Get to Know the New Eviction Lab Data: A Summary of What’s Inside

Get to Know the New Eviction Lab Data: A Summary of What’s Inside
Matthew Desmond, Ashley Gromis, Lavar Edmonds, James Hendrickson, Katie Krywokulski, Lillian Leung, Adam Porton
Publication Date:
Find Full Text

The Eviction Lab at Princeton University released a new publicly available dataset with aggregated counts and rates of eviction filings and judgments across the United States. The researchers sought landlord/tenant records from courts in each state and other data providers, removed irrelevant cases, and aggregated them into counts by state, county, and other geographic levels.

The 13 states that provided court data are Alabama, Connecticut, Hawaii, Iowa, Indiana, Minnesota, Missouri, Nebraska, New Jersey, Oregon, South Carolina, Pennsylvania, and Virginia. Some tracked the filing of cases but did not note the resolution.

These data were supplemented the purchase of public eviction records from LexisNexis Risk Solutions (excluded North Dakota and South Dakota) and American Information Research Services, Inc. The researchers primarily used the LexisNexis data since they were the most complete. To validate the data and provide additional downloadable statistics, the Eviction Lab also collected county-level statistics from 27 states, New York City, and the District of Columbia. To convert the counts into rates, the researchers used data from the decennial Census and American Community Survey.

The data collection effort yielded 82,935,971 court records, which were narrowed to 38,564,127 unique cases. Cases that were later merged or omitted included those related to security deposits or other non-eviction matters, duplicate records from the same case number, and blank or test records. The researchers retained the serial eviction cases, in which the household and address are the same but the action date is not. Among the unique cases, some were further excluded during data analysis, including commercial cases and those outside of the years 2000 to 2016. The methodology paper further notes the states in which inconsistent data coverage led to exclusion or the substitution of data from state-reported statistics.

Some data, including from Arizona, California, Connecticut, Hawaii, Idaho, Maryland, New Hampshire, New Jersey, Vermont, Washington, Wyoming, and the District of Columbia, are likely to underestimate evictions for a variety of reasons. In the online map, figures for these places are shown but have a tool tip noting the undercount. Many other states have suspected underestimates in remote or rural areas. Entities with additional data are encouraged to submit them to the Eviction Lab to improve the resource.

A nonpublic version of the dataset includes name and address. For research purposes, the Eviction Lab accepts data merge requests and will return a de-identified merged dataset to researchers for free.

About the dataset

  • The data include eviction filings and eviction judgements, where data were available.
  • The individual level records, which are not shared publicly by the Eviction Lab, include court records for renter households but do not indicate how many people reside in each affected home.
  • Eviction counts in 2016 may include projected counts for areas in which the last data available were from 2015.
  • In total, the researchers found 5,461,151 households with serial eviction filings at the same property address—indicating that landlord/tenant court hearings often operate as a collection agency without necessarily resulting in an eviction.
  • 66,512,547 records came from LexisNexis.
  • 12,870,070 records came from 13 states directly.
  • 10,606,467 records came from American Information Research Services, Inc.
  • 597,327 commercial eviction cases were excluded.