ABSTRACT
Our topic is the planning of research. I’ll discuss what’s called the replicability crisis, then focus on Open Science because that needs attention from the very start of planning. I’ll move on to discuss pilot testing, and the formulation of sampling and analysis plans. Much of the rest of the chapter is about the vital issue of choosing N: Of course, we’d like big N, but there are costs as well as benefits, and sometimes big N is impossible. I’ll take two approaches to choosing N. First, using estimation, we take the precision for planning approach by asking “How large an N do we need to estimate the effect we are studying with a certain precision, say within ±0.2 units of d?”, where Cohen’s d is the effect size measure we’re using. Second, using NHST, we can take the statistical power approach by asking “How large an N do we need to have an 80% chance of achieving statistical significance at the .05 level when we test the null hypothesis of δ = 0 in the population, if the population effect size is really, say, δ = 0.4?”
