JAMA Internal Medicine published an RCT examining whether remediation of abandoned housing in Philly reduces surrounding crime. Quick take: Mostly well-conducted RCT finds reductions in gun-related crime that are moderately encouraging but not yet reliable (might be due to chance).

Program & Study Design:

• The study randomly assigned 63 groups of abandoned houses to (1) full remediation (installing working windows/doors, cleaning trash, weeding); (2) trash cleanup & weeding only; or (3) no intervention (control). It measured surrounding crime 18-months pre- & post-intervention.

Findings:

• The study found that Full Remediation produced statistically significant reductions of 8% in weapons violations and 13% in gun assaults, vs the control group, but no significant effects on the 5 other primary outcomes. Trash clean-up only had no significant effects vs control.

• However, the effects on weapons violations & gun assaults lost statistical significance under an alternative (trend-adjusted) analysis. They also lost significance when the main analysis adjusted for the study's measurement of many outcomes (which can lead to false-positives).

Comment:

• Bottom line — I think the effects are moderately encouraging, but the study can't convincingly rule out the possibility they're due to chance (as opposed to the program). This seems like a great candidate for a (larger) replication RCT, to hopefully confirm the results.

• As to study quality - I think generally good, with one qualification: Crime outcomes were measured a few months later in the Full Remediation group vs the Control group. Statistical analyses may not be able to fully correct for this difference & prevent bias in the results.