“It also makes clear why an observational study needs to collect lots of ancillary information about each participant so that the kind of balancing required can be attempted. In a true experiment, with random assignment, such information is (in theory)… — Howard Wainer Copy Share Image
