Tidy dispersion parameters?

[bwiernik]

Would it be possible to add dispersion parameter information to the tidy output? Ideally, it would also be nice to have the option to exponentiated and exponentiate+square root transform the estimates and confidence intervals.

[bbolker]

In principle using cond=c("cond","zi","disp") should work (haven't tested recently)? There is an exponentiate argument as well (also may not be well-tested). Out of curiosity, what would be the use-case for an exp+sqrt transform?

[bwiernik]

Currently, there is a check that errors if cond contains something other than "cond" or "zi".

For the exponentiate argument, it'd be nice to be able to specify it separately for cond vs zi vs disp. For example, if I fit a linear regression model with a dispersion model, I'd like to exponentiate the disp parameters, but not the cond parameters.

For the exp+sqrt transform, for a linear regression model like I describe above, the dispersion parameter are in log(variance) units. An exp+sqrt transform expresses them in (multiplicative) standard deviation units.

[nthun]

@bbolker it seems like the component= "disp" doesn't work currently:

Error in tidy.glmmTMB(mean_model, component = c("cond", "disp")) : only works for conditional and (partly for) zero-inflation components