Powersim for logistic regression

[Kirsten Barnicot]

Hi I wonder if anyone can help with a very simple question? When we use Powersim to generate and analyze the data using a binomial model with a logit link function, as in Example 3 in the Powersim manual, does the effect size b that we specify refer to the odds ratio or to the unexponentiated coefficient? Thank you in advance for your help - this will really be very helpful to know :-)

[Leonardo Guizzetti]

Per the help for -powersim-:

expb Effect sizes specified in option b() are interpreted as exp(b); for example, this could be useful if the power wants to be determined in the context of a logistic regression and the user wants to specify effect sizes as odds ratios, rather than as log odds ratios.

[Kirsten Barnicot]

Amazing, thanks so much Leonardo - that's very helpful.