Why is determining the impact of housing so challenging?

This is the third part in a three-part series Recidiviz has written summarizing recent high-quality academic work about policies and programs that affect justice-involved individuals. The goal is to make research more accessible to practitioners, helping them make evidence-based decisions. As always, we welcome thoughts and feedback.

A naive approach involves comparing outcomes for those who spend time in transitional housing to those who do not. However, this method suffers from selection bias: those who stay in transitional housing may be different in important ways from those who do not, which itself could drive differences in outcomes. Perhaps those who are not required to stay or opt-out are more likely to have a supportive family member they could live with or more means to afford stable housing.

A credible way to study the effects of transitional housing requires an experimental approach. Ideally, we’d have a large pool of people, randomly assign transitional housing to some of them (the treatment arm), and compare their outcomes with the rest of the pool (the control arm). As far as we know, such an experiment has not been conducted in over four decades.¹

The next best thing is a natural experiment–some setting in which an individual’s decision to stay in transitional housing is strongly influenced by a policy or something else out of the individual’s control. Thus, when otherwise similar people vary only in whether they stayed in transitional housing, we can assume that differences in their outcomes are caused by housing itself. Next, we review an article that exploits a natural experiment in Iowa.

First, Lee documents that case managers strongly influence whether someone will stay in residential housing or simply be released to parole as usual. While high-tendency case managers required half of their cases to live in residential housing, low-tendency case managers only send one in four.

So, what are the effects of residential housing on recidivism? Lee finds that being assigned to residential housing does not reduce reincarceration. This is both surprising and disappointing given that housing programs are 14 times more costly than parole and are intended to provide a stable place to live and reduce recidivism. In fact, Lee finds that assignment to residential housing actually increases violent crime by about 7 percentage points. Moreover, individuals assigned to residential housing are much more likely to be reincarcerated for technical violations. He posits that stricter rules and monitoring in halfway homes may provide greater opportunities to fail.

Justice-involved individuals often report that stable housing is critical for reintegration. The papers we review here make it clear that simply providing a place to live is not enough: the type of housing, supervision rules, and peer influence should all be taken into consideration. Research on the effects of providing free or subsidized housing (e.g., providing housing vouchers to the housing insecure) is limited and should be expanded. If programs that reduce crime and subsequent incarceration can be identified, the societal benefits will easily justify the additional costs.

² Although researchers have studied multiple housing programs, none have experimentally tested programs that provide housing vouchers to people leaving prison.

Related posts:

When it comes to screen adaptations of novels, it’s often a hit or miss. The Harry Potter films are considered to rival the books, depending on who you ask, and yet the Percy Jackson films are not to…

True to my generalist ways, I fall down a lot of internet rabbit holes. The latest is Hot Ones, a show where celebrities with a book, movie, or lifestyle line to hawk eat progressively hotter chicken…

Git is an open-source software and distributed version control system. It helps developers easily handle different versions of a source code. With it, you can know who did what, when, and why…