The fields of randomized experiments and probability sampling are traditionally two separated domains of applied statistics. Both fields share one similarity, which makes them unique from other areas of statistics: design plays a crucial role in this type of empirical research. While design of randomized experiments is traditionally focused on establishing the causality between differences in treatments and observed effects (internal validity), design of probability samples is focused on generalizing results observed in a small sample to an intended target population (external validity). Many experiments conducted with the purpose to improve survey methods are small scaled or conducted with specific groups. The value of empirical research into survey methods is strengthened as conclusions can be generalized to populations larger than the sample that is included in the experiment. This can be achieved by selecting experimental units randomly from a larger target population and naturally leads to randomized experiments embedded in probability samples. This results in experiments that potentially combine the strong internal validity from randomized experiments with the strong external validity of probability sampling. Generalizing conclusions observed in an experiment to a larger target population can be achieved with a design-based inference framework known from sampling theory. In this webinar a general design-based framework for the analysis of single factor and factorial randomized experimental designs embedded in general complex probability samples is presented. Methods are illustrated with real life applications conducted at Statistics Netherlands.
22 September 2022