sidenote: I’d be interested to what extent ACE now uses Bayesian reasoning in their estimates, e.g. by adjusting impact by how likely small sample studies are false positives.
Our current methodology uses an alternative approach of treating cost-effectiveness estimates as only one input into our decisions. We then take care to “notice when we are confused” by remaining aware that if a cost-effectiveness estimate is much higher than we would expect based on the other things we know about an intervention or charity, that may be due to an error in our estimate rather than to truly exceptional cost effectiveness.
Our current methodology uses an alternative approach of treating cost-effectiveness estimates as only one input into our decisions. We then take care to “notice when we are confused” by remaining aware that if a cost-effectiveness estimate is much higher than we would expect based on the other things we know about an intervention or charity, that may be due to an error in our estimate rather than to truly exceptional cost effectiveness.
We admit that Bayesian techniques would more accurately adjust for uncertainty, but this would require additional work in developing appropriate priors for each reference class, and this process may not generate worthwhile differences in our evaluations, given our data set. See this section of our Cost-Effectiveness Estimates page for details on our thinking about this.