Model Comparison: How to constrain a model to positive mu-differences?

[junpenglao]

@Olaf

From what I gathered from a tutorial in R about rat growth that @OriolAbril provided, in R this formula would be kind of this σproj∼(condition∗block∗direction)/(1|trial)+(1|user)\sigma_{proj} \sim (condition * block * direction)/(1|trial) + (1|user) , meaning trials are viewed as nested random effects, and users as random effects. It seems patsy doesn’t support nesting and random effects.

IIUC, you have condition, block, direction which potentially be different, and user could also display different strategy - in that case, I recommend you to try parsing the linear formula σproj∼(condition∗block∗direction*user) in patsy, and use indexing for trial (the random effect.)

In general, frequentist testing like ANOVA are GLMs, and GLMs are pretty straightforward to express in Bayesian - I would love to give you a bit more concrete suggestion, maybe it is easier you share a jupyter notebook with some simulation data?