My dissertation work examined the theoretical foundation of “social” priming and its implications for the replicability of research in this area. A PDF of the full paper is available here. And here is the abstract:
Social priming has been a fundamental methodological practice in experimental social psychology for decades; however recent failures to replicate priming studies have cast some doubt on the robustness of subtle primes’ influence on social behavior. The present research attempts to resolve this uncertainty by advancing the perspective that the robustness of social priming effects can be predicted a priori from the strength of the concept association under study. Across eight studies, reported without publication bias, I found that the observed strength of concept pair associations (Studies 1 and P1) from the priming literature predicted my capacity to observe (Studies 2, P2, and P5) or not (Studies 2, P3, P4, and P5) social priming effects. An attempt to strengthen a concept association and test the causal role of association strength (Study 3) resulted in no difference in concept associations and no priming. Overall, I find that social priming effects occur only for the most strongly related concept pairs and that easily measured explicit associations are useful for predicting these effects. I conclude that estimating the strength of underlying concept associations is critical for planning, interpreting, and reporting priming studies.
The featured image above is a summary of the effect sizes from all of the studies in the dissertation, and the scatterplot below represents a different take on the same data. The central take-away is that priming effects only seem to occur for really strongly associated concept pairs. And even then, with common recall-based manipulations, they are fairly small.