More and more we’re hearing about the importance of effect size in statistical analysis This was basically ignored in my PhD stats program (although truth be told, we were too busy flipping coins to predict heads or tails and learning how to program in SAS–no pulldown menus for this gal!).
The Data Colada blog today has a great post on this very topic. Author Uri Simonsohn outlines the challenges of using both P and Bayesian methods in Hypothesis testing. He includes commentary from other smart people. It is a good read, and I’m going to spend more time with it this afternoon.
At the same time, I’m reading the book “Big Data” and the suggestion is that in our Big Data future, we won’t even have to do any stats analysis because n=All and all we need to know will be how to do correlations. So there’s that.