explore how models meaningfully affect inferences

[drbenvincent]

Explore how the Rachlin versus the Modified Rachlin discount function affect inferences. We will do this by simulating datasets with known differences in (k, s) or (kappa, s) and examining our inferences of the effect sizes.

With the iteration over true effect sizes, this is going to take a long time to compute.

• try variational inference?
• understand and fix the occasional error we get

TODO:

• get summary stats on inferred effect size
• wrap into a function
• iterate over true effect sizes
• plot true vs inferred effect sizes
• scatter plot of true (kappa, s) parameters for the simulated populations.

Then

• make it all work for original Rachlin also, so we get plots size by side
• check accuracy of effect size calculation
• add in more examples