Using Youth Risk Behavior Survey data to analyze housing instability among Delaware public school students

ABSTRACT

This study provides information about the prevalence, living conditions, and demographic characteristics of housing instability among students in the state of Delaware. Data were obtained from the Delaware Youth Risk Behavior Survey (YRBS) for the 2011–2017 academic years and included 23,819 youth respondents in grades 6–12. Descriptive and logistic regression analyses were conducted to characterize student demographics. Results show that 3.6% (about one out of 30) of Delaware students, grades 6–12, experience housing instability, and this rate was consistent across urban and rural areas. Findings also reveal that males and students of color were more likely to experience housing instability. Results obtained are utilized to understand the prevalence of housing instability and to inform schools, communities, and policy makers of ways to support and strengthen services and programming for students’ academic needs and well-being.

KEYWORDS:

• Housing instability
• McKinney-Vento
• public schools
• youth development
• youth risk Behavior survey (YRBS)

Introduction

In the 2015–2016 school year, national estimates indicated that approximately 1.35 million school-aged children and youth were experiencing housing instability¹ (Ingram et al. 2016; National Center for Homeless Education 2019). When children and youth experience housing instability, they may also experience difficulty in maintaining their student roles, as they are more likely to fall behind in their classes, be suspended from school, and receive supportive special education services than their stably housed peers (Institute for Children, Poverty, and Homelessness [ICPH] 2017; Murphy and Tobin 2011). According to the McKinney-Vento Homeless Assistance Act of 1987 Subtitle VII-B (henceforth referred to as McKinney-Vento), children and youth experiencing housing instability are required to receive equal access to a free and appropriate public education. However, despite the existence of this law, students experiencing housing instability are frequently under-identified (U.S. Government Accountability Office 2014). Students are denied their right to a free and appropriate public education as required by McKinney-Vento when schools fail to appropriately identify them (Brown, Duffield, and Owens 2018).

The purpose of this study is to fill gaps in our understanding about prevalence, living conditions, and demographic characteristics of students experiencing housing instability in public schools in Delaware through the examination of Youth Risk Behavior Survey (YRBS) data. Findings from this research will inform recommendations for researchers, schools, and policymakers to guide and strengthen the implementation of McKinney-Vento, with the goal of contributing to positive educational outcomes for students experiencing housing instability.

We begin with a description of the federal education policy created to address educational access issues arising from housing instability among students in K–12 schools (i.e., McKinney-Vento) and a discussion of housing instability’s impact on students’ educational experiences and trajectories. Next, we discuss the YRBS and its utility in identifying unstably housed students. Finally, research questions, analyses, and findings are presented, concluding this discussion with implications for researchers, schools, and policymakers.

McKinney-Vento

Housing instability is defined as the absence of a fixed, regular, and adequate nighttime residence (McKinney-Vento Homeless Assistance Act of 1987). Those experiencing housing instability include children and youth who: (1) share the housing of other persons (i.e., doubled up) due to loss of family housing and economic hardship; (2) live in motels, hotels, trailer parks, and camping grounds; and (3) reside in emergency and transitional shelters, are abandoned in hospitals, or stay in public/private spaces not typically used as housing for humans, such as cars, parks, abandoned buildings, and train stations (Every Student Succeeds Act of 2015). In an effort to ensure equal access to a free and appropriate education for students experiencing housing instability, Congress passed the McKinney-Vento Act in 1987. This federal law has since been modified under the Improving America’s Schools Act (1994), No Child Left Behind Act (2002), and Every Student Succeeds Act (2015). McKinney-Vento outlines the following federal directives in regard to providing support and eliminating barriers to an equal, free, and appropriate, public education:

(a) Each state educational agency shall ensure that each child of a homeless individual and each homeless youth has equal access to the same free, appropriate public education, including a public preschool education, as provided to other children and youths; (b) In any state where compulsory residency requirements or other requirements, in laws, regulations, practices, or policies, may act as a barrier to the identification of, or the enrollment, attendance, or success in school of, homeless children and youths, the State educational agency and local educational agencies in the state will review and undertake steps to revise such laws, regulations, practices, or policies to ensure that homeless children and youths are afforded the same free, appropriate public education as provided to other children and youths; (c) Homelessness is not sufficient reason to separate students from the mainstream school environment; (d) Homeless children and youths should have access to the education and other services that such children and youths need to ensure that such children and youths have an opportunity to meet the same challenging state academic standards to which all students are held (McKinney-Vento Homeless Assistance Act of 1987).

McKinney-Vento includes funding appropriation for a vast array of services such as shelter, food, and medical care (McKinney-Vento Homeless Assistance Act of 1987). Prior to McKinney-Vento, there were virtually no federal, state, or local standards within schools to ensure the educational rights and needs of children and youth experiencing housing instability. Therefore, children and youth were often denied entrance into schools because they did not have documentation such as Social Security cards, proof of permanent residence, and medical records (Ely 1987; Murphy and Tobin 2011; U.S. Government Accountability Office 2014). McKinney-Vento requires that districts appoint liaisons who are responsible for identifying students and then provide and/or connect students to services that support their ability to engage in their educational pursuits (McKinney-Vento Homeless Assistance Act of 1987). Despite this enacted federal regulation, McKinney-Vento has been consistently underfunded (National Alliance to End Homelessness 2019). Further, schools are often unaware of or ill-equipped to meet students’ needs (Ingram et al. 2016; Hallett and Skrla 2016). Moreover, there is a high mobility rate associated with homelessness, which has severe educational consequences. (Murphy and Tobin 2011; Kull et al. 2019). The educational disruption students encounter when faced with housing instability—coupled with the negative impact of high mobility and instability on their developmental trajectories—indicates a need to better understand the prevalence of housing instability among students in order to provide and enhance appropriate academic and social supports and services.

Educational impacts

Students experiencing housing instability are often subjected to unplanned mobility (e.g., eviction, moving abruptly due to shelter changes, conflicts with family/friends with whom they are doubled up) that negatively impacts their educational engagement, participation, and outcomes (Hallett and Skrla 2016). Lack of a stable home impacts the physical, psychological, and educational needs of these students (Murphy 2011; Rafferty 1999). Furthermore, unstably housed students score lower than their stably housed peers on achievement tests and are less likely to be promoted (ICPH 2017). When families are trying to find stable housing, they are continually on the move, and often, the children are not able to attend school regularly. Lack of a stable home has been found to result in frequent school absences and school transfers (Hallett and Skrla 2016; Miller and Bourgeois 2013). When experiencing housing instability, students often worry about their home situation, endure continuous disruption to school attendance, and feel socially isolated from their peers (Murphy and Tobin 2011). Additionally, transferring a student from one school to another has detrimental effects on the student’s education. For example, some research finds that frequent school transfers and mobility result in the loss of about three months of learning and increases students’ risk of graduating late or not at all (Mehana and Reynolds 2004; Rumberger and Larson 1998; Sparks 2016; Voight, Shinn, and Nation 2012).

The report Hidden in Plain Sight: Homeless Students in America’s Public Schools (Ingram et al. 2016) examines the growing problem of student homelessness by interviewing and surveying housing-unstable students and the state coordinators and local liaisons assigned to help them under the requirements of McKinney-Vento. The report found that:

• 82% of students surveyed say being homeless had a big impact on their life overall, 72% on their ability to feel safe and secure, 71% on their mental and emotional health, 62% on their physical health, and 69% on their self-confidence.

• 61% say they were never connected with any outside organization while homeless; 87% of those who were connected found the help valuable.

• 60% say it was hard to stay in school while they were homeless.

• 58% say their schools did only a fair or poor job or should have done more to help them stay in and succeed in school.

• When asked what’s very or fairly important, students identified the following: having someone to talk to or check in with for emotional support (86%), connecting with peers or maintaining friendships (86%), and participating in school activities, including sports, music, art, and clubs (82%) (Ingram et al. 2016).

The frequent instability that unstably housed children and youth endure makes it important for schools to serve as anchors of consistency and support in their lives, thus facilitating safety, constancy, and positive educational experiences and outcomes (Aviles de Bradley 2015; Hallett and Skrla 2016; Murphy and Tobin 2011).

The youth risk behavior survey (YRBS)

The purpose of the YRBS is to assess the prevalence of health-risk behaviors, identify health-risk behavior trends, and improve and inform programs and policies related to youth (Brener et al. 2013). The survey was developed in 1990 by the Centers for Disease Control and Prevention (CDC). The CDC conducts a national biennial survey and supports states, large urban regions, tribal organizations, and territories interested in administering the high school survey sampled from among their respective entities. The CDC has also developed a middle school survey but relatively few states elect to participate. States can customize the survey by supplementing core questions with optional questions, including questions on housing instability. When such questions have been included, the YRBS has demonstrated to be a useful tool for understanding the prevalence of housing instability. For example, Perlman et al. (2014), Brown, Duffield, and Owens (2018), and their respective co-authors found that the number of youths identified as housing unstable by the YRBS are notably higher than those identified under McKinney-Vento.

Furthermore, the YRBS provides valuable information about the specific living conditions of youth experiencing housing instability. For example, using the Delaware 2011, 2013, and 2015 YRBS for middle and high schoolers, researchers are able to distinguish between youth who are experiencing homelessness with or without their parents (also referred to as accompanied and unaccompanied youth) and whether they are living doubled up with another family, which is information not always accessible through record collection under McKinney-Vento. This information is valuable for schools and policymakers because students experiencing housing instability without parents often do not have support or guidance from a consistent, caring adult (Aviles de Bradley 2014, 2015). Therefore, students who are housing unstable and not living with their parents may need further support and assistance at school.

In addition to providing valuable prevalence and living conditions data, the YRBS also enables researchers to compare differences in housing instability based on demographic information, such as county, grade level, gender, and race/ethnicity. This information can help districts, cities, and states understand which youth are more likely to experience housing instability. While national research about youth who are more likely to experience housing instability has been conducted, the YRBS data are especially valuable because they provide information for a specific location in the country (i.e., based on county, city, and state), rather than research from other parts of the country that might be different from the location of interest. For example, research shows that Hispanic, non-White, and Black or African American youth are more likely to experience homelessness, and homelessness affects rural and urban areas at similar rates (Morton, Dworsky, and Samuels 2017). While these findings provide a description of youth homelessness, they are limited in their ability to provide recommendations at the district, city, and state levels. Having more context-specific information can better inform the local subtleties of McKinney-Vento implementation in supporting students experiencing housing instability.

Current study

The current study estimates the prevalence, living conditions, and demographic characteristics of housing instability among students in Delaware using YRBS data. This research provides a descriptive model for understanding who is more likely to experience housing instability among students and will discuss how results from the YRBS data can inform research, school-based practices, and policies. Guiding research questions are:

1. What is the prevalence of housing instability among youth in Delaware?

2. What are the demographic characteristics of students reporting housing instability in Delaware?

3. Are there differences among students experiencing housing instability by grade level, gender, race/ethnicity, county, and absence of parents?

4. Are there differences in the absence of parents among students experiencing housing instability by grade level, gender, race/ethnicity, and county?

Research context

At approximately 100 miles long and 30 miles wide, Delaware is the second smallest state in the U.S. It is divided into three counties: New Castle, Kent, and Sussex; New Castle county is predominantly urban and suburban and is more densely populated while Kent and Sussex counties are more rural in make-up. In 2016, Delaware had a total population of 952,065, with a median income of $61,017 (U.S. Census 2016). In Delaware, 24% of all people experiencing homelessness were children under the age of 18 (Housing Alliance Delaware 2017), and the number of students experiencing housing instability also increased from 3,847 during the 2014–15 school year (SY) to 4,047 during the 2015–16 SY (KIDS COUNT in Delaware 2016).

Despite the severe poverty and violence found in Delaware, specifically in the city of Wilmington, dubbed by some as “Murder Town USA” (Jones 2014; Linderman et al. 2017), there have been few studies focusing specifically on housing instability among school-aged children and youth in the state. Given the continued increase of housing instability in rural contexts (Harvard T. H. Chan School of Public Health 2019), a foundational description and analysis of housing unstable populations in predominantly rural and less densely populated areas are needed to better understand the scope and severity of its impact on this sub-population of students. This information will allow for increased awareness among researchers, schools, and policymakers in an effort to better respond to, and provide services and programming for, this often-overlooked population (Witz 2015). It is important to understand the characteristics of housing instability among smaller populations across the U.S., as most research has focused on large cities, such as Los Angeles, New York, and Chicago. Moreover, Delaware YRBS data is especially valuable as Delaware is one of a handful of states to conduct a middle school survey in addition to the high school survey and, therefore, among the few states able to analyze housing instability data among this younger student population.² Other studies that use the YRBS to examine student housing instability include only high school students, with no information known about middle school students’ housing instability. More importantly, research in this context could provide insights for developing solutions and best practices in serving the educational needs of students experiencing homelessness and housing instability.

Method

Data sources

In Delaware, the Center for Drug and Health Studies (CDHS) at the University of Delaware administers the high school YRBS survey on behalf of the state’s Department of Health and Social Services Division of Public Health. The YRBS is not a mandatory survey required by the Delaware Department of Education, but all public school districts and charter schools are invited to participate. School participation rates for YRBS over the 2011, 2013, 2015 and 2017 school years range from 64% up to 91%. Data for this project were obtained from the Delaware YRBS for the academic years 2011–2017. During that time frame, the Nemours Foundation also contracted with CDHS to conduct the Delaware Middle School YRBS, which is a distinct instrument from the high school survey. The instruments were developed with input from a multidisciplinary, statewide advisory group of youth advocates, representatives from CDHS, members of the Division of Public Health, and the CDC. The CDC core survey was supplemented with questions from the optional question bank and Delaware-developed questions. Both surveys were administered in the spring of odd numbered years to a random sample of classes within a census of public schools. Due to Delaware’s small size and population, researchers do not use a two-stage procedure to draw the YRBS sample. Rather, following guidelines from Westat Inc., national YRBS technical assistance advisors, researchers have drawn a sample of classrooms within a census of public schools throughout the state. The census of public schools is updated biannually by the state coordinator. All procedures for the conduct of these surveys were approved by the University of Delaware Institutional Review Board.

Participants

The data for the present study come from the 2011, 2013, 2015, and 2017 Delaware YRBS, which included 23,819 response sets from Delaware public school students in grades 6–12. Participation is voluntary and survey responses are anonymous, with experienced survey administrators from CDHS conducting the data collection to provide an added degree of confidentiality to students participating. Parental notice regarding the survey is provided one to two weeks in advance of survey administration, and parents are given the opportunity to opt out on behalf of their child. In addition, students are also given the opportunity to opt out at the time of administration. Research indicates that adolescents answer the YRBS accurately, and the survey has demonstrated adequate levels of test-retest reliability (Brener et al. 1995; 2013).

Measures

For this paper, analysis of the YRBS focused on students’ housing status, grade level, gender, race/ethnicity, county, and the absence of parents.

Housing status

In 2011, 2013, and 2015, participants were asked, “Where do you typically sleep at night?” The response set included the following: (1) At home with your parent(s) or guardian(s); (2) At a friend’s or relative’s home with your parent(s) or guardian(s); (3) At a friend’s or relative’s home without your parent(s) or guardian(s); (4) Somewhere else (such as a shelter, transitional housing, public place, hotel, car) with your parent(s) or guardian(s); (5) Somewhere else (such as a shelter, group home, foster care home, public place, car, hotel) without your parent(s) or guardian(s). In 2017, the question was modified and participants were asked “During the past 30 days, where did you usually sleep?” The response set included the following seven responses: (1) In my parent’s or guardian’s home; (2) In the home of a friend, family member, or other person because I had to leave my home or my parent or guardian cannot afford housing; (3) In a shelter or emergency housing; (4) In a motel or hotel; (5) In a car, park, campground, or other public place; (6) I do not have a usual place to sleep; and (7) Somewhere else.

To ensure comparability of the data given the two different versions of the housing question, respondents from the YRBS data sets were dichotomized into two categories: unstably housed students and stably housed students. Respondents who indicated they typically slept at home with parents/guardians were categorized as “housing stable”; those who indicated they slept elsewhere (e.g., at a friend’s/relative’s home, motel, or somewhere else) were categorized as “housing unstable.” This distinction was determined based on McKinney-Vento criteria.

Grade level

YRBS respondents were categorized into two grade levels: middle school or high school. Middle school students were given options to indicate whether they were in the traditional middle school grades (6th, 7^th, or 8^th) or in “Other Grade/Ungraded.” Similarly, high school students were given options to indicate a traditional high school grade (9^th, 10^th, 11^th, or 12^th) or in “Other Grade/Ungraded.” YRBS data provided information on this variable for all years of sampling. Due to the small percentage of students reporting housing instability, the analysis focused on whether a student was attending middle school or high school as opposed to individual grade levels.

Gender

YRBS respondents were asked their sex and were able to respond by indicating “male” or “female.” YRBS data provided information on this variable for all years of sampling.

Race/ethnicity

Participants were asked to indicate their race and ethnicity. The State of Delaware public school student population is 45% White, 30% African American, 17% Hispanic/Latino, and 8% other minority (Rodel Foundation of Delaware 2018). Based on these demographic characteristics, the ethnicity variable was collapsed into Black, White, Hispanic/Latino, and Other, where Other consists of American Indian, Asian, Hawaiian/Pacific Islander, and multi-racial. YRBS data provided information on this variable for all years of sampling.

County

Participants were asked to indicate in which county they reside. In Delaware, there are three counties: New Castle, Kent, and Sussex³, representative of urban and rural contexts. YRBS data provided information on this variable for all years of sampling.

Absence of parents

To determine whether students were without a parent at night, the YRBS variable that asked respondents where they typically slept at night was recoded. For students who were categorized as “housing unstable,” if they indicated they slept elsewhere without their parents, they were further categorized into “yes” for absence of parents, and if they indicated they slept elsewhere with their parents, they were categorized into “no” for absence of parents. This option was only available for YRBS sampling years 2011, 2013, and 2015. We could not dichotomize absence of parents for sampling year 2017 as the question and response options had changed, and all response options did not include language on the absence of parents.

Data analysis

Because housing instability is a relatively rare event (e.g., prevalence < 5%), large sample sizes are necessary to provide precise estimates of population prevalence. To improve the precision of our estimates, we pooled YRBS data across four administrations (i.e., 2011, 2013, 2015, and 2017) to produce a total sample size of 23,819 submitted surveys. While it is possible that a student may have taken a YRBS survey in more than one year, the likelihood of this is very small. Given that the YRBS includes a random sample of about 5,700 students each year, selected from a population of about 68,000 students, the sampling rate is about 8.4%. Furthermore, with each administration, new 6th and 7th graders enter the population, while 11th and 12th graders exit the population. Taken together, the probability that any one student would appear in the data set more than once is only 6%, and this probability applies to all subgroups in the data. Thus, the point estimates (i.e., percentages, means, and regression coefficients) based on pooled data will be unbiased, while our standard errors and p-values will be biased downward, just slightly, given that 94% of the sample satisfies the independence of observations assumption.

To test the research questions, frequencies and percentages were generated to characterize the demographics of students experiencing housing instability and students who were stably housed. Then, a series of logistic regression models were run because the outcomes of interest were dichotomous. Odds ratios were calculated to determine the magnitude of differences in significant associations between demographics and housing status, as well as predicting the absence of parents for students experiencing housing instability based on demographic characteristics. Descriptive statistics and regression models were estimated using the sample weights for YRBS provided by the CDC, with the original weights normalized with a sum equal to the study sample size. The normalized weights correct non-response rates and ensure that the final sample reflects student population demographics while still maintaining the original sample count of 23,819.

Results

Results are presented first by basic demographic characteristics of all students, followed by findings from logistic regression models that examine differences in grade level, gender, race/ethnicity, and county by housing status (i.e., housing unstable and housing stable). Finally, we present findings from the planned contrast to further probe significant differences in the absence of parents among students experiencing housing instability.

Demographic characteristics

A total of 23,819 students participated in the study and 838 (3.6%) were classified as housing unstable. (See Table 1 for additional information about type of housing instability experienced by the unstably housed student sample). The unstably housed student sample is comprised of 40.5% middle school students and 59.5% high school students. The percentages of males and females were 55.6% and 39.7%, respectively, and about half (54.1%) of respondents were from New Castle County. Respondents were 30.6% White, 31.8% Black, 19.8% Hispanic/Latino, and 13.1% Asian, American Indian, Hawaiian/Pacific Islander, and multi-racial. See Table 2 for the demographic characteristics.

Table 1. Type of housing instability among students by grade level.

	Middle school		High school		Total
	n	%	n	%	n
2011, 2013, and 2015
At a friend’s or relative’s home with your parent(s) or guardian(s)	112	13.3	66	7.9	178
At a friend’s or relative’s home without your parent(s) or guardian(s)	118	14.1	111	13.1	229
Somewhere else (such as a shelter, transitional housing, public place, hotel, car) with your parent(s) or guardian(s)	51	6.1	28	3.3	79
Somewhere else (such as a shelter, group home, foster care home, public place, car, hotel) without your parents or guardians	61	7.3	71	8.5	132
2017
In the home of a friend, family member, or other person because I had to leave my home or my parent or guardian cannot afford housing	33	3.9	51	6.1	84
In a shelter or emergency housing	7	0.9	7	0.9	14
In a motel or hotel	5	0.6	6	0.7	11
In a car, park, campground, or other public place	2	0.2	19	2.3	21
I do not have a usual place to sleep	31	3.7	12	1.4	43
Somewhere else	25	3.0	22	2.6	47

Table 2. Demographic characteristics of Delaware students by housing status.

Variable	Housing unstable (n = 838; 3.6%)		Housing stable (n = 22,385; 96.4%)		Total
	n	%	n	%	n
Grade level
Middle school	339	40.5	9,408	42.0	9,747
High school	499	59.5	12,976	58.0	13,475
Missing	–	–	–	–	–
Gender
Male	466	55.6	11,045	49.3	11,511
Female	333	39.7	10,955	48.9	11,288
Missing	39	4.7	385	1.8	424
Race/ethnicity
White	257	30.6	10,208	45.6	10,465
Black	267	31.8	6,027	26.9	6,294
Hispanic/Latino	166	19.8	3,561	15.9	3,727
Other^a	110	13.1	2,040	9.1	2,150
Missing	38	4.7	549	2.4	587
County
New Castle	454	54.1	12,932	57.7	13,386
Kent	212	25.3	4,977	22.2	5,189
Sussex	171	20.4	4,442	19.8	4,613
Missing	1	0.2	34	0.3	35

Note. ^aOther = Asian, American Indian, Hawaiian/Pacific Islander, and multi-racial; “–” denotes not applicable.

Item non-response was 2.4% or less, and survey refusal was 1.8% across all four years, therefore potential bias due to missing data is negligible (Allison 2002).

Logistic regression

See Table 3 for logistic regression models. Odds ratio calculations indicated that males are 1.38 times more likely to experience housing instability than females (p < .001). Furthermore, compared to white students: Black students are 1.86 times more likely (p < .001), Hispanic and Latino students are 1.89 times more likely (p < .001), and students of other races (Asian, American Indian, Hawaiian/Pacific Islander, and multi-racial) are 2.15 times as likely (p < .001) to experience housing instability. In addition, compared to New Castle County (i.e., an urban county in Delaware), students from Sussex County (i.e., a rural county in Delaware; OR = 1.22, p = .035) and Kent County (i.e., a mixed urban and rural county in Delaware; OR = 1.28, p = .005) were significantly more likely to experience housing instability. Finally, odds ratios demonstrated that students in high school are not more likely than middle school students to experience housing instability (OR = 1.09, p = .236).

Table 3. Logistic regression of housing status on demographic characteristics.

				95% CI
Demographic characteristics	B	SE	OR	Lower	Upper	p-value
Male	.33	.07	1.38	1.20	1.60	.001**
High school	.09	.08	1.09	.94	1.26	.236
Race/ethnicity
Black	.62	.09	1.86	1.55	2.22	.001**
Hispanic/Latino	.64	.10	1.89	1.54	2.32	.001**
Other^a	.77	.12	2.15	1.71	2.72	.001**
County
Kent	.25	.09	1.28	1.08	1.52	.005*
Sussex	.20	.10	1.22	1.01	1.47	.035*

Note. ^aOther = Asian, American Indian, Hawaiian/Pacific Islander, and multi-racial; SE = Standard error; OR = Odds ratio; CI = Confidence interval; *p < .05; **p < .001.

Absence of parents

Logistic regression

Additionally, the data were narrowed to look specifically at students experiencing housing instability to examine the absence of parents, given that students experiencing housing instability without parents often do not have support or guidance from a consistent, caring adult. Therefore, students who are housing unstable without parents may need further support or assistance at the school level, motivating our investigation of differences based on demographic characteristics. Among students who are experiencing housing instability, 63.5% (n = 391) report the absence of parents (see Table 4). We conducted a logistic regression to further probe differences in the absence of parents across grade level, gender, race/ethnicity, and county for students experiencing housing instability. The results showed that none of the variables predicted the probability of absence of parents (see Table 5).

Table 4. Absence of parents among students with housing instability.

Variable	Housing unstable with Parents (n = 225; 36.5%)		Housing unstable without Parents (n = 391; 63.5%)		Total
	n	%	n	%	n
Grade level
Middle school	103	40.1	154	59.9	257
High school	122	34.0	237	66.0	359
Missing	–	–	–	–	–
Gender
Male	119	52.9	207	52.9	326
Female	93	41.3	116	29.7	209
Missing	13	5.8	68	17.4	81
Race/ethnicity
White	80	35.5	119	30.4	199
Black	63	28.0	120	30.7	183
Hispanic/Latino	37	16.4	82	21.0	119
Other	37	16.4	52	13.3	89
Missing	8	3.7	18	4.6	26
County
New Castle	120	53.3	208	53.2	328
Kent	55	24.4	101	25.8	156
Sussex	50	22.3	82	21.0	132
Missing	–	–	–	–	–

Note. Absence of parents could not be identified for 222 students experiencing housing instability. Responses from 2017 were not included due to instrument change that did not delineate the presence of parents.

Table 5. Logistic regression of absence of parents on demographic characteristics for students experiencing housing instability.

				95% CI
	B	SE	OR	Lower	Upper	p-value
High school	.29	.18	1.33	.94	1.89	.104
Male	–.07	.18	.93	.66	1.32	.699
Race/ethnicity
Black	.24	.22	1.27	.83	1.95	.278
Hispanic/Latino	.35	.25	1.42	.86	2.32	.170
Other	–.06	.27	1.11	.56	1.59	.820
County
Kent	.10	.22	1.11	.72	1.69	.642
Sussex	–.03	.23	.97	.63	1.51	.898

Note. SE = Standard error; OR = Odds ratio; CI = Confidence interval.

Discussion

YRBS data in this study provided prevalence, living conditions, and demographic characteristic information among middle and high school students experiencing housing instability in Delaware. Results from this study indicate that (a) housing instability was 3.6% overall, or about one out of every 30 students; (b) students of color were more likely to experience housing instability than their White counterparts; specifically Black students in Delaware were 1.86 times more likely, Hispanic/Latino students 1.89 times, and Other races 2.15 times; (c) male students were 1.38 times more likely to experience housing instability; and (d) students in Sussex County and Kent County were more likely to experience housing instability compared to students in New Castle County by 1.22 and 1.28 times, respectively; (e) there were no significant differences in housing instability when comparing middle and high school students; and (f) there were no differences in the absence of parents based on demographic characteristics among students experiencing housing instability.

Prevalence of housing instability

The results demonstrated that 3.6% of youth experienced housing instability. This translates to approximately one out of every 30 students—suggesting that many or perhaps most middle and high school classrooms include at least one student experiencing housing instability. Additionally, as supported by previous research (Brown, Duffield, and Owens 2018; Perlman et al. 2014), YRBS data identified a higher percentage of students experiencing homelessness (3.6%) as compared to students identified through McKinney-Vento-based methods (2.3%) in Delaware (U.S. Interagency Council on Homelessness n.d.). Nationally, high school YRBS data illustrates that “4.9% of students surveyed in the 17 states experienced homelessness at some point during the 2016–2017 school year, while public schools reported only 2.57% of their students as experiencing homelessness” (SchoolHouse Connection 2020a). According to SchoolHouse Connection (2020a), the significant under-identification indicated when comparing the McKinney-Vento required data to the YRBS data means that as many as one million students throughout the U.S. experiencing homelessness are not receiving the services that are their right under federal law. As indicated by these results, about one-third of students⁴ in Delaware who are housing unstable are not on their school’s radar to receive services through McKinney-Vento. While this federal policy has been in place since 1987, one potential reason why students are under-identified is that many educators, parents, and students are not aware that it exists (Miller 2011; Aviles de Bradley 2014; Ingram et al. 2016). The under-identification of students in schools has implications for in-service teacher awareness regarding prevalence of housing instability among middle and high-school students; it can also be inferred that teacher preparation programs (in addition to schools and school personnel) must be diligent in their efforts to increase awareness and accountability. As noted earlier in the paper, findings from Hidden in Plain Sight: Homeless Students in America’s Public Schools (Ingram et al. 2016) indicate that students would benefit from in-service and pre-service teachers who are aware of the federal requirement McKinney-Vento places on schools to meet the educational, emotional, and basic needs of this sub-population of highly mobile students.

For example, one way to increase awareness is by requiring teacher preparation programs to include McKinney-Vento information in their program curriculum (Heybach and Aviles de Bradley 2014), similar to the way in which pre-service teachers must be informed about their responsibility to students with disabilities under the Individuals with Disabilities Education Act (2004). In Delaware, this could be accomplished by collaborating with universities that offer teacher preparation programs. Researchers could partner with programs to provide workshops, seminars, and/or lectures about McKinney-Vento. The information could focus on the state overall and then specifically on the areas in which student teachers will be placed. This would ensure that all student teachers in Delaware know this information, which, in turn, may potentially increase the number of students identified and who subsequently receive services through the Act. This model could be expanded and implemented nationally. Historically, Delaware was one of nine states to collect information on unstably housed students in 2015 and one of 17 states in 2017 (SchoolHouse Connection 2019). Recently, the CDC announced that the YRBS will include a standard question on homelessness in its 2021 high school questionnaire. However, the question designed to identify unaccompanied youth remains optional (SchoolHouse Connection 2020b). Including a question about homelessness as standard on the high school questionnaire should allow for a more thorough examination of high school YRBS data for unstably housed youth across the nation. Moreover, as previously noted, Delaware continues to be one of the few states to collect housing status data for middle school students; other states are strongly encouraged to include housing status questions (inclusive of accompanied or unaccompanied status) for middle schoolers to more accurately identify and provide services as required through McKinney-Vento.

Demographic characteristics of unstably housed students

Students of color

Prior research indicates that Black and Brown students comprise a significant portion of students experiencing housing instability (Aviles de Bradley 2014; Aviles and Heybach 2017; Morton, Dworsky, and Samuels 2017; Kull et al. 2019). Black and Brown students are consistently disproportionately represented among unstably housed students (Morton, Dworsky, and Samuels 2017). Findings from our research demonstrated that Black students in Delaware are 1.86 times more likely than White students to be housing unstable, which is considerably lower than the respective national rate of 2.67 times (SchoolHouse Connection 2019).

The YRBS data also capture self-report of Hispanic/Latino/a students who are living doubled up and/or moving frequently (often staying with friends and relatives), common forms of housing instability among this student population (Conroy and Heer 2003). Historically, Hispanic/Latino/a students and families are undercounted among individuals experiencing homelessness, referred to in the literature as the “hidden Hispanic homeless” (Conroy and Heer 2003). Our results found that Hispanic/Latino/a students in Delaware were 1.89 times more likely to experience housing instability than their White peers, which is higher than in the national YRBS data, where Hispanic/Latino/a students are 1.68 times more likely (SchoolHouse Connection 2019). However, it is possible that Hispanic/Latino/a students in Delaware, similar to Latino families nationwide, may still have been undercounted due to fear of increased scrutiny by government agencies because of the increase in discriminatory policies and practices targeting Latino families. These findings may also have implications for counting undocumented Latino/a students experiencing housing instability (Esquivel 2007). These racial/ethnic representations should be further explored and analyzed to develop and provide culturally responsive and relevant services to students and families experiencing housing instability across various racial groups in Delaware.

Males

The U.S. Interagency Council on Homelessness (2018) reported that the majority of unaccompanied youth experiencing homelessness are young men or boys. Further, approximately 66% of unsheltered unaccompanied youth and 55% of sheltered unaccompanied youth are male. Our data supports previous research that demonstrates the overrepresentation of males experiencing housing instability, as male students in this study were 1.38 times more likely to experience housing instability than females. Schools should be aware of these differences and work to identify and provide male students with the necessary support for their academic success and well-being, beginning with the requirements outlined under McKinney-Vento. This finding should not diminish the need to identify female students; rather, it is an indicator of the gender dynamics at play within this particular student population.

High school students

Our findings reveal no significant difference in rates of housing instability among middle and high school students. This finding is important as most states do not collect the housing status of middle school students. These data suggest services and programming for unstably housed students are critical throughout the trajectory of a student’s academic career. Research recognizes that housing instability influences high school completion (Aratani and Cooper 2015), and youth who do not obtain a high school diploma or GED are at a 346% higher risk of experiencing housing instability as adults (Morton, Dworsky, and Samuels 2017), creating a negative cyclical effect on students experiencing housing instability and their future housing outcomes and opportunities. Therefore, educators, service providers, and other significant adults must be persistent in identifying students in order to provide services and supports that increase successful matriculation at both the middle and high school levels.

County

Data show that students in the urban/rural and rural counties are more likely to be housing unstable than students in the urban county of Delaware. This finding highlights the fact that housing instability is not a phenomenon relegated to urban areas in Delaware. Most research on youth experiencing housing instability focuses on urban “street youth,” and there has been limited research about youth in suburban or rural areas (e.g., a majority of studies conducted have been in Chicago, Los Angeles, and New York City; Hallett 2012; Aviles de Bradley 2015; ICPH 2017). More recently, research has begun to focus on unstably housed youth in rural areas indicating that youth homelessness affects urban and rural areas at similar rates (Morton, Dworsky, and Samuels 2017). However, this study expands on the research about urban and rural homelessness to demonstrate that, in Delaware, students are more likely to experience housing instability in rural counties. One potential reason for this finding is the wording of the YRBS housing instability question, which may more effectively capture youth living in rural areas than previous studies by offering the response choice: “(2) In the home of a friend, family member, or other person because I had to leave my home or my parent or guardian cannot afford housing.” Morton and colleagues (2017) highlight that rural youth rely more on couch surfing than shelter systems, and therefore, based on the invisible nature of their housing instability, may be under-identified by McKinney-Vento-based methods.

Further investigation is needed to understand why rural rather than urban housing instability is more likely in Delaware and how well McKinney-Vento is serving these students. This information can help schools understand the resources and barriers within their respective communities and areas so they may adequately meet the needs of students across urban and rural regions. For example, the Delaware Department of Education (DDOE) should be informed about this finding to ensure that the data is disseminated to districts, schools, and McKinney-Vento liaisons. The DDOE should examine the ways in which McKinney-Vento funding is distributed across the state to ensure that adequate funds are being provided to rural schools for services, such as transportation, which could cost more in rural than urban areas. In addition, research should be conducted in rural and urban Delaware with students experiencing housing instability to understand the unique challenges occurring in these locations that may affect the rate of housing instability.

Parental absence among unstably housed students

Finally, the results indicate no difference of parental absence based on demographic characteristics for students experiencing housing instability. These findings both support and challenge previous research. Prior studies have demonstrated that youth are equally likely to experience housing instability in rural and urban areas (Morton, Dworsky, and Samuels 2017). In addition, Morton and colleagues (2017) found that one in 30 youth ages 13–17 were unaccompanied, which demonstrates that there are consistent rates of homelessness across high school and a portion of middle school. Therefore, our findings extend these results by including all middle-school-aged students.

However, contrary to our findings, previous research has demonstrated that Black and non-White Hispanic unaccompanied youth are more likely to experience housing instability compared to their White peers (Morton, Dworsky, and Samuels 2017). Furthermore, data collected by the U.S. Department of Housing and Urban Development indicates that males are more likely to be unaccompanied than females (U.S. Interagency Council on Homelessness 2018). While it is unclear why our study’s results both support and challenge previous research, one potential reason for the difference in findings is that some of these unaccompanied youths might have relocated to larger cities outside of Delaware, such as Baltimore and Philadelphia, due to the limited services available in Delaware for unaccompanied youth, and thus they might not be represented in our sample.

Finally, YRBS data from 2011, 2013, and 2015⁵ demonstrate that 14.1% and 13.1% of middle school and high schoolers, respectively, are unstably housed at a friend’s or relative’s home without their parent or guardian; 7.3% of middle schoolers, and 8.5% of high schoolers are residing “somewhere else (such as a shelter, group home, foster care home, public place, car, hotel) without your parents or guardians.” (Centers for Disease Control 2015, 3). As youth mature into adulthood, a number of physiological, psychological, and cognitive changes occur (Walker, Gambone, and Walker 2011), and supportive adults can productively contribute to youths’ positive developmental trajectories (Reeg 2003). As such, the prevalence of students who are unstably housed without a parent or guardian present is an important dynamic to consider. Caring adults in the lives of youth experiencing housing instability can have a significant positive impact on their academic and emotional well-being (Hart-Shegos 1999; Ingram et al. 2016). Schools play an integral role in fulfilling this need by providing services and/or referrals to services, such as tutoring and mentoring, that support the developmental and academic needs of students experiencing housing instability who are without consistent adult contact and care.

Limitations

A main limitation of this study is that the YRBS data were collected during the school day and therefore did not account for students who were absent from school or youth who stopped attending school altogether. The YRBS captures information about the students enrolled in school who are in attendance, but these data are not representative of school-aged youth who have dropped and/or been pushed out⁶ (Fine 1991; Morris 2015) of school. As noted previously, unstably housed students endure continuous disruption in school attendance (Murphy and Tobin 2011). Future research that captures McKinney-Vento implementation in schools, including counting students who have left school, could greatly inform policies, practices, and services for re-engaging youth experiencing housing instability who are not enrolled in school.

Another limitation of this study is that among the data instruments only the 2017 Delaware YRBS asked a question that could be used to identify housing instability among transgender students. As a result, we were unable to analyze the data due to the small sample size. Nonetheless, it is important to note that LGBTQ youth are 120% more likely to report homelessness than cisgender or heterosexual youth (U.S. Interagency Council on Youth Homelessness 2018). Therefore, states should continue including housing instability risk factors (e.g., identifying as LGBTQ) on the YRBS. With the continued use of gender and sexuality identity items, we underscore that future analysis could provide a better understanding of living conditions and demographics of students who are especially vulnerable to housing instability.

Finally, the YRBS data analyzed is restricted in its ability to inform our understanding of McKinney-Vento implementation at the school level. The data presented inform our understanding of the scope and severity of housing instability among students in Delaware. McKinney-Vento requires that states receiving funds under McKinney-Vento assure that each student experiencing housing instability shall have access to a comparable free, appropriate public education in the mainstream school environment, including transportation services, gifted and handicapped educational services, school meal programs, vocational education, bilingual programs, and before and after school programs (U.S. Department of Education 2002). The data identified here provide a general understanding of housing instability among Delaware students, which is a necessary first step to providing them with the services they are entitled to receive through McKinney-Vento.

Implications

As indicated by the results, the YRBS can provide valuable information to districts, cities, counties, and states regarding the prevalence of housing instability and the living conditions and demographics of students experiencing housing instability. This is particularly significant given the dearth of research completed in less populated states such as Delaware. The prevalence of housing instability among K–12 students warrants creating school systems and structures that best support and meet their academic, emotional, and well-being needs. YRBS data have the ability to inform and strengthen advocacy efforts to increase resources for unstably housed students. For example, McKinney-Vento requires training for teachers and other school personnel (Pavlakis and Duffield 2017) to ensure that the needs of students are met. However, there is no requirement currently in place for pre-service teachers and school personnel (such as counselors and coaches) to learn about McKinney-Vento (Heybach and Aviles de Bradley 2014). Given this lack of awareness, educating pre-service teachers about McKinney-Vento and providing them with demographic characteristics of students experiencing housing instability in their future school districts could improve their knowledge of the law and increase the number of students identified and eligible for services.

In addition to providing pre-service teachers with legal and demographic information, educators should be informed about the impacts of housing instability on students’ school participation and academic outcomes, as well as tools and resources that can be employed to better support their educational, emotional, psychological, and other basic needs (Murphy and Tobin 2011; Kull et al. 2019). This is especially significant for students not in the care of a parent or guardian. In light of the developmental shifts occurring for middle- and high school = aged students (Scales 2010; Simmons and Blyth 2008), it is critical for their emotional and cognitive well-being to receive support, have access to consistent adults in the school building, and/or be given assistance navigating school systems, thus ensuring that their educational rights are not violated. These developmental shifts illuminate a finding by Uretsky and Stone (2016) that adolescent-aged students should be considered a unique subpopulation of housing unstable students, and intervention strategies should be adjusted accordingly. These strategies would be critical to ensure that students remain enrolled in school, as youth who lack a high school diploma or GED are 4.5 times more likely to experience housing instability (Morton, Dworsky, and Samuels 2017). It is essential that schools create the necessary conditions to facilitate positive youth development leading to improved academic outcomes and youth trajectories toward becoming healthy, stable adults.

As noted previously, a significant purpose of YRBS is to improve and inform programs and policies related to youth (Brener et al. 2013); McKinney-Vento policy awareness and implementation can be bolstered by YRBS administration and evaluation. McKinney-Vento continues to be limited in its effectiveness due to lack of awareness among teachers, school personnel, parents, and students. If states continue to adopt the optional questions regarding housing status on the YRBS, in tandem, McKinney-Vento too can expand its impact and effectiveness for unstably housed students. Moreover, research should further investigate other pertinent factors shaping and influencing educational experiences and outcomes for students identified as housing unstable. For example, examining YRBS data constructs related to mental health, coping behaviors, interface with law enforcement/juvenile justice system, violence, and adult supports (parent, teacher, etc.) would provide insights regarding risk and other behavioral/health disparities associated with this population of students; such analyses should include similarities and differences among accompanied and unaccompanied students. Future research that compares the educational and mental health outcomes of unstably housed children and youth to their stably housed peers—particularly accounting for differences of race, gender, gender identity, age/grade, and parental presence—is necessary to inform our understanding of within-group needs. This understanding can guide schools and social service programs to ways in which they can further create equitable structures, systems, and services that facilitate positive developmental outcomes and educational trajectories for middle and high school students experiencing housing instability.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes on contributors

Ann M. Aviles, PhD, is an Associate Professor of Human Development and Family Sciences in the College of Education and Human Development at the University of Delaware. Dr. Aviles’s research engagements include examining policies, services, and programs that impact the educational opportunities, material realities, and mental health of youth of color experiencing homelessness/instability; Latina/o education; education policy; education equity; critical/justice-based teacher preparation; school-community partnerships; participatory action research; Critical Race Theory (CRT); Latina/o Critical Theory (LatCrit); and positive youth/community development.

Kathleen McCallops, MS, is a doctoral student in the Department of Human Development and Family Sciences at the University of Delaware. Ms. McCallops’s research focuses on the educational opportunities and mental health of youth and families experiencing housing instability and food insecurity.

Maryam Hussain, PhD, is currently a postdoctoral fellow in health psychology at the University of California, Merced. Dr. Hussain’s research broadly focuses on the psychosocial determinants of health in underserved populations; specifically, she looks at how ongoing and daily experiences of cultural resiliency and adversity predict cardiometabolic disease risk over time.

James Highberger, MPA, is a research associate at the Center for Drug and Health Studies and the acting data manager for the Youth Risk Behavior Surveys, Delaware School Surveys, and Youth Tobacco Surveys. He received his Master's in Public Administration from the University of Colorado and is currently working toward his PhD in Criminology. He focuses his work on adolescent substance use and prevention and also has a passion for geographic information systems, mapping out how resources and substance use rates vary throughout the state of Delaware to better direct prevention efforts.

Rachel Ryding, MA, is a PhD student in the Department of Sociology and Criminal Justice at the University of Delaware, where she works with the Center for Drug and Health Studies on the Delaware School Survey projects. Her research interests are focused on substance use, mental health, and research methods.

Sharon Merriman-Nai, MC, is a researcher, evaluator, and project director at the Center for Drug and Health Studies at the University of Delaware. From 2015 to 2019, she led the Center's school-based survey efforts, including the Delaware Youth Risk Behavior Survey, and conducts evaluation for Center projects designed to promote prevention of suicide, substance use, and other mental health challenges. Previously, Ms. Merriman-Nai served as a community mental health counselor and consultant, working with adults and children who had experienced childhood trauma and adversities.

Henry May, PhD, is Director of the Center for Research in Education and Social Policy (CRESP) and an associate professor in the School of Education at the University of Delaware. Dr. May specializes in the application of modern statistical methods and mixed-methods in randomized experiments and quasi-experiments studying the implementation and impacts of educational and social interventions and policies.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Notes

1 In this paper, the authors will use the term housing instability whenever possible to provide a broader understanding of temporary housing and to expand upon the traditional perspective regarding homelessness (often used in relation to persons who reside in a shelter, abandoned building, car, etc.) by including temporary housing situations such as staying with friends or relatives, couch surfing, or moving frequently.

2 National Consortium on Homelessness and the YRBS Meeting, personal communication, September 13, 2019.

3 Our definition of New Castle County (NCC) as urban is based on the large portion of urbanized area in NCC, according to the map and definition provided by the U.S. Census Bureau. The density portion of the census definition of “Urban” requires a population density of 1,000 people per square mile. New Castle meets that standard with a density of 1,555 per square mile. Kent County is described as urban/rural because it has only one major metropolitan area (Dover), and the rest is largely rural. Sussex is defined as rural because according to the map, the majority of its areas are considered rural by the Census Bureau, and according to census.gov it has a population density of only 210 people per square mile.

4 The authors performed a logistic regression to test for differences in rates of homelessness across years in the sample. The results showed there were no differences in the rates of homelessness by year (X2(3) = .672, p = .880).

5 Data from 2017 could not be included here due to changes in the item response options that year. Specifically, the question included in the 2017 survey did not allow for a distinction between parent presence or absence.

6 Pushout refers to practices that contribute to students leaving school. These include unwelcoming and uncaring school environments and overreliance on zero tolerance school policies.