AEFP 45th Annual Conference

Toward a Meaningful Impact through Research, Policy & Practice

March 19-21, 2020

Homeless Students and their Educational Outcomes: Evidence from Tennessee

Monica Hernandez, Vanderbilt University,

Homeless students are particularly at risk for poor educational outcomes because of the barriers they face to accessing schooling as well as achieving success once enrolled. Local residency requirements, the need for medical records, and inadequate transportation commonly impede school access. Further, socioemotional issues that result from the stigma of being homeless, challenges stemming from an unstable living situation, and the likelihood of inconsistent attendance are serious impediments to students’ academic success (Mawhinney-Rhoads and Stahler, 2006; Cowen, 2017).
This paper documents the prevalence of homelessness among public school students in Tennessee and measures the relationship between homelessness and school performance. We identify homeless students using a longitudinal, administrative database of students enrolled in Tennessee public schools from the 2016/17 through 2018/19 academic years. The data includes information on whether a student was ever identified as homeless during the academic year and the type of night residence upon identification.
In Tennessee, the average rate of homelessness amongst public school students in preschool through 12th grade is 1.7%. Rates of homelessness are 60% higher among Kindergarten students in relation to students in other grades (2.8%) and decrease as students reach 11th grade (1.1%). The homelessness rate increases again in 12th grade (1.4%), when some students lose stable housing after surpassing the mandatory age for school enrollment and become unaccompanied youth (Figure 1). This pattern of homelessness in Tennessee students contrasts with patterns in the rates of economic disadvantage (ED), as the differences in ED rates across grades are less drastic than homelessness rates (Figure 2). For instance, Kindergarten students are 4% more likely to be economically disadvantaged, but 35% more likely to be homeless than their 1st-grade peers. Similarly, 12th grade students are 6% less likely to be economically disadvantaged, but 25% more likely to be homeless than their 11th-grade peers. These differences do not seem to be driven by student characteristics, as demographic characteristics of homeless students are similar to those of economically disadvantaged students (Table 1).
Using school-year fixed effects models, we compare the educational outcomes of homeless students with those of their economically disadvantaged (ED) peers who live in poverty and were enrolled at the same time in their same school. With this methodology, we are able to control for some factors that affect students’ educational outcomes but are not related to housing instability. Such is the case of time-invariant school characteristics (e.g. school quality, location, and peers background) and chronic risks related to living in poverty.
Preliminary findings show that homeless students are absent from school at rates 38% higher than their ED peers, placing them at 60% higher risk of being chronically absent. Because of their high mobility rates, they are 2 times more likely to switch schools within an academic year and 3 times more likely to switch districts, as compared to their ED peers (Table 2). Their performance on Math and ELA standardized achievement tests is significantly lower than that of their ED peers, ranging from 0.15 to 0.20 standard deviations below. These differences are significant, considering that their ED peers already have low scores that are 0.3 standard deviations below the mean (Table 3).
We plan to continue this analysis to examine other academic outcomes of homeless students, such as their special education status, grade progression, discipline, and drop-out rates, and assess how their outcomes vary based on their grade. Examining outcomes by grade is important since unaccompanied homeless students enrolled in 12th grade might be at particular risk of struggling in school relative to homeless students in other grades.
This paper highlights the particular state of vulnerability of homeless students that places them at a very high risk of failing in school. Findings highlight the necessity of connecting local school systems to other systems that offer tailored social services for highly mobile populations. More research is needed to understand the contexts and settings conducive to making such connections successful.



This is a compelling and visually appealing display of information about students experiencing homelessness. Does Tennessee have any policies to ease transitions between schools and districts? Is there additional evidence you would need before making recommendations regarding that the transitions? I'd be interested to know more about what types of school rules are being violated, in an effort to better understand what kind of social and emotional support these students may require. Thank you for sharing your work! If you have any questions about my comments, you can reach me at

Interesting work! I have two questions. First, since identified homeless students typically receive services, do you try to differentiate relationships with being receiving services as opposed to homelessness? Second, have you looked at different outcomes by the length of homelessness? You can reach me at if you have comments or questions for me.

This is very interesting and compelling work. I'd be interested to know how accurate is the measure of homelessness and if/how it can be verified. Then given your findings, what are some school, district, and state policy levers that can be pushed on to address this issue? For individual teachers, what can they do and what support can be provided to help teachers to help their homeless students?

Interesting work y'all! I'm wondering about the potential for interactions or subpopulation analysis. For example, currently the poster looks at the ways economic disadvantage moderates the relationship between homelessness and the outcomes of interest. I wonder how other student characteristics might also moderate the relationship (e.g., race, gender). I find the overall results compelling but also wonder about the heterogeneity in the relationship.

I'm really excited to see how this work is developing. A couple of questions. How much within-student variation is there across years? Could you estimate models with student FE to identify the degree to which homelessness impacts outcomes for a given student over time? Another interesting direction would be to assess how homelessness affects outcomes in subsequent years, and especially whether homelessness among young children (where it appears to be more common) negatively affects their outcome trajectories even in subsequent years when they are not homeless.

Hi Olivia, Thank you for checking out our paper! We appreciate your engagement with this work! Best, Ela and Mónica

Hi Dominique, Thank you for checking out our paper and for your comments. We agree that examining heterogeneity would be a great next step. I'll send over a follow-up email to chat further! Thanks so much. Best, Ela and Mónica

Hi Tuan, Thank you for checking out our paper and for your comments. The indicator for homelessness is from the Tennessee Department of Education. It is a binary indicator for whether a student was ever identified as homeless during a given school year. There are a few ways a student can be identified as homeless, such as from enrollment paperwork or identification during the school year by school personnel. We agree that logical next steps of this analysis is identifying different policies that can help homeless students. We had a few ideas on how to do this. I'll send over a follow-up email to share further! Thanks so much. Best, Ela and Mónica

Hi Andrew, Thank you for checking out our paper and for your questions. We agree that it'd be a great next step to examine heterogeneity amongst homeless students who do and do not receive special education services. We also agree that looking at the length of time a student has been homeless would be an interesting next step. I will send a follow-up email to chat further! Thanks so much. Best, Ela and Mónica

Add new comment