What kind of assumptions need to be realistic and how to test them: A response to Tsang (2006)

Tsang (2006) contends that certain core assumptions of a theory, which are typically about people's behaviors or thoughts, need be realistic, because they determine the viability of the mechanism that generates a hypothesized relationship. While Tsang's (2006) article rightly emphasizes the importance of realistic assumptions, it neglects the issue that certain kinds of assumptions are necessarily unrealistic for the roles that they play in theory development and testing. Therefore, researchers should not be criticized for making unrealistic assumptions of the latter kinds. Furthermore, by deliberating on the assumptions underpinning a theory, researchers can construct theories with better explanatory power and further develop existing theories. Tsang (2006) also suggests two approaches for testing assumptions, namely, a structural model and direct inquiry approaches. Although these approaches have certain merits, they also have limitations that may render the evidence gathered unreliable under certain situations. Two alternative approaches, namely, the experimental‐causal‐chain and the moderator‐of‐process designs, address these limitations. The researcher could consider adopting these designs as well in order to improve the rigor of assumption testing. Copyright © 2010 John Wiley & Sons, Ltd.