When more of a good thing is better: reply to Hayes et al. (1996)

More of a good process component is better when clients are not already getting enough. Process components may be in short supply because of limited resources or ignorance, because they are evaluations themselves, or because they are difficult-to-attain subgoals on the way to valued outcomes. Levels of crucially important process components that are not in short supply (including most commonly used therapeutic interventions) are unlikely to be statistical predictors of outcome. Conversely, then, null correlations with outcome do not impugn a component's therapeutic importance. Many of A.M. Hayes, L.G. Castonguay, and M.R. Goldfried's (1996b) examples of positive correlations usefully illustrated clients' cognitive and emotional processes that were treatment subgoals or evaluative indexes. However, by focusing on the (relatively rare) positive correlations, they overlooked the more common null or inconsistent correlations of theoretically important process components with outcome, which are the central topic of the responsiveness critique.