search

elmacartney / Droso_social_env_behaviour

MIT License
0 stars 1 forks source link

Prior structure #3

Open elmacartney opened 1 year ago

elmacartney commented 1 year ago

I have now adjusted the model to include the binary startle response, because the data set is severely reduced if we used the difference in response across startles as many individuals do not startle.

The model now has family = c("gaussian", "categorical", "poisson") but needs the priors adjusted (@itchyshin mentioned that it needs priors for the fixed effects).

Would either @szymekdr or @itchyshin be able to look at the model and adjust the priors? Code found at end of RMD_2.rmd under "correlation analysis for manuscript"

:)

itchyshin commented 1 year ago

@szymekdr

Could you add gelman prior, please

https://rdrr.io/cran/MCMCglmm/man/gelman.prior.html

@elmacartney has "categorical" because 0/1 for startle (as she explained above)

Without setting gelman stuff - things do not usually coverage so well and fixed effect estimates are often strange.

Thanks!

Shinichi