Premium
Why successful replications across contexts and Operationalizations might not be good for theory building or testing
Author(s) -
Trafimow David
Publication year - 2019
Publication title -
journal for the theory of social behaviour
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.615
H-Index - 51
eISSN - 1468-5914
pISSN - 0021-8308
DOI - 10.1111/jtsb.12211
Subject(s) - replication (statistics) , context (archaeology) , randomness , psychology , relevance (law) , universe , epistemology , cognitive psychology , social psychology , computer science , mathematics , statistics , paleontology , philosophy , physics , political science , astrophysics , law , biology
Despite wide acceptance that replications across operationalizations and contexts is positive for psychology, the present goal is to suggest a more nuanced view that sometimes successful replications can be to the detriment of progress in psychology. The present manuscript distinguishes between replication in two idealized universes. In one such universe, the only difference between original and replication studies is randomness. In the other such universe, there are systematic differences between original and replication studies. Of course, replication in the real universe is subject to both randomness and systematicity. Although it is desirable to maximize, to the extent feasible, the probability of replication in the idealized universe where only random effects differ between original and replication studies; this may or may not be so with respect to systematicity. An important factor that determines the desirability of replications in the context of systematic differences pertains to auxiliary assumptions. A famous crisis in social psychology, with respect to the influence of attitudes on behaviors, illustrates the relevance of auxiliary assumptions in the replication context where there is serious consideration of systematic likeness or unlikeness between original and replication studies.