PD&R Edge Banner

May 17, 2022

Tenant Screening With Criminal Background Checks: Predictions And Perceptions Are Not Causality

  Image of Calvin Johnson, Deputy Assistant Secretary for the Office of Research, Evaluation, and Monitoring. Calvin Johnson, Deputy Assistant Secretary for the Office of Research, Evaluation, and Monitoring.

Nearly 1 in 3 adults in the United States has a criminal arrest record that often serves as a barrier to housing, employment, and a range of social services. People with criminal records face landlords who refuse to rent to them even when they are financially qualified, employers who will not hire them even when they have the necessary skills, and service providers who deny them access to needed services even when they are eligible to receive them. Those with criminal records are consistently shut out from housing, employment, and services.

Criminal history is the compilation of criminal records. Criminal history data typically are used in risk assessment and screening tools across criminal and juvenile justice settings, but they have not been studied as a predictor of housing retention. Nevertheless, housing providers and their management agents routinely conduct criminal background checks as part of their tenant screening processes, believing that criminal history is an accurate predictor of criminal behavior that could harm other tenants. Housing providers and their management agents rely on findings from recidivism studies that show rates of rearrests following release from jails and prisons to support their stance, even though recidivism may not be a good proxy for housing problems.

Criminal history is not a good predictor of housing success. A study of housing outcomes among tenants participating in an intervention based on the Housing First model found that the performance of tenants with a criminal history was similar to that of participants without a criminal history. Although few studies examine the association between criminal history and housing success, no empirical evidence exists to justify a ‘blanket exclusion’ of people with criminal histories from housing.

Risk Assessments: Prediction Is Not Causality

Researchers in the behavioral and social sciences have a long history of using risk assessment and screening tools to standardize approaches to estimating the risk of future harm among justice-involved and other forensic populations. These tools are intended to reduce implicit bias, increase fairness, and allow consistency and transparency in juvenile and criminal justice decisionmaking. These tools, however, have also been criticized for using historical outcome measures that are inherently biased because of discriminatory criminal justice practices. Despite an extensive behavioral and social science literature on risk assessment and screening principles and practices, stakeholder confusion about how accurately these tools estimate the risk of future harm leads to their misuse and has nontrivial impacts on decisionmaking.

Specifically, risk assessment and screening tools are prognostic in nature, using information about past exposure to the outcome of interest and other risk factors, exposure (past and present) to preventive intervention services in response to past exposures, and other social context factors to predict the likelihood of a particular outcome. There are currently no empirically validated tools predicting the risk of harm a rental applicant might present to other tenants and property available to housing providers and their property managers. Instead, housing providers and management agents use their perception of risk an applicant with a criminal record might pose to other tenants and property. More and more housing providers and their management agents are recognizing the limitation of this approach and are assessing criminal histories alongside additional contextual information like the seriousness of the crime, whether arrest led to conviction, history of participating in preventive interventions and other rehabilitative programs, history of participating in behavioral health interventions, length of current employment, ties to social and community supports, civic and social engagement, and other protective factors. In so doing, a financially capable 35-year-old rental applicant who was arrested at age 22 for possession of marijuana would not be automatically excluded from rental consideration. Instead, the housing provider or their management agent might take into account the applicant’s 10 years of employment as a bus driver for the public transit system, completion of a drug court program, completion of the CDL certification, and active participation in PTA and faith-based ministries.

As prediction is not causality, we have to accept that predictions look backward to estimate an outcome that has not yet occurred and may never occur. One risk assessment tool widely used within the criminal and juvenile justice system highlights this point. Using data from a large county in southern Florida, researchers from Florida State University assessed the recidivism rates for nearly 6,000 people released from jail and awaiting trial.


Table 1. Rearrest Among Pretrial Releasees by Risk Classification
All Risk LevelsLow RiskMedium RiskHigh Risk
Individual classified----57%25%18%
Any arrest29.6%18.0%38.4%61.0%
Violent arrest only6.8%3.0%4.8%12.5%

The takeaway from this table is that predicting future criminal involvement is a complicated business. Even using the best assessment and screening tools that undergo regular validations and enhancements, predictions are often wrong.

As table 1 shows, there is significant variability in rearrest. Although nearly 30 percent of the sample was rearrested for any arrest, 18 percent of individuals classified as low risk were rearrested compared with 61 percent of individuals classified as high risk. The breakout for violent rearrest only shows that 6.8 percent of those classified as low risk were rearrested compared with 12.5 percent of those classified as high risk. Note that nearly 60 percent of the sample was classified as low risk compared with less than 20 percent classified as high risk.

Tenant screening services have adopted similar actuarial approaches to assessing risk of tenant performance, including the likelihood of default or owing a balance when moving out. I have yet to identify research on tenant screening that focuses on assessing the risk of harm to other tenants or the property itself.

Why Does This Matter?

Tenant screening with criminal background checks is a form of risk assessment that has two primary components: an assessment of an applicant’s creditworthiness and an assessment of the applicant’s potential safety risk to other tenants and property. As such, the criminal background check is what housing providers and their management agents use to formulate their perception of the risk of future harm an applicant might pose to other tenants. They do not have a validated assessment or screening tool. As described above, more and more housing providers and their management are recognizing the limitation of that approach and are taking into account contextual and other social factors that have empirically been shown to reduce criminal involvement. And prediction is not causality — they are often wrong, even when validated assessment and screening tools are used to assess the risk of future offender.

Equity Lens

Tenant screening “raises the bar to entry for all renters.” People of color, however, are disproportionately affected by tenant screening practices that include a criminal background check. Disproportionate patterns of arrests in poor communities and communities of color mean that using criminal background checks to screen tenants disadvantages renters who are people of color. Even when these screening practices are applied to all potential renters, housing providers and their management agents often have found discriminatory workarounds, including the following practices:

  • Signaling: Landlords selectively mentioned credit and criminal background check requirements to black testers;
  • Preferential treatment: White women posing as rental applicants with criminal records received preferential treatment compared with similarly situated black women.

Jurisdictions nationwide are realizing the harm discriminatory criminal justice practices have caused as well as the potential for repeated harm in tenant screening with criminal background checks. As of 2018, four jurisdictions — Newark, New Jersey; San Francisco; Washington, DC; and Seattle — have instituted “Ban the Box” laws regulating rental housing providers’ use of criminal records. Laws in these jurisdictions either prohibit questions about criminal history on rental applications or permit only an individualized approach the use of criminal history information later in the application process.

Source:

Bureau of Justice Statistics. 2014. “Survey of State Criminal History Information Systems, 2012: A Criminal Justice Information Policy Report.” ×

Source:

Daniel K. Malone, 2009. “Assessing criminal history as a predictor of future housing success for homeless adults with behavioral health disorders,” Psychiatric Services 60:2, 224–30. ×

Source:

Daniel K. Malone, 2009. “Assessing criminal history as a predictor of future housing success for homeless adults with behavioral health disorders,” Psychiatric Services 60:2, 224–30. ×

Source:

Chelsea Barabas, Madars Virza, Karthik Dinakar, Joichi Ito, and Jonathan Zittrain. 2018. “Intervention over prediction: reframing the ethical debate for actuarial risk assessment,” Proceedings of Machine Learning Research, 81, 1–15; Alex Chohlas-Woods. 2018. “Understanding Risk Assessment Instruments in Criminal Justice,” Brookings Institution; Aron Shlonsky and Dennis Wagner. 2005. “The next step: Integrating actuarial risk assessment and clinical judgment into an evidence-based practice framework in CPS case management,” Children and Youth Services Review 27:4, 409–27. ×

Source:

Alex Chohlas-Woods. 2018. “Understanding Risk Assessment Instruments in Criminal Justice,” Brookings Institution; Julia Angwin, Jeff Larson, Surya Mattu, and Lauren Kirchner. 2016. “Machine Bias,” ProPublica. ×

Source:

Thomas Blomberg, William Bales, Karen Mann, Ryan Meldrum, and Joe Nedelec. 2010. “Validation of the Compas Risk Assessment Classification Instrument.” ×

Source:

Anna C. Reosti. 2018. “Tenant Screening and Fair Housing in the Information Age,” Dissertation, University of Washington. ×

Source:

Seattle Office of Civil Rights. 2018. “2018 Testing Program Report.” ×

Source:

The Equal Rights Center. 2016. “Unlocking Discrimination: A DC Area Testing Investigation About Racial Discrimination and Criminal Record Screening Policies in Housing.” ×

 
 
Home Page Archives