How transportability analysis can be useful for cumulative theory testing in management research

Abstract

Moving management science forward by publishing research aimed at theory testing and refinement is the mission of the Journal of Management Scientific Reports. One way to test and refine a theory is through constructive replication studies. If X→Y (the causal effect of X on Y) is found in an initial study, what assumptions and mechanisms are required to expect that X→Y would also be found in a subsequent study, when the population differs between the initial and subsequent studies? What structural discrepancies between source and target populations would render transporting the knowledge obtained from one study to another impossible? Whereas generalizability is concerned with making inferences based on a possibly biased sample drawn from a focal population about the focal population, transportability refers to the conditions under which a transfer of causal knowledge across populations is valid. Researchers now have access to methods for identifying the conditions under which causal information learned from experiments would be valid in studies where ideal experiments cannot be conducted and where only passive observations can be collected. In this commentary, I explain how transportability analysis and its associated diagnostic tools from machine learning and artificial intelligence can be useful for cumulative theory testing in management research.