Homelessness and Disability in Public-School Students

Data are from the Federal and State Departments of Education. We initially attempted to compile data from all 50 states and territories, but many did not have publicly available school enrollment data, or data were in a format that was not usable (eg, only state-level aggregate data rather than by school district). Because of data availability, we focused on the Mid-Atlantic and Northeast, where we were able to access data for 7 states and 1 district: Connecticut; Washington, DC; Delaware; Massachusetts; Maine; New Jersey; New York; and Rhode Island. We attempted to obtain data from other states, but data were not publicly available, and we did not get responses to data requests.

For our numerator (students with a disability experiencing homelessness), McKinney-Vento data were available by school district through the National Center for Homeless Education (NCHE) Web site (nche.ed.gov). These data are submitted to the Federal Department of Education by December, reporting on the previous school year (eg, data collected between September 2019 and July 2020 and reported in fall 2020). The McKinney-Vento liaison in each district identifies students experiencing homelessness on the basis of the federal definition of homelessness and reports the observation to later be sent to the Federal Department of Education. Liaisons identify students through outreach in the school and through partnering with other organizations that serve children and the unhoused.¹⁰  A full description of the identification process is available from the NCHE Homeless Liaison Toolkit (https://nche.ed.gov/wp-content/uploads/2020/09/Local-Liaison-Toolkit-2020.pdf). Data are reported by the liaisons and are reviewed and checked for quality by state coordinators of homeless education, then reviewed by EDFacts and the NCHE. If issues are found, corrections are made by the liaison, state coordinator, and data stewards, and the data are reviewed.¹⁵ 

The homelessness data included the cumulative number of students enrolled in each school district, the number experiencing homelessness in each district, the number experiencing homelessness who had disabilities, the number experiencing homelessness who learned or are learning English as a second language, the number experiencing homeless for each race, those who were unaccompanied minors, and the number of students in hotels or motels, shelters, or transitional housing, and doubled up. If a student experienced homelessness and then found stable housing within a school year, they are still counted in the homelessness data. All data were aggregated by district, so we could not determine how many students were in multiple categories (eg, experiencing homelessness, disabled, and English language learner). If there were <3 students in any category for a district, the data were suppressed to prevent identifiability (ie, there was no value for the data, but it was in the range of 0 to 3). Twenty-five percent of districts listed in enrollment data did not report homeless data (9.5% of all students lived in districts that did not submit homeless data). However, based on NCHE data guidance,¹⁶  missing school districts were considered to have 0 homeless students. Districts are required to report “non-zero” counts under the McKinney-Vento act, so no reported data are an indicator of no homeless students in the year. The US Department of Education works with districts and states to resolve any data irregularities and align district and state data.¹⁶  New York City geographic district data were not included in the NCHE Web site, so data were added from the Department of Education data Web site (eddataexpress.ed.gov).

For our denominator, enrollment and special education data for the 2019 to 2020 school year were available from the Department of Education Web site for Connecticut; Washington, DC; Delaware; Massachusetts; New York; and Rhode Island. The same sets of data were available on request from Maine. Like the homelessness data, IDEA data are compiled in the fall for the previous year (eg, 2019 to 2020 data are sent in fall 2020). These datasets were edited to isolate the pertinent information (year, state, county, school district, total enrollment, and special education enrollment) and merged by district with the homelessness dataset that had been edited to contain the disaggregated homeless counts for each school district.

We calculated the percentage of students with disabilities, the percentage of students experiencing homelessness among the disabled population, the percentage experiencing homelessness without a disability out of all enrolled students without a disability, the overall percentage experiencing homelessness, and the ratio of students experiencing homelessness with a disability to those without a disability. We calculated these estimates by school district for all states. For all ratio measures, we calculated standard errors and 95% confidence intervals. Because we had data from 2018 to 2019 for Massachusetts, we calculated the difference over time for the whole state and by county. Our institutional review board deemed this work not human subject research. We conducted a sensitivity analysis to explore the effect of suppressed data. We mapped data using ARC GIS. All data are presented in Supplemental Figure 2.

In 7 states and Washington, DC, 1 051 701 students were identified as having a disability out of 5 510 704 students enrolled in public schools for the 2019 to 2020 school year (Table 1). A total of 214 062 students experienced homelessness, as defined by the McKinney-Vento Act, of which 49 953 had an identified disability.

The percentage of students experiencing homelessness with and without disabilities is presented by state in Table 1. In all states, a greater percentage of students with disabilities experienced homelessness compared with students without disability. The highest percentage of students with disabilities experiencing homelessness was in Washington, DC, at 9.4% and the lowest was in Connecticut at 1.2%. Of note, in New York City geographic districts, 14.3% of students with disabilities and 15.0% of students without disabilities experienced homelessness. In our sensitivity analysis examining the effect of cell suppression, we found that, if all suppressed districts had 3 students experiencing homelessness with disability and 0 experiencing homelessness without disability (our most extreme assumption), we would have 909 more students with disabilities identified (1.8% larger than our original estimate), with biggest differences being in Maine (62% more cases) and no difference for Washington, DC, or Massachusetts. School district-level maps of the percentage of students with disability, the percentage of students with disability experiencing homelessness, and the ratio comparing homelessness in those with and without disability for each state, are presented in Supplemental Figure 2A–2G.

We found that students with disabilities were more likely to experience homelessness compared with students without disabilities across all 8 states and districts (risk ratio 1.58, 95% confidence interval 1.57–1.59). The risk ratios varied between states, but in all cases, students with disabilities were more likely to experience homelessness than students without disabilities (Fig 1). Compared with students without disabilities in the respective states, students with disabilities were 2.49 times as likely to experience homelessness in Rhode Island and 81% more likely to experience homelessness in Massachusetts.

Between the 2018 to 2019 and 2019 to 2020 school years in Massachusetts, there were 27 278 fewer students enrolled (Table 2). There was a slight increase of 0.5 percentage points (1% relative increase) in reported disabilities, as well as no change in percent of homelessness among students with disabilities. There was a 0.5-percentage-point decrease in homelessness for students without disabilities in the 2019 to 2020 school year compared with 2018 to 2019 (27% relative decrease). Although reports of homelessness decreased for both students with and without disabilities, those without disabilities had a greater decrease, corresponding to an increased risk ratio in the 2019 to 2020 school year compared with 2018 to 2019 (16.7% relative increase).

Throughout the Northeastern and Mid-Atlantic United States in the 2019 to 2020 school year, we found that students with a disability were more likely to experience homelessness than their abled peers. The school system is vital for this doubly marginalized population because high-quality education and appropriate support in the form of special education or connections to social services are health protective.¹⁷  Understanding the distribution of the student population at the intersection of homelessness and disability is critical for future policy and funding decisions, and our findings emphasize the need for comprehensive data collection on the state level.

We examined data collected during the initial phase of the COVID-19 pandemic, and the results need to be interpreted within that context. Our data encompass March to July 2020, a period in which a national eviction moratorium was in place.¹⁸  The pandemic’s effects on housing are likely larger in the 2020 to 2021 school year as compared with the 2019 to 2020 school year. In fact, the Massachusetts data reveal little change across the years. Nevertheless, it is important to note the impact that the COVID-19 pandemic has had on the educational experience of this population. Virtual education is a challenge to both students experiencing homelessness and poverty,¹⁹  in general, as well as students with disabilities whose special education programs may be more hands-on.²⁰  Other studies have revealed a decrease in the overall academic performance of students over the course of the past several years. There has also been a substantial increase in youth mental health diagnoses that could be classified as “emotional disturbances” and necessitate accommodations in school.²¹  Economic issues exacerbated by the pandemic have increased housing instability and led to increased rates of homelessness nationwide.^22,23  We hypothesize that, in the 2020 to 2021 school year, we will see an increase in students with disabilities experiencing homelessness because of the COVID-19 pandemic, and it will be crucial to continue to monitor these data and gain access to states without accessible district-level enrollment data.

Our results prompt us to examine the relationship between homelessness and disability. These data alone do not provide evidence for a causal or temporal relationship between the 2 outcomes, but several can be hypothesized. Caring for a child with a disability can be expensive in the United States and can create a financial burden,^24–26  which can lead to housing instability over time. Access to care for a child with a disability, or a suspected disability, depends greatly on the kind of health insurance a family has.²⁷  For example, an uninsured or underinsured family may find themselves unable to get their children to specialists for early intervention and therapies that can ameliorate symptoms of some developmental disabilities to the point of no longer meeting disability criteria. There are also environmental factors associated with low socioeconomic status, such as air and drinking water pollution, that have been linked to an increased risk of disability.^28,29  Longitudinal studies are needed to clarify and estimate the impact of potential mechanisms.

The NCHE and McKinney-Vento Act and IDEA administrators published a best-practices report on navigating the intersection of the McKinney-Vento Act and IDEA in 2008,³⁰  with an updated brief in 2022.³¹  There have been tensions in enforcing the 2 laws together because of a lack of specificity in the laws and communication issues between McKinney-Vento and IDEA administrators. The report listed 4 best practices to support students with disabilities experiencing homelessness: (1) create and promote policies and practices for regular ongoing communication among IDEA and McKinney-Vento staff, (2) review and revise state and local policies to address complex situations creatively, flexibly, and expeditiously, (3) work as a team to problem solve, and (4) repeat steps 1 and 2 annually.³⁰  With this insight, districts can work to remove barriers to service access, including absenteeism preventing disability evaluation, delays in services after a child moves into the district because of a lack of disability reports from the previous district, or transportation for students with disabilities that temporarily move out of district.

The overall socioeconomic composition of these states provides an important context for our findings. In Washington, DC, the jurisdiction with the highest percentage of students with a disability experiencing homelessness, the median annual income per household is $90 842. Connecticut, the state with the lowest percentage of students with disabilities experiencing homelessness, the median annual income per household is less than that of Washington, DC, at $79 855.³²  However, in Washington, DC, there is a well-documented disparity between its wealthiest and most disadvantaged residents. The percentage of residents in each state living below the poverty line better relates to our findings; 15% of Washington, DC, residents live in poverty in comparison with 9.7% of Connecticut residents.³²  The urban versus rural population of each state also appears to be reflected in these data because the homeless population tends to be higher in urban areas than in rural areas.

This work expanded on earlier research that focused on the Commonwealth of Massachusetts.¹³  We saw relatively little change in homelessness rates between years which may suggest that this population was stable over this time. Tracking the change in this population is a useful indicator of the efficacy of laws like IDEA and McKinney-Vento and can be used to monitor the impact of larger-scale events such as the COVID-19 pandemic and economic issues.

This study is confined to the Northeastern and Mid-Atlantic United States, which may limit its application to the rest of the country. Differences in reporting practices between states and districts should also be considered because states and districts may differ in their processes for collecting and reporting data. This creates the potential for misclassification. Housing status and disability may be differentially reported and identified, with those experiencing homelessness potentially being more likely to not be identified as such because of different interpretations of homelessness or the stigma that comes with sharing an unstable housing status. Children with certain disabilities may be less likely to be identified as having a disability if they are moving more often or if other factors prevent assessment. Although more work is needed to improve accuracy in reporting, our work is important in describing current services and disparities using current reporting practices. Results may not accurately represent the connection between homelessness and disability in all populations. The disaggregated data for the homeless population in each school district is not available and much of it would be suppressed because of smaller numbers of students, so we cannot speak to the impact of this issue on a particular population. Despite potential misclassification, we present data as collected from the educational system which represents how services are currently being allotted.

Education plays an important role in supporting positive long-term outcomes for vulnerable students. Students with disabilities who are experiencing homelessness are a particularly vulnerable population and quantifying this population for the first time will provide policymakers with valuable information to be able to act to better support these students.

COVID-19

coronavirus disease 2019

IDEA

Individuals with Disabilities in Education Act

McKinney-Vento

McKinney-Vento Homelessness in Education Act

NCHE

National Center for Homeless Education