An recently posted RCT of the Denver Basic Income Project - which provided unconditional cash of $12K over 1 year to unhoused people - reported "groundbreaking findings," covered in national and local press. Quick take: The study actually found minimal effects on housing and other outcomes at 10 month mark.
Study Design:
The study randomized 807 unhoused people in Denver to: Group A (paid $12K over 1 year in equal monthly installments); Group B (paid $12K over 1 year, with half paid as an up-front lump sum); or Group C (control, paid a nominal $600 over 1 year in equal monthly installments). The study pre-specified that it would compare outcomes for Groups A and B to Group C (control).
Findings:
The study states that it found "minimal differences" in outcomes among the 3 groups. For example, the Project's website features this graph showing that housing stability for Groups A and B was very similar to C at the 10-month follow-up ("Timepoint 3" - dark green bar):
The study nevertheless cites these results as evidence of success by ignoring the RCT design and pointing to the improvements from baseline for all 3 groups, including control (a common occurrence that could be due to "regression to the mean"):
In short, the study misreported disappointing RCT findings as positive. The misreported positive findings unfortunately received wide press coverage (e.g., NPR, CBS, CNN, Newsweek, Colorado Sun, Denver Post, Denver Gazette).
Comment:
We highlight this study due to the misreporting. But we'd also note the study lost ~50% of its sample to attrition at the 10-mo follow-up, so would regard its findings - even if accurately reported - as suggestive but not reliable.