x

Please fill out this short user survey of only 3 questions in order to help us improve the site. We appreciate your feedback!
Pilot Study: Improving Court Outcomes Improving Criminal Justice Outcomes for People Experiencing Housing Instability in Shasta County, California
Last registered on October 20, 2020

Pre-Trial

Trial Information
General Information
Title
Pilot Study: Improving Court Outcomes Improving Criminal Justice Outcomes for People Experiencing Housing Instability in Shasta County, California
RCT ID
AEARCTR-0006606
Initial registration date
October 16, 2020
Last updated
October 20, 2020 7:39 AM EDT
Location(s)

This section is unavailable to the public. Use the button below to request access to this information.

Request Information
Primary Investigator
Affiliation
University of California, Irvine
Other Primary Investigator(s)
PI Affiliation
Claremont Graduate University
Additional Trial Information
Status
In development
Start date
2020-12-01
End date
2021-02-01
Secondary IDs
Abstract
Existing evaluations of text- message based reminders of court appearances have been found to substantially reduce the probability that a defendant fails to appear (FTA) in court. Reducing FTAs can potentially minimize the link between criminal conviction and persistent poverty, as FTAs for minor misdemeanors lead to arrest warrants being issued, increased fines, or even incarceration. The research team has submitted a large proposal to evaluate two potential strategies to reduce FTAs among the people charged with misdemeanors associated with homelessness. This complementary pilot proposal is intended to replicate an existing successful text-message based reminder system, implemented in New York City, in Shasta County.

Given the institutional and population differences across these two studies, it is not obvious that the findings of Cooke et al. (2018), which have generally been replicated in other large urban jurisdictions, will hold in smaller, rural court. Further, the target population, people at risk of homelessness, may be less likely to receive text messages and may respond to those messages differently. Our focus on those experiencing housing instability is crucial because these individuals are the most likely to miss court and experience economic distress.

We propose to implement this pilot study over two months, with a one month follow up period, during which we will produce a short summary of the pilot results, comparing FTA rates across control and treatment groups, and for those charged with misdemeanors in general and those charged with misdemeanors related to homelessness. These results will allow us to place the findings of our larger study in the larger context of research on reducing FTAs.


External Link(s)
Registration Citation
Citation
Owens, Emily and CarlyWill Sloan. 2020. "Pilot Study: Improving Court Outcomes Improving Criminal Justice Outcomes for People Experiencing Housing Instability in Shasta County, California." AEA RCT Registry. October 20. https://doi.org/10.1257/rct.6606-1.0.
Sponsors & Partners

There are documents in this trial unavailable to the public. Use the button below to request access to this information.

Request Information
Experimental Details
Interventions
Intervention(s)
Pilot Proposal Narrative

Our larger proposed experimental evaluation will identify the effect of two court outreach programs specifically targeted at people experiencing homelessness. Both interventions are likely to involve substantial costs relative to what existing field experiments have increasingly shown effectively reduce FTAs in the general population: text message-based reminders.

There is a consensus among practitioners and researchers who work in this area that text reminders are unlikely to be effective for people experiencing homelessness. Still to our knowledge, this reasonable conjecture has not been explicitly tested.

If the results of Cooke et al. (2018) were generalizable to the Shasta County context, we would expect an approximately 28% reduction in FTAs in a treatment group receiving this message relative to a control group. We do not necessarily anticipate Cooke et al. (2018) to be generalizable, however, for two reasons. Most importantly, eligibility for the Cooke et al. (2018) evaluation was predicated on providing a cell phone number to the court. Secondly, Cooke et al. (2018) was conducted in New York City, as opposed to the less dense, more rural Shasta County.

As a result, while we can appeal to existing experimental studies to estimate that text-message based reminders can lower the FTA rates in misdemeanor cases by up to 30%, we do not have a clear prior about the size of τ, the impact of text-message based reminders on FTA rates for people experiencing homelessness.

Assignment to Treatment

We, therefore, propose a short pilot study that will attempt to replicate the results of a recent text message evaluation, Cooke et al. (2018), in Shasta County. Once court dates are scheduled, we will construct a stratified random sample of people charged with general misdemeanors and homelessness-related misdemeanors to receive treatment. All individuals with assigned court dates are currently sent automated voice messages via PhoneTree. Through an app developed by a third party, treated individuals will also be sent text message reminders.

We will use a slightly modified version of the most effective message in Cooke et al. (2018), modified slightly to reflect institutional constraints in Shasta County, which both reminded individuals of the consequences of failing to appear, and prompted recipients to plan how they would appear.

You have court on (*date, three days from now*, at *time*) at 1500 Court Street, in Redding. What time should you leave to get there by *time*? Any other arrangements to make? Missing court can lead to your arrest.

We will evaluate this study by comparing the FTA rates across control and treatment groups, accounting for the within-charge type randomization. We will also test whether the impact varies across people more or less likely to be experiencing housing instability.

To inform our planned larger study, after the completion of the experimental period, we will compare the individuals in our sample, in both control and treatment groups, to rosters of individuals receiving services at a large local service provider. This will provide insight into the feasibility of providing additional reminders through local service providers in a subsequent larger study, including providing parameters for a power calculation.

We also intend to use data in this pilot to explore potential heterogeneity in effects along various socioeconomic and criminal justice related dimensions to inform any potential blocking designs in a subsequent larger study.


References
Cooke, Brice, Binta Zahra Diop, Alissa Fishbane, Jonathan Hayes, Aurelie Ouss, and Anuj Shah. 2018.
Using Behavioral Science to Improve Criminal Justice Outcomes: Preventing Failure to Appear in Court.” Chicago. University of Chicago Urban Lab Report.


Intervention Start Date
2020-12-01
Intervention End Date
2021-02-01
Primary Outcomes
Primary Outcomes (end points)
Failure to Appear in Court (FTA), and how FTAs vary by charge (homelessness related or not)
Primary Outcomes (explanation)
The Shasta County Superior Court has identified a set of misdemeanors it considers related to homelessness. In addition to comparing how our intervention affects FTA rates, we will test for differences in the rate at which the intervention affects people who are more or less likely to be experiencing housing instability.
Secondary Outcomes
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
Analysis Plan

Our pilot analysis aims to obtain a benchmark against which to compare existing text-based reminders to our results, and to verify the extent to which people likely to be experiencing homelessness deferentially respond to these reminders. We will calculate intent to treat effects, which are the most practically relevant statistic, using the following linear probability model

FTA_IT=Α_T+HOMELESS_IT Θ_1+ INTERVENTION_I Θ_2+(INTERVENTION_I X HOMELESS_IT) Θ_3+U_IT

where i indexes individual i charged with a misdemeanor scheduled on date t.

Recall that we will randomly assign cases to reminder or control status separately for people charged with homelessness and non-homelessness related misdemeanors. Our estimate of Θ_2 will represent the average intent-to-treat effect of reminders in Shasta County that is more comparable to previous research, which excluded people likely to be housing insecure. Our estimate of Θ_3 will provide us with a qualitative sense of how less likely housing insecure people are to respond to text messages; while we may not have a large sample to formally test the hypothesis that texts a differentially effective across group, for the purposes of this experiment the statistical significance of Θ_3 is perhaps less important than is size relative to Θ_2 .

Between November of 2018 and January of 2020, 4,465 cases were for charges unrelated to homelessness, compared to 442 homelessness related offenses. The relative FTA rates for these cases was 52% and 80%, respectively. Over this period, people had an average of 1.6 court appearances. We, therefore, anticipate that, over the proposed two-month pilot, we will not observe more than one individual per case.

Based on these base rates, if text-messages reduced FTAs by 28% in both groups, and we randomly assigned 50% of non-homeless people and 50% of homeless people to treated status, we would be able to correctly reject the null hypothesis of no change in FTAs in sample sizes of 491 non-homeless cases and 179 homeless cases, with 80% power. Based on existing records, over two months we would expect to have an experimental sample of 594 people charged in non-homeless cases and 60 homeless cases. While we would have limited power to test the hypothesis that text messages reduced FTAs among the population of people experiencing homelessness, we would be adequately powered to distinguish between the homeless and non-homeless populations if the FTA reduction, on average, differed by more than 8 percentage points (a 28% reduction versus a 20% reduction).
Experimental Design Details
Not available
Randomization Method
Randomization will be done by a computer. Randomization will be stratified by misdemeanor charge type (homelessness-related or not)
Randomization Unit
Individual
Was the treatment clustered?
Yes
Experiment Characteristics
Sample size: planned number of clusters
approximately 660, 600 non-homeless, 60 homeless
Sample size: planned number of observations
We do not anticipate individuals appearing in court more than once, so we expect the sample size to be roughly equivalent to the number of clusters.
Sample size (or number of clusters) by treatment arms
300 non-homeless control, 300 non-homeless treated
30 homeless control, 30 homeless treated
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Overall, a 28% reduction in FTAs, a difference in outcomes across groups of more than 8 percentage points.
IRB
INSTITUTIONAL REVIEW BOARDS (IRBs)
IRB Name
IRB Approval Date
IRB Approval Number