Allegiance effects in treatment research: A commentary.

Allegiance effects are ubiquitous in treatment outcome research. Luborsky and colleagues (this issue) extend the methods of assessing investigator allegiance and find, once again, that it predicts response in a powerful and robust fashion. They further speculate about the mechanisms that may account for this phenomenon. My own sense is that allegiance effects are most likely to occur (and be most problematic) when investigators compare some preferred modality in which they have both interest and expertise with some other alternative intervention in which they have neither. Rather than reflecting some underlying misrepresentation of the scientific process, I suspect that they more often reflect honest differences in competence in executing the competing approaches. To the extent that this is true, the solution lies not in eliminating allegiances but in balancing them, largely by making sure that colleagues are added to the research team who have the necessary interest and expertise to ensure that each modality tested is adequately implemented.