That is, if both effects are in the data (i.e., the true model is M1: het‐GLM‐skew) these are detected with a power close to 1.00, and if one of the effects is in the data, that effect is detected with an acceptable power, and the other effect (which is not in the data) is detected with a power close to the level of significance (indicating that the Type I error rate is not increased).
← all excerpts
A heteroscedastic generalized linear model with a non-normal speed factor for responses and response times.
1
—
—