Useful analyses for presocial0.1

[jkbhagatio]

• Which statistical test(s) to use to compare distributions for easy and hard patch cumulative wheel distance?
  • Wilcoxon rank-sum?
• When doing stats and viz:
  • Look at cumulative wheel distance data post threshold change.
  • For a particular lambda, look at:
    • all sessions together
    • all mice for a particular session number together
    • all sessions for a particular mouse together

[jerlich]

I think we did something like compute delta_prefA = log2 ((postA/preA) / (postB/preB)) where pre and post are the wheel distance in some time interval (30 min or 1 hour) before and after a change. If there was no change you expect this to be normal with mean 0.

then for aggregate stats you can do a mixed effect model like delta_prefA ~ delta_threshold + (delta_threshold | subjid) so you can get an estimate of the sensitivity of each mouse to the change. (or just do delta_prefA ~ delta_threshold separately for each mouse. You can use this code if you want to use python. Otherwise lme4 in R or MixedModels.jl