An empirical validation of the Quadruple Process Model of implicit attitudes against alternate, theoretically defensible specifications.

The Quadruple process (Quad) model is a multinomial processing tree that specifies the joint contribution of four qualitatively distinct cognitive processes to responses on implicit measures. The way in which these processes interact to drive responses was initially specified according to theory, and the construct validity of this specification of the model has been demonstrated across a wide variety of studies. However, there are other theoretically-defensible ways in which these processes might interact. The purpose of the present research was to compare the standard version of the Quad model against alternate specifications in order to determine which model best fits data from the Implicit Association Test. Three different versions of the Quad model were applied to very large samples of real participants’ data across three content domains: racial attitudes, sexual orientation attitudes, and gender stereotypes. The standard model provided best fit for racial attitudes and gender stereotype data. However, other versions of the model provided equivalent fit to sexual orientation attitudes data. Taken together, these analyses indicate that the standard version of the Quad model provides best fit to data from the Implicit Association Test in general, but that alternate specifications may be appropriate for some content domains and participant populations.