delay-discounting-analysis

Hyperpriors

The effect of doing hierarchical estimation, can be seen in the figure below. We will always have uncertainty about a participant’s true discount rate (seen here with log(k)) but we can improve our estimates when we assume participants are drawn from a population. When we do this, our belief of the group-level discount rate (bottom panel, thick line) acts to ‘shrink’ our participant-level estimates toward the group mean.