Improving psychological science

Psychological science recently faced a reckoning: Researchers slowly began to realize that the way in which they did research was not particularly well suited to discover new knowledge. Psychological theories are often weak and non-specific, our measurements are unreliable and invalid, and researchers often analysed data until they found the desired result (a practice known as p-hacking). In this research line, I contribute to an extensive literature suggesting solutions to this ever growing number of problems. My focus here is on data pre-processing and multiverse analyses. Through scientific research demonstrating that data pre-processing decisions massively influence statistical results and theoretical conclusions (Primbs, Holland, Quandt, & Bijlstra, in prep.), a Shiny app for teaching, and conference and invited talks I try to highlight that all of the small decisions researchers make when analysing their data can impact the conclusions they draw from their experiments. In other work, I caution readers against misinterpreting effect sizes (Primbs et al., 2023a), argue for the need for consensus meetings to establish field-specific smallest effect sizes of interest (Primbs et al., 2023b) and contributed to work demonstrating the careless use of theories in psychological science (McPhetres et al., 2021). I also regularly contribute to large team science projects and serve as Assistant Director for Translation and Cultural Diversity at the PSychological Science Accelerator.