Five different “Cohen’s d” statistics for within-subject designs

Jeff Rouder poses an “effect size puzzler” where the puzzle is simply to compute a standardized effect size for a simulated dataset where subjects make 50 responses in each of 2 conditions. He offers a sweatshirt prize (???) to anyone who can do so “correctly.” (Update: Jeff posted his answer to the effect size puzzler here.) … Continue reading Five different “Cohen’s d” statistics for within-subject designs