This project concerns research synthesis, or more specifically: Bayesian evidence synthesis (BES). In this method, the results of independent analyses serve to evaluate the relative support for a fixed set of competing informative hypotheses. For example, with three means A, B and C, we could have the following informative hypotheses:
- H1: Mean A > Mean B > Mean C versus
- H2: Mean A < Mean B < Mean C versus
- H3: Mean A = Mean B = Mean C.
If we measure the same construct, we can expect the ordering of means to be independent of the measurement instrument at hand. Hence, we do not need to harmonize independent datasets as is required for other aggregation methods. By aggregating the relative support for a set of hypotheses over multiple independent datasets, we can draw conclusions about the hypothesis best supported by all datasets simultaneously. The resulting robust conclusions are essential for the progress of science.
