Life is all about making choices, and stats is no exception. You can either select the best predictive model out of a big set OR you can look at p-values from a model. Not both! Good news is choosing between prediction vs inference often streamlines workflows and makes writing more logical.
Jo PottsAnother day another example of AIC model selection then interrogating p-values to look at significant effects 😞 (No slight on the authors, a common problem and often journals request—force?— reporting of p-values)