due either to the disability itself or to disability-related structural or social factors (7, 8). We also use person-first language (i.e., “Student with a disability”) throughout this paper in accordance with the APA style guide (9) and because many individuals with disabilities, including authors of this research, prefer to be a person who also has a disability (10). However, we acknowledge considerable debate over the use of person-first language, and general preference for identity-first language in the autistic (11), deaf (12), and blind (13) communities. Challenges faced by undergraduate students with disabilities in STEM Legislation requiring universities to provide equal access for students with disabilities (14, 15), as well as increased diagnoses and prevalence of disabilities in the population (16), have contributed to an increase in the proportion of students with disabilities in undergraduate STEM education. However, students with disabilities still face substantial challenges in higher education, including longer times to degree completion (17), lower GPAs (18), and high attrition rates (19–21), compared to their peers who do not identify as having disabilities. Barriers to success for students with disabilities are particularly pronounced in STEM. STEM instructors report challenges in meeting the needs of students with disabilities and may be less willing (if given a choice) to provide accommodations than non-STEM instructors (2, 22, but see 23). Students also report perceiving that their STEM instructors are less amenable to providing accommodations, which can discourage them from self-advocating and receiving accommodations (24). Combined, these factors may contribute to data showing that STEM students receive fewer accommodations than their peers in other fields, despite legal mandates to accommodate students with disabilities (25, 26). Furthermore, standard accommodations were designed for traditional lecture courses but are less applicable to many current STEM courses. For example, there have been recommendations to transition STEM courses from lectures to student-centered active-learning with multiple low-stakes assessments, but typically no standard accommodations exist for these courses (27–29). Additionally, many STEM courses are now fully or partially online (30–32), and students with disabilities report feeling inadequately accommodated in online environments compared with in-person settings (33–35). Finally, undergraduate participation in STEM research experiences for course credit can pose challenges for students with disabilities, as students are often unaware of the accommodations they could receive (36) and are likely to conceal their disabilities, increasing the risk of negative experiences (27, 34). While this literature highlights broadly negative experiences for students with disabilities in undergraduate STEM settings, it rarely considers other identities these students hold, leaving it unclear whether these patterns are consistent for students who hold an additional marginalized identity. Who are students with disabilities (beyond their disability)? The first step to understanding the diverse experiences of students with disabilities in STEM is to characterize the demographic identities represented within this population, yet existing data provide an incomplete and sometimes contradictory picture. National surveys suggest that the population of students with disabilities is more racially diverse and includes more women than the population without disabilities (37), but they do not report identities such as first-generation status, socioeconomic background, or LGBTQ+ (lesbian, gay, bisexual, transgender, and queer) identities. Research in the general population shows higher disability prevalence among low-income (38) and LGBTQ+ populations (39), although these patterns may differ in higher education due to barriers to accessing diagnoses. For example, accommodation rates of students at universities that primarily support high-income students have risen sharply, whereas accommodation rates at colleges and universities that primarily serve low-income students have stayed the same, indicating that low-income students may have less access to accommodations in higher education (40). The largest (N = 189) and most comprehensive national study of diversity among college students with disabilities found that white, first-generation, and LGBTQ+ students were overrepresented relative to the general student population (41). Whether these demographic patterns hold within STEM fields, which have historically underserved low-income, first-generation, LGBTQ+, women, and students of color (42, 43), remains unclear. What experiences do multiply marginalized students have in higher education? Understanding how holding an additional marginalized identity shapes the experiences of students with disabilities in STEM requires integrating research on both disability and twice marginalization (44–46). Studies across higher education consistently show that twice-marginalized students, or those who hold two marginalized identities (47), have more negative educational experiences than students with only one, as they experience marginalization across multiple axes (48). While not specific to disability, these findings suggest that students with an additional marginalized identity may have different experiences than students whose only marginalized identity is their disability, including distinct embodied disability experiences, challenges navigating their courses given their disability, and a general sense of belonging at their institution. Students’ embodied experiences of their disability are shaped both by their disability itself and by cultural and social factors (7, 8). Healthcare research shows that individuals with fewer resources and those facing stigma due to their identities (e.g., women, racial minorities, and LGBTQ+ individuals) report greater disability severity and higher levels of pain than individuals from more privileged groups (49–51). In higher education, more painful conditions are associated with lower perceived social support (52), and more severe disabilities reduce the likelihood of remaining enrolled (53). These patterns suggest that students with another marginalized identity who experience greater disability-related severity or pain may face additional academic challenges. Twice-marginalized students may also encounter unique challenges navigating college as a student with a disability. Holding an additional marginalized identity can create opportunities for compounded stigma, as members from one identity can hold negative attitudes toward another (54, 55), increasing perceived discrimination related to their disability (56). This can often lead to differences in how likely an individual from a given racial group is to self-identify as having a disability (57). This can discourage students from using accommodations or seeking support (58), leading to worse academic outcomes (59). Additionally, students who hold both a typically visible identity (e.g., race/ethnicity) and an invisible one (e.g., most disabilities in higher education) often perceive the invisible identity as less central to their self-concept (60). Because identity centrality predicts self-advocacy (61), which, in turn, supports better academic and social outcomes (62), students with multiple marginalized identities may engage in less self-advocacy and perceive less support in their college experiences. Consistent with these patterns, twice-marginalized students report lower campus belonging (63) and higher psychological distress (64) than students with only one marginalized identity, which can negatively affect academic outcomes (65). Although theoretically grounded in the literature, these hypothesized relationships between holding an additional marginalized identity and disability experiences have rarely been examined empirically. We identified only seven studies that assess whether students with disabilities who also hold additional marginalized identities experience different challenges than students whose only marginalized identity is disability. Most of these studies identify more negative experiences for multiply marginalized students (but see reference 66), including higher barriers to education for non-white (67–69), first-generation (70), and low-income (71) students, and more reported discrimination among LGBTQ+ (72) and Latinx (69) students with disabilities. These studies have primarily been small-scale interview studies at single institutions (but see reference 72) and include students of all majors, so it is unclear whether and how these experiences translate to measurable differences with a nationally collected sample of students in STEM. Together, these gaps underscore the need for larger-scale studies that characterize both the demographic composition and diversity of lived experiences of students with disabilities in STEM. To address this need, we conducted a national survey of students with and without disabilities to explore the following. RQ1: How does the representation of different demographic groups (gender, race/ethnicity, first-generation, LGBTQ+, and socioeconomic status) differ between STEM students with and without disabilities? RQ2: How does holding an additional marginalized identity shape STEM students with disabilities (i) embodied experience of their disability, (ii) disability-related experiences on campus, and (iii) their broader campus experiences? Positionality Some of the authors identify as having a physical disability and/or identify as having conditions that would be considered disabilities to be accommodated, including ADHD, dyscalculia, anxiety, and depression. Our team also includes individuals who identify as women, nonbinary, and LGBTQ+. Our research questions, survey design, analyses, and interpretations are informed by our lived experiences navigating STEM with these identities, as well as by mentoring and supporting students with disabilities in diverse higher education settings. We recognize that the experiences of students with disabilities are complex and cannot be fully captured by any single study. Our goal is to raise awareness and highlight areas for growth in improving accessibility in STEM higher education. METHODS We conducted a national survey of STEM students with and without disabilities in February 2022. Participant recruitment and data cleaning We emailed all publicly listed disability resource center coordinators (N = 1,234) and asked them to forward our study recruitment information to recruit students who were registered with their university’s disability resource center. We also contacted all publicly listed undergraduate life sciences coordinators (N = 1,181) to recruit students without disabilities. As an incentive, participants could opt into a drawing for one of four $50 gift cards. In total, we received 1,954 student responses from 85 institutions. We removed incomplete or low-quality responses (answered <10% of items or completed survey in <3 min; N = 381). We additionally removed responses with missing or incomplete data relevant to our research questions, including non-STEM majors, graduate, or post-baccalaureate students (N = 364), and anyone who did not report their university affiliation (N = 147). Students identifying as multi-racial (N = 67) or as a race not listed (N = 26) were also removed from the study sample, as it was unclear whether their experiences were more closely aligned with those of people of color or white students. Since our focus was on institutional experiences, we excluded students with disabilities who were not affiliated with their disability resource center (N = 22), although we acknowledge that there are barriers and stigma associated with diagnosis. This resulted in a final study sample of 947 students, including 379 students with disabilities and 568 without disabilities. We use the full study sample (N = 947) for RQ1, while only students with disabilities who also reported the full set of demographic factors collected in our study (N = 317) are included in analyses for RQ2. Demographics We collected data on five self-identified demographic factors known to influence student experiences: gender, race/ethnicity, LGBTQ+ identity, first-generation status, and socioeconomic status (73–76). Students identified their gender as “woman,” “man,” or “nonbinary.” Race/ethnicity was self-reported from five options (see Supplement A), but due to small sample sizes, we consolidated categories into “white” or “person of color” (POC). Asian students were grouped with other underrepresented groups due to more similar rates of disability in higher education compared to white students (77) and reported shared experiences of exclusion and racism in higher education settings (78–80). Students self-identified as LGBTQ+ or not. While we use LGBTQ+ as an umbrella term to be inclusive of individuals who do not identify as either straight or cisgender, we recognize that each identity within this group can confer different experiences (81). Students were classified as first generation if no parent held a bachelor’s degree. Finally, students were grouped into two socioeconomic categories based on their family’s income: low income (<$50,000) and middle/high income (≥$50,000). Students self-reported whether they had a disability, selecting from a list of disability types previously identified by disability resource directors as categories of students that may require different accommodations (82). Some students chose to self-describe their disability in the “other” category, and we assigned these responses to our pre-determined categories when applicable. Survey development We developed a survey to capture the experiences of college students with disabilities, including their embodied experiences, disability-related campus experiences, general campus experiences, and negative mental health outcomes. All items used a 6-point Likert scale (1 = strongly disagree, 6 = strongly agree) unless otherwise noted (see Supplement A for analyzed questions). Only students who identified as having a disability responded to these items, and those with multiple disabilities were asked to answer considering all their disabilities collectively. Disability characteristics We assessed how disabilities influenced students’ daily life and self-perception using five measures. These included three single-item measures to explore perceptions of disability impairment (severity “My disability significantly affects my daily life”, pain “I experience physical pain because of my disability”; 1 = never, 5 = always, and noticeability “Other people can easily tell that I have a disability”; 1 = never, 5 = always [83]). We characterized students’ disability identity using a one-item scale to measure disability salience (“I spend a lot of time thinking about my disability” [84]) and a six-item scale to measure centrality of the disability identity (e.g., “My disability identity is an important reflection of who I am” and “In general, my disability is an important part of my self-image” [84]). Disability-related experiences We measured six aspects of students’ experiences related to stigma, disclosure, and support. Perceived stigma: eight items to measure students modified from the Stigma Scale for Chronic Illnesses (e.g., “I feel embarrassed because of my [disability]-related limitations” [85]). Class stigma: five items to assess perceived stigma from instructors (e.g., “I feel my instructors are not willing to provide requested accommodations”), adapted from the “Class stigma” construct of the 43-item “College Students with Disabilities Campus Climate” (CSDCC) survey (86). DRC support: four-item construct from the CSDCC survey evaluated students’ satisfaction with disability resource center (DRC) services (e.g., “I feel supported by the staff at my DRC”). Use of accommodations: four-item construct from the CSDCC assessing accommodation use (e.g., “I only use accommodations when absolutely necessary”). We removed the item “I request faculty notification letters from Disability Services” since all our retained respondents were affiliated with their disability resource center, and there are institutional differences that complicate the interpretation of this item: students sometimes have selective control over which instructors receive letters, while at other institutions, the DRC sends letters to all of the students’ instructors. Comfort with disclosure: seven items modified from Baker et al. (87), examining comfort discussing disability in academic contexts. Items were adapted to clearly refer to comfort disclosing to both students and professors (e.g., “I am reluctant to disclose the nature of my disability(ies)” was replaced with “I prefer not to disclose the nature of my disability(ies) with my professor” and “I prefer not to disclose the nature of my disability(ies) with other students”). Comfort on campus: four-item construct from the CSDCC survey (86) capturing overall campus support (e.g., “I feel the overall campus environment is supportive of students with disabilities”). General campus experiences We used the six-item Perceived Cohesion scale (88) to assess sense of belonging and morale (e.g., “I see myself as part of my university’s community” [belonging] and “I am enthusiastic about my university” [morale]). Data analysis All analyses were performed using RStudio (89). Treatment of demographic and categorical variables For demographic analyses (RQ1), we used all three gender categories (“man,” “woman,” and “nonbinary”), but for regression analyses (RQ2), we consolidated them into “men” and “women/nonbinary.” This was done because all nonbinary participants also identified as LGBTQ+, which created multicollinearity in regression models. Additionally, women and nonbinary individuals are both historically excluded from STEM and likely share similar experiences of gender-related stigma (90, 91). Confirmatory factor analysis To assess that the survey items operated as intended within the defined survey constructs and with our student population, we conducted confirmatory factor analyses (CFA) on all multi-item factors for students with disabilities (92; for a comprehensive explanation of our CFA methods and specific fit indices, see Supplement C). When low factor loadings or model modification indices highlighted dysfunctional items, we excluded or correlated related items if they were theoretically distinct from other items. With these modifications, all but one of our factors met the stringent “good” model fit criteria (CFI & TLI > 0.95, SRMR <0.6 [93]). “Perceived stigma of disability” had a slightly lower model fit, which still met the criteria of “fair” model fit (94). To account for variable loadings for different items, we computed weighted factor scores (95) using “lavPredict” (package “lavaan,” version 0.6-19 [96]). Scale reliability within our studied population was confirmed through measures of McDonald’s omega (all ω > 0.7 [92, 97]) (Supplement Table C2). We tested measurement invariance to verify that our survey functioned consistently across demographic groups within the students with disabilities subset of data (98). Due to severe imbalance in group sizes, we could not test invariance for race/ethnicity or gender (99). Most of our measures met acceptable criteria for measurement invariance, although minor deviations occurred at the metric level for “comfort on campus” and “disclosure” by first-generation and income status, and at the scalar level for “perceived stigma” and “class stigma” by LGBTQ+ identity (see Supplement Table C3-C5 for fit statistics). This could indicate that these student groups have slightly different interpretations of the items in these scales; however, given the small magnitude of these differences, we proceeded with latent mean comparisons. RQ1 analysis: demographic comparisons We compared demographic characteristics between respondents with and without disabilities using chi-square tests and calculated odds ratios to quantify effect sizes (“emmeans” version 1.11-2.8 [100]). RQ2 analysis: differences in experiences To examine how being twice-marginalized shapes the experiences of students with disabilities, we fit linear models for each of our 12 response variables (“glmmTMB” version 1.1.14 [101]). We removed students without disabilities from these analyses as they were not administered these disability-specific questions. Models used Gaussian (weighted factor scores) or Poisson (ordinal responses) error distributions as appropriate (see Supplement B, Table B1). All five demographic variables (gender, race/ethnicity, LGBTQ+ identity, first-generation status, and socioeconomic status) were initially included as fixed effects. However, race/ethnicity and first-generation status did not improve model fit in any model (ΔAIC < 2 in all cases, and typically ΔAIC < −1), and were therefore excluded from final analyses. A priori power analyses indicate that our study is sufficiently powered to detect medium effects for each of our fixed effects (α = 0.05, d = 0.5, power = 0.80; approximately 63 participants required in each group). A priori power analysis also indicated that detecting a small two-way interaction effect (α = 0.05, d = 0.2, power = 0.80) would require approximately 400 participants total. Therefore, we focused on main effects rather than testing interaction terms. For response variables that may differ based on the context of their institution (e.g., comfort on campus), institution type was included as a covariate. To account for multiple comparisons across many models, and to reduce the type I error rate, we conducted Benjamini–Hochberg corrections on regression P-values (102). Given the number of statistical tests performed, and the number of small effects, results with adjusted P-values (Padj) between 0.05 and 0.10 are described as “suggestive,” recognizing that they do not meet conventional thresholds for statistical significance after correction. RESULTS Students with disabilities in STEM are more likely to be white, LGBTQ+, and nonbinary The demographic makeup of students with disabilities in STEM (N = 379) differed from that of students without disabilities (N = 568; Fig. 1A; Supplement D, Table D1). Students with disabilities were 30% more likely to identify as nonbinary and 14% less likely to identify as a woman compared to students without disabilities, while the percent of students identifying as men did not differ between groups (Supplement D, Table D2). Students with disabilities were 2.3× more likely to identify as LGBTQ+ and 61% less likely to be people of color than students without disabilities. The percentage of students who identified as first-generation or low-income did not significantly differ between groups. Demographic characteristics of survey respondents: (A) distribution of demographic identities among students with and without disabilities, (B) number of conditions associated with disability per student, and (C) frequency of reported disability conditions (see Supplement D, Table D1, for statistics). Among students with disabilities, most reported more than one condition (Fig. 1B). Mental health conditions (e.g., anxiety, depression) were most common (73.9% of reported conditions), followed by learning or cognitive disabilities (55.6%; e.g., ADHD, dyscalculia) and chronic conditions (26.5%; e.g., diabetes; Fig. 1C). Physical disabilities were reported by 12.5% of students, and fewer than 10% reported hearing or vision disabilities. Experiences of students with disabilities are not uniform across demographic groups Students’ embodied experiences of their disabilities differed across intersecting identities (Fig. 2; Supplement D, Table D3). Low-income students reported 24% higher disability-related pain than middle-/high-income students. There was suggestive evidence (P = 0.01, Padj = 0.07) that LGBTQ+ students reported greater severity of their disability than non-LGBTQ+ students. There was also suggestive evidence (P = 0.01, Padj = 0.07) that women/nonbinary students reported higher salience of their disability identity than men. Regression coefficients showing the effects of twice marginalization on embodied disability experiences among students with disabilities. Points represent coefficients for historically excluded groups (women/nonbinary, LGBTQ+, low-income) relative to reference groups (men, non-LGBTQ+, and middle/high income). Error bars indicate 95% confidence intervals, and symbols correspond to Benjamini–Hochberg adjusted P-values. Full model results are provided in Supplement D, Table D3. † P < 0.1; * P < 0.05. Disability-related college experiences varied across demographic identities, particularly for LGBTQ+ students (Fig. 3; Supplement D, Table D3). Compared to non-LGBTQ+ students with disabilities, LGBTQ+ students with disabilities were about twice as likely to avoid using accommodations, reported almost twofold lower comfort with disclosing their disability, and 1.5× greater discomfort with the campus disability climate. There was suggestive evidence (P = 0.02, Padj = 0.09) that LGBTQ+ students with disabilities reported higher disability stigma in academic settings. Students across all groups reported high satisfaction with their campus disability resource centers (Fig. 3; Supplement D, Table D3). The general college experiences of students with disabilities were more consistent across demographic groups (Fig. 3, Supplement D, Table D3). However, low-income students reported approximately 1.5× lower sense of belonging than middle-/high-income students. Institution type was not a significant predictor in any analysis. Regression coefficients showing the effects of twice marginalization on college experiences among students with disabilities. Points represent coefficients for historically excluded groups (women/nonbinary, LGBTQ+, and low income) relative to reference groups (men, non-LGBTQ+, and middle/high income). Error bars indicate 95% confidence intervals, and symbols correspond to Benjamini–Hochberg adjusted P-values. Full model results are provided in Supplement D, Table D3. † P < 0.1; * P < 0.05. DISCUSSION To our knowledge, we present the largest and most comprehensive depiction to date of the diverse experiences of undergraduate STEM students with disabilities who are registered with their university’s disability resource center. Our results show that LGBTQ+ students are more highly represented among students with disabilities compared to students without disabilities, and report substantially more negative experiences, especially regarding perceived stigma and ease of using their accommodations, than non-LGBTQ+students with disabilities. LGBTQ+ and nonbinary individuals are more represented among STEM students with disabilities compared to students without disabilities Our results show that LGBTQ+ students are more represented among STEM students with disabilities, with half of this group identifying as LGBTQ+, compared to only 20% in the population without disabilities. This proportion far exceeds recent estimates of LGBTQ+ representation, which ranged from 24% in the general population of individuals with disabilities (103) to 37% among students with disabilities in higher education (41). We also found that ~14% of students with disabilities in this study were nonbinary. Although national data on nonbinary students with disabilities are lacking, this result aligns with prior research indicating high rates of disability among the transgender population (104). Since our recruitment was not targeted at LGBTQ+ students, and the 20% LGBTQ+ representation among students without disabilities in our sample aligns with recent national surveys of college students overall (105, 106), it is unlikely that our survey reached a disproportionate number of LGBTQ+ students compared with previous ones. If anything, we expected lower-than-average representation of LGBTQ+ students in our study, since they leave STEM for other disciplines at higher rates (107–109). Although our data cannot identify mechanisms driving this pattern, individuals with disabilities may be more accepting of their LGBTQ+ identity (110), or LGBTQ+ individuals may be at greater risk of mental health conditions due to discrimination compared to non-LGBTQ+ students (111). Taken together, it is likely that previous work has underestimated the representation of LGBTQ+ and nonbinary students with disabilities in higher education. These findings highlight the importance of considering LGBTQ+ and nonbinary identities when conducting research on the experiences of students with disabilities, and students more broadly. Although research on LGBTQ+ students has expanded (112, 113), demographic questions about LGBTQ+ identity remain absent from many higher education research surveys (81). This has limited our understanding of their prevalence, particularly among students with disabilities. Similarly, survey questions about gender have historically only included binary man/woman categories (114), and some states require binary categories for data collection (115), limiting data on nonbinary students. Even when nonbinary options are included, these respondents are often excluded from analyses due to small sample sizes (114, 116). Because students with disabilities are disproportionately likely to be nonbinary, our results highlight that the frequent exclusion of nonbinary individuals in research may disproportionately remove responses from students with disabilities, masking disparities. To improve accuracy in studies reflecting the experiences of students with disabilities, researchers should ensure that LGBTQ+ and nonbinary students are identified, surveyed, and included in analyses. When sample sizes permit, nonbinary identities should be analyzed as a distinct category; when they do not, grouping nonbinary and women students would be preferable to exclusion (117). Among students with disabilities, LGBTQ+ students report more negative experiences than non-LGBTQ+ students Our data expose a consistent pattern: among students with disabilities, LGBTQ+ students report more negative experiences than both non-LGBTQ+ students and students who hold other marginalized identities. These findings may help explain the previously identified negative educational outcomes (59) and higher attrition rates of LGBTQ+ and students with disabilities from higher education (21, 108). Notably, we found that LGBTQ+ students with disabilities may be under-utilizing accommodations that could help support their success in higher education, given that LGBTQ+ students are unlikely to disclose their disability to their instructors. One possible explanation is that LGBTQ+ students often make deliberate, context-specific decisions about disclosure, choosing not to “come out” in environments perceived as risky, including STEM classrooms (118). This aligns with qualitative research showing that LGBTQ+ students with disabilities who view both identities as concealable often perceive nondisclosure as the “path of least resistance” to acceptance and belonging (119). Notably, this finding presents some contradiction to earlier research findings that LGBTQ+ students are more likely than non-LGBTQ+ students to disclose depression to instructors in online science courses, especially to receive accommodations. Our results suggest that patterns about disclosing depression may not translate to disabilities broadly. Together, these findings suggest that barriers previously shown to drive LGBTQ+ students to leave STEM majors (107–109) may also disproportionally constrain students with disabilities. This is particularly concerning because LGBTQ+ students comprised ~50% of students with disabilities in our sample, highlighting the need for targeted interventions to improve retention and diversity in STEM. While research into factors increasing retention for LGBTQ+ students with disabilities is limited, we can identify potential interventions from studies on LGBTQ+ populations. Prior research demonstrates that LGBTQ+ representation, such as instructors openly identifying as LGBTQ+, is associated with increased sense of belonging, increased interest in STEM careers, and reduced stigma among LGBTQ+ students (120–122). Representation can also be fostered when non-LGBTQ+ faculty highlight scientists with LGBTQ+ identities and disabilities (123, 124) or design inclusive curricula, particularly in biology courses where sex and gender intersect with content (125, 126). Finally, visible signals of LGBTQ+ inclusion—such as rainbow pins or flags—have been shown to increase perceptions of safety and belonging (127, 128), so faculty or DRC staff displaying acceptance of LGBTQ+ identities may increase students’ comfort level with utilizing campus resources. Patterns of access, belonging, and well-being for low-income students with disabilities Despite comparable representation and accommodation use, low-income students with disabilities still experience barriers that may undermine STEM persistence. Contrary to previous research, low-income students in our study were equally represented among students with and without disabilities and were just as likely to use accommodations (38, 40), suggesting increased access to accommodations and diagnoses over time. However, they also report markedly higher disability-related pain and lower belonging on campus. Higher levels of pain have been demonstrated among low-income individuals with disabilities in the general population (50) and may indicate lower access to care, even among university students. Importantly, low-income students with disabilities were the only multiply marginalized group in our sample to report lower belonging on campus overall, a known predictor of STEM persistence for low-income students (129). Taken together, this suggests that even as access to accommodations improves, low-income disabled students may still leave STEM at higher rates due to medical and social barriers. Limitations While our data set represents one of the most diverse national samples of students with disabilities to date, we faced limitations when aggregating demographic groups. We were unable to test whether nonbinary students reported different experiences than women with disabilities, despite research showing that nonbinary students face different challenges than their cisgender LGB (lesbian, gay, bisexual) and binary transgender peers (130, 131). We were also unable to consider all twice-marginalized identities, such as multiracial students with disabilities or heterogeneity within racial identities, limiting our ability to fully explore the diversity of this population. Similarly, since we did not test for interactive effects, we were unable to capture the full range of intersectional experiences within our groups. In the context of our results, there is likely additional heterogeneity among the LGBTQ+ population (e.g., variation across racial identities in this group) that we could not test for (110). Additionally, since women tend to participate more in research (132), and women were the majority of our sample, our data may underrepresent the experiences of men compared to the population of STEM students with disabilities. We also grouped students with all types of disabilities together, a necessity given the limited diversity of conditions in our sample and the comorbidity of conditions. Yet, students with different types of conditions (e.g., visible vs. invisible) may have distinct experiences (82), and some conditions may be more prevalent among some marginalized groups (104). Finally, because we focused only on students with official diagnoses and university accommodations, the patterns observed in this study may not generalize to students with disabilities who are not formally diagnosed or registered. Thus, this research serves as a call to action for future research on the experiences of nonbinary, racially diverse, and non-registered students with disabilities. Conclusions Our findings demonstrate that experiences of students with disabilities are not homogeneous: LGBTQ+ students specifically report worse academic and social outcomes than non-LGBTQ+ students. Our study illustrates the importance of considering LGBTQ+ identities and including nonbinary individuals in research, as omitting them can disproportionately exclude both LGBTQ+ students and students with disabilities. We also observed no effect of institution type on our results, suggesting these findings are broadly applicable. Notably, because half of the students with disabilities in our sample also identified as LGBTQ+, some patterns historically attributed to disability may instead reflect challenges because of LGBTQ+ identity, emphasizing the need to consider the additional marginalized identities that students hold when characterizing their experiences.