An R/brms tutorial on using Regularised Prediction and Poststratification for test norming

Background: Sample representativeness and generalisability issues remain neglected in psychology. These issues are doubly relevant for test norming as a) improving the representativeness of norms increases their accuracy (e.g., when using them to calculate IQ scores) and b) representative test norms can in turn be used in research to obtain more generalisable inference (e.g., about effect size). Up-to-date, representative norms are rare in psychology. Solution: We provide an accessible tutorial on applying Multilevel Regression and Poststratification (MRP) and extensions of it (Regularised Regression and Poststratification) using the Stan & R package brms for the purpose of deriving test norms or any descriptive estimates of population entities (e.g., prevalence of psychological disorder). Potential benefits: Adoption of RPP can make representative, openly shareable norms more widely available as RPP can help us derive such norms based on inexpensive online convenience samples. These norms can improve the state of (measurement) standardisation in psychology and beyond.