The Current State of Eviction Data

We believe the primary challenges with U.S. eviction data can be broken up into the following categories:

Multiple data points at the beginning and the end of the eviction process are never collected and generally do not leave a discoverable trail.

These data include pre-eviction information that could serve as an early warning that a tenant is in trouble: for example, rental assistance requests, alerts that tenants are behind on rent, complaints against the landlord, notices to quit, and so forth. They also include data on what happens after an eviction judgment, including information as to whether a tenant was ultimately evicted (and whether the eviction was conducted by law enforcement, or if the tenant moved out on their own accord), where the tenant moved after the eviction, how long it takes them to find a new home, and how long they stay at that home.

Non-existent data also include pieces of information that courts do not collect, but that would be helpful to service providers, policymakers, and researchers in targeting renter assistance. Examples include the renter’s race or ethnicity, citizenship status, age and gender, as well as the composition of the household (e.g., are there children living at the address).

Finally, data on landlords (i.e., the landlord’s record of evicting tenants, the landlord’s history of code violations, the residency of the landlord) and on the reason for eviction are often not collected.

Some courts never digitize eviction data, and it remains in hand-written form. Other courts do digitize it, but it remains stored on a local computer or in PDF printouts, and never shared to a database. Converting this data into a readily usable format (e.g., Excel, CSV, or API) is extremely time consuming and difficult to do at scale. Sometimes vendors like American Information Research Services and Premonition do collect and aggregate this data, which they then sell to tenant screening services and credit agencies. As a result, accessing these data is generally cost prohibitive for the public, researchers, service providers, or policymakers to access. These data may also not be as comprehensive as those collected by local researchers.

Other courts do share their records with legal services like Westlaw and Lexis-Nexis, which allow licensed users to search dockets by case number, name, and in some instances by case type. A challenge is that this system is geared towards searching for specific cases, one by one, and absent sophisticated web scraping software, it is difficult to analyze aggregate eviction data. Another challenge is that access to these legal databases is expensive, and so these data are effectively unavailable to the public.

As one stakeholder put it: “A court’s job is to adjudicate the cases and move on.” With no mandate and no resources to curate data into usable databases, most courts don’t bother.

According to experts, certain states, including Alabama, Colorado, Kansas, Kentucky, Louisiana, Mississippi, Nebraska, South Dakota, and Utah, do aggregate the data into databases but charge fees for its access. Even where jurisdictions aggregate eviction data into purportedly public databases, these databases are not usually easy to access. Locating them requires a significant allocation of time and resources, and often necessitates connecting with the right government official or office and convincing them to share the data.

Why is there so much variability? Because courts have neither mandate nor funding to curate and store eviction data. As one stakeholder put it: “A court’s job is to adjudicate the cases and move on.” With no mandate and no resources to curate data into usable databases, most courts don’t bother.

A combination of human entry error, data stewardship disparities, and quirks in the data themselves results in eviction records that lack fields like addresses and disposition information, contain multiple repeat entries, or miss entire blocks of entries. Cleaning and interpreting these data is a lengthy process and sometimes requires localized knowledge.

Because there is no national mandate or framework for stewarding eviction data, jurisdictions vary dramatically in types of data they collect and make available, the terms of access to those data, and even the definitions of eviction-related terms. There is no shared taxonomy for eviction data, even within the same state.

As a result, it is difficult to make cross-jurisdictional comparisons or to compile data from the county level to the state or national level.

Eviction data is often examined in isolation, not in conjunction with other datasets that could paint a fuller picture of the challenges leading to and arising from eviction. These data could include court data like criminal records, debt records, and guardianship records, social services data like homelessness records, or housing databases.

As discussed in Challenge A above, pre-eviction data largely does not exist, posing major challenges for eviction prevention work. However, even eviction records are not always up-to-date. While some courts upload and populate eviction records weekly, others may only update their systems quarterly. Even if courts do upload records on a consistent basis, for-profit aggregators may not make data available for purchase immediately.

Court data reveal personal information about both the landlord and the tenant, including full name and address. To some parties, for example, service providers who can deliver rental assistance or connect evicted tenants to new housing, this data is extremely useful. However, to many others, including researchers and policymakers looking to understand neighborhood or city-level trends, they are unnecessarily revealing. Particularly when linked with other data, these details can reveal highly personal information that the data subject may be uncomfortable sharing publicly. Currently, very few eviction data systems have permissions and access levels to differentiate the type of information available to different parties.