Response to GiveWell’s assessment of StrongMinds
Updates
November 2024: Update to our analysis
We have made a substantial update to our psychotherapy report. This 83 page report (+ appendix) details new methods as well as an updated cost-effectiveness analysis of StrongMinds and Friendship Bench. We now estimate StrongMinds to create 40 WELLBYs per $1,000 donated (or 5.3 times more cost-effective than GiveDirectly cash transfers). We now estimate Friendship Bench to create 49 WELLBYs per $1,000 donated (or 6.4 times more cost-effective than GiveDirectly cash transfers).
See our changelog for previous updates.
Summary
GiveWell reassessed HLI’s cost-effectiveness analysis of StrongMinds and reached a figure 83% lower, 1.8 WELLBYs per treatment against our 10.5. That reduction is the product of seven researcher-made adjustments rather than new evidence. Of the 83%, we think 5% is clearly supported by the evidence and about 45% is plausible but of uncertain size, mainly the depression to life-satisfaction conversion and the loss of effectiveness at scale. We are unsympathetic to roughly 35%, which turns on household spillovers where our reading of the evidence differs, and we do not think the evidence supports the remaining 15%, the discount to durability. On costs we agree with GiveWell, and StrongMinds’ own reporting suggests a figure lower still than the $105 they use.
GiveWell also concludes that the Against Malaria Foundation is 4x more cost-effective than StrongMinds. Comparing a life-saving intervention against a life-improving one depends heavily on the account of the badness of death and the location of the neutral point, and GiveWell adopts the assumptions most favourable to saving lives without flagging that other reasonable positions reverse the ranking. HLI takes no institutional stance on these philosophical questions, but they are decision-relevant and deserve more attention than the report gives them.