My understanding is that the IDinsight study had about 35 percent of beneficiaries choosing saving lives over any presented amount of cash transfers, while the respondents who switched based on the # of transfers per life saved expressed preferences that were (on the median) roughly similar to preexisting GiveWell weights.
So I think a model that gives about 35 percent weight to a lives-saved-only calculus and about 65 percent weight to a tradeoff calculus would accurately reflect the beneficiary preferences in the survey. That isn’t too far off from the 60/10/30 model.
Going solely by the preferences in this one survey might affect some GiveWell decisions on the margin, but the bulk of their top charities’ rated effectiveness already comes from saving lives under five.
That’s still not accurate because the YLLs adjustment mostly matters on the intensive margin (saving kids vs adults), and as GiveWell states, their lives vs income tradeoff was essentially unchanged compared to their previous moral weights.
This is a good point.
My understanding is that the IDinsight study had about 35 percent of beneficiaries choosing saving lives over any presented amount of cash transfers, while the respondents who switched based on the # of transfers per life saved expressed preferences that were (on the median) roughly similar to preexisting GiveWell weights.
So I think a model that gives about 35 percent weight to a lives-saved-only calculus and about 65 percent weight to a tradeoff calculus would accurately reflect the beneficiary preferences in the survey. That isn’t too far off from the 60/10/30 model.
Going solely by the preferences in this one survey might affect some GiveWell decisions on the margin, but the bulk of their top charities’ rated effectiveness already comes from saving lives under five.
That’s still not accurate because the YLLs adjustment mostly matters on the intensive margin (saving kids vs adults), and as GiveWell states, their lives vs income tradeoff was essentially unchanged compared to their previous moral weights.