As is clear from the readings, to reasoning about statistical inference for post-stratified/matched designs requires that we can reason about statistical inference for pre-stratified/blocked randomized experiments. Today we mostly spend our time understanding the simplest form of statistical inference for randomized studies — Fisher’s randomization inference for tests of the sharp null hypothesis of no effects. Here is the pdf version of the classwork for today with comments . Here is the Sweave source code of the classwork for today .