| Abstract: | When analyzing superpopulation data, inferential statistical methods should be used. Empirical tests of hypotheses are subject to a variety of stochastic processes, or “errors”, even if the data involved in those tests are not the product of random sampling. Assumptions about the nature of these processes have to be an explicit part of the analysis and need to be justified. Using several examples of published research, we make transparent that all substantive conclusions are conditional on the assumptions about the nature of the stochastic processes that are at work in generating superpopulation data. |
