Prior Evictions Among People Experiencing Homelessness in Delaware.

Eviction is frequently a precursor to homelessness. This is an exploratory study that looks at a group of homeless adults who stayed in Delaware homeless shelters in 2019 and the extent by which their homelessness is preceded by an eviction filing. Specifically, we match records of homeless shelter use with records from a court-based database of eviction filings, both in Delaware, to determine the frequency and correlates of prior eviction among adults staying in Delaware shelter and/or transitional housing facilities in 2019. Results show that 21 percent of the people in the study group had records of eviction filings in the 2-year period prior to initial homeless services use. Recent history of eviction filings was much more prevalent among study group members who were homeless with their children (i.e., with families), who were Black, and/or who were female. These findings are consistent with prior research and demonstrate the potential of interventions designed to mitigate eviction to also reduce homelessness, especially among families with children.

Introduction

Eviction and homelessness are related in that eviction precipitates displacement and homelessness frequently sustains it. These two phenomena combine to first remove and then deprive households (single individuals or families) of their rootedness in place, and the resulting disequilibrium contributes to a range of adversities, including an array of undesirable health outcomes and inequities.–

Households that face eviction often face an uphill battle to gather the resources required (e.g., first and last month’s rent, security deposits, utility deposits, moving expenses, etc.) to relocate into other housing. Furthermore, the mark of an eviction on their credit report makes it more difficult to find landlords that are willing to rent to them. Many evicted households will find temporary, makeshift accommodations with family or friends, or take up other precarious living situations. Some will, upon eviction, immediately become homeless and seek shelter and other homeless services. However, a longer trajectory from eviction to homeless shelter is more common, where households move to other living arrangements and these arrangements fall apart as combinations of economic and interpersonal strain intensify over time.

Here we provide an exploratory study that looks at a group of homeless adults who stayed in Delaware homeless shelters in 2019 and the extent by which their homelessness is preceded by an eviction filing. Specifically, we match records of homeless shelter use with records from a court-based database of eviction filings, both in Delaware, to determine the frequency and correlates of prior eviction (within two years of first using homeless services) among adults staying in shelter and/or transitional housing in 2019.

Background

Previous reports that included findings on the extent to which evictions have preceded homelessness have typically come from surveys of people experiencing homelessness. These studies have indicated that homeless households often had prior experiences of eviction and that such eviction experiences served as a primary catalyst for subsequent spells of homelessness. For example, separate studies reported that 11 percent each of surveyed homeless adults in Los Angeles and homeless families in San Francisco cited evictions (legal or extralegal) as a primary cause of their homelessness. Reports based upon surveys of homeless populations in other parts of the country found corresponding proportions as high as 30 percent in the Houston area and 45 percent across Massachusetts.

These findings are based largely upon the responses of individuals experiencing homelessness and carry distinct limitations. Respondents may have differing understandings of what constitutes an eviction, such as whether “eviction” is limited to formal, court-filed evictions or also includes other, less formal actions or threats from landlords., Problems with memory recall may distort the time between when an eviction occurred and how much time subsequently elapsed until the onset of homelessness. People may underreport their experiences with eviction due to the stigma related to being evicted, or may overreport them in the hope that this may render them eligible for housing or other services. Respondents may not make connections between eviction and homelessness when they initially find alternative housing to avoid homelessness–like moving in with family or friends–and then become homeless.

This study differs from the survey-based studies as it determines the extent of eviction experiences among a homeless population solely through matching administrative records. This standardizes the definition of eviction to instances when an eviction is formally filed in the court system. It also operationalizes homelessness to an instance where a household received shelter from a homeless services provider. The records provide much more specific time frames between the date of eviction filing and the subsequent date of shelter entry. However, despite the greater precision involved here, the evictions identified in this study do not include situations where a household is forced or coerced out of their rental arrangements by extra-legal means, and thus will provide conservative assessments of the connection between eviction and homelessness that do not include instances in which it is the threat of eviction, in the absence of a legal filing, that displaces people. What results is a different perspective from those provided by surveys, and an approach that is unique among the research that we are able to locate. Taken together, this study promises to broaden (and not replace) the understanding of how extensive eviction histories are among homeless populations, and identifies associations between individual characteristics and the relative likelihoods of having experienced an eviction filing prior to homelessness.

Delaware has high levels of both evictions and homelessness. From 2020 to 2022, the state’s homeless population has more than doubled. Prior to the COVID-19 pandemic, there were approximately 18,000 evictions filed annually in Delaware’s Justice of the Peace court system. Based upon this, Princeton University’s Eviction Lab reported an annual eviction filing rate of 16%, as compared to a national eviction filing rate of 2%. Low filing fees and plaintiffs’ ability to pursue evictions without obtaining legal representation are among the factors that account for high eviction filing rates in Delaware. By looking at the extent to which eviction filings preceded the onset of homelessness, this study provides a foundation for assessing whether and to what extent policy interventions to provide legal and financial assistance to mitigate evictions could potentially impact the numbers of people subsequently experiencing homelessness.

Data and Methods

The basis for this study is a merge of two administrative datasets. The first is the Community Management Information System (CMIS), a repository of data on homeless services provided in Delaware and the families and individuals who use these services. The Housing Alliance Delaware (HAD) manages the CMIS and has provided access to it for this study. The second database, maintained by Delaware’s Justice of the Peace (JP) court system, contains legal filings for evictions in Delaware. These eviction records are publicly accessible in a searchable online database.

For this study, we identify, using records in the CMIS, a study group of adults who have an initial record of receiving emergency shelter or transitional housing (i.e., homeless services) sometime in 2019. These records are matched with the JP Court database based on first and last name for cases when an eviction filing was initiated within a two-year period preceding the date of the initial record of homeless services use. Eviction records in which a name matches one from the CMIS study group, but which either have a longer gap between the filing date and the earliest homeless services date, or where the eviction filing date follows the earliest date of homeless services receipt, were not considered for this study.

Based on this match with the eviction filings, a subgroup of the study group with a matching eviction record is identified and the characteristics of the individuals in this subgroup is compared to the characteristics of the study group as a whole. Risk ratios and chi-square tests of difference are used to assess individual differences between those with and without prior eviction records, and a multivariate logistic regression model is fitted to systematically assess the associations between individual characteristics and having a recent (within two years of earliest homeless services receipt) record of an eviction filing.

This study was approved as exempt from full review by the Institutional Review Board of the University of Delaware.

Results

The CMIS database yielded records for 1,052 adults for whom the earliest record of a shelter or a transitional housing stay (i.e., index stay) occurred in 2019. Of these, 221 (21 percent) had a matching record of a JP Court eviction filing that occurred within a 2- year period prior to the index stay.

Table 1 shows the differential distributions of individual characteristics between the study group as a whole and the subgroup of 221 people with recorded eviction histories (i.e., eviction subgroup). Chi-square tests showed statistically significant differences (p< 0.05) for the eviction subgroup in the distributions by race, gender, household type and disability.

Table 1

Individual Characteristics for the Overall Study Group and the Subgroup with Recent Histories of Eviction Filing

	Adults in Study Group	Adults in Study Group with Previous Eviction Filing	Risk Ratio
Total	1052	221
Race (chisq=23.99; d.f.=3) p<0.001
Black	58%	74%	1.98
White (reference)	38%	25%
Other	2%	1%	0.61
Unknown	2%	0%

While 58 percent of the overall study group was comprised of Black race, that proportion rose to 74% among the eviction subgroup, yielding a 1.98 risk ratio (RR) when compared to those in the study group of White race. The study group was just over half female (51 percent), while that proportion rose to 62 percent among the eviction subgroup (RR=1.55). Only 27% of the study group were experiencing homelessness as part of a family (i.e., accompanied by minor children), while 43% of the adults in the eviction subgroup were part of a family (RR=2.06). Along with these significant bivariate differences in race, gender, and household composition, the difference in rates of identified disability is substantively similar (48 percent and 47 percent) but marginally significant (p=0.04) due to different levels of missing data.

Table 2 shows the extent to which the proportions of those with recent eviction histories vary based upon selected characteristics and then combinations of these characteristics. In this respect, the substantial differences between Black and White races (27 percent and 14 percent); men and women (25 percent and 16 percent); and family (with children) and individuals (34 percent and 16 percent) mirror the findings from Table 1. Combining these characteristics to construct even more precisely defined subgroups underscores how differences can layer upon each other, such as with gender and household type. However, such combinations only marginally raise the rates of recent eviction history among a particularly defined subgroup from the 34 percent rate shown among adults in families.

Table 2

Proportions of the Study Group with a Recent Eviction Filing Broken Down by Select Individual Characteristics

	Adults in Study Group	Adults with a Recent Eviction Filing	Proportion with Recent Eviction Filing
Total	1,052	221	21%

The logistic regression results on Table 3 demonstrate the robustness of the impacts associated with race, gender and household type on the odds of having a recent eviction history. In addition, the two older age groups are associated with significantly higher odds of having an eviction history than the youngest age group, a relationship that did not manifest itself in Table 1. Finally, adding interaction terms to the model between race, gender and household type measures did not contribute measurably to the interpretability of these results and are not included in the Table 3 results.

Table 3

Logistic Regression Results for Independent Impacts of Individual Characteristics Upon the Likelihood of having a Prior Eviction History

Covariate	Coefficient	Odds Ratio	Confidence Interval
Race (White as reference category)
Black***	0.77	2.12	0.4-1.140
Other	-1.07	0.343	-2.53-0.39

* p<0.05; ** p< 0.01; *** p<0.001

Discussion & Conclusion

A data match between eviction and homeless services records show that, among the 1,024 people whose initial Delaware homeless services use was recorded sometime in 2019, 21 percent were defendants in evictions filed in the Delaware JP Courts during the 2-year period prior to initial homeless services use. Such eviction filings were much more prevalent among study group members who were homeless with their children (i.e., with families), who were Black, and/or who were female.

In the eviction literature, these characteristics are commonly associated with higher risk for eviction. Desmond, in his study of eviction in Milwaukee, finds that “Black women disproportionately experienced the mark and the material hardship of eviction” (p. 112) and argues that “eviction is to women what incarceration is to men: a typical but severely consequential occurrence contributing to the reproduction of urban poverty” (p. 88).

Racial disparities in who becomes homeless are also well documented in the homeless literature. In contrast, there are no bodies of literature showing homelessness as disproportionately affecting women or families. Instead, there is a consensus that dynamics and experiences of homelessness are qualitatively different among women, who are most often homeless as single parents accompanied by children, and men, whose homelessness is experienced predominantly as a single adult. Among the homeless population, adults in families are more likely to have had spells in which they lived in leased housing and as such were more vulnerable to eviction, whereas unaccompanied adults, due to higher prevalence of disability, more institutional placements, and a greater flexibility with sleeping arrangements, are less likely to enter into leasing arrangements and face eviction.

While this study cannot conclusively show that eviction filings led to subsequent homelessness among the study group members, circumstances would suggest that the two phenomena are related and that intervening with housing assistance and other measures to mitigate evictions could also prevent subsequent homelessness among some proportion of those assisted. Given that one-third of the adults in families in this study received an eviction notice prior to their homelessness, tenant assistance measures that allow households to maintain stable housing represent a potential means to markedly reduce the number of families who become homeless, and thereby would prevent the onset of two experiences that are often devastating and traumatic, especially among children.

This study has limitations that must be taken into account when considering the results. First, matching criteria were limited to first and last name, creating a situation where matches are prone to both type 1 and type 2 errors, as different people may share common names, and spelling inconsistencies may preclude a person’s records from being matched across data systems. While there is no way to systematically validate matches, the datasets that were matched were relatively small in size, which reduces the incidence of separate people having identical names, and the sequencing guidelines (evictions preceding homelessness) provides some safeguard against type 1 error. The datasets were also small enough to permit manual review of the matches, the great majority of which involved names that appeared likely to be unique to one person.

A second limitation is that the CMIS dataset has large enough gaps in its coverage of homeless services to preclude having a comprehensive dataset of Delaware’s entire homeless population. This means that we were unable to use this data to determine how many people with eviction filings subsequently became homeless and that the study group we created does not include everyone receiving homeless services for the first time in 2019. Thus, the study group, while at 1,024 persons is of substantial size, has qualities of a convenience sample. More broadly, the generalizability of such a sample, from a small state, is limited, especially given the variation in eviction laws and policies across states.

Even with these limitations, the extent of evictions experienced in the study group is consistent with the range of findings in previous studies, and the individual characteristics associated with greater propensity to have an eviction history are consistent with both the homeless and eviction literatures. Given this, the findings of this study establish a link between homelessness and prior eviction, although it leaves much to be further explored. And finally, in Delaware, this study bolsters efforts to pass statewide right to counsel legislation for tenants facing eviction, as this study suggests that such legislation can have the effect of reducing homelessness, especially among families, women, and racial minorities.