Dance p 3 Mar09

档案导读

Geoff Cumming uses a simulation to show how wildly the p value varies from one replication to the next because of sampling variability. He argues for estimation and confidence intervals over null hypothesis significance testing.