Experimental analyses of the performance of verbal subjects often include verbal reports, obtained during post-session interviews, about within-session covert verbal behavior (e.g., hypotheses about the contingencies). But such post-session reports are not necessarily accurate, and procedural details of how the samples were obtained are typically inadequate. Even when the post-session reports are accurate, the within-session hypotheses do not have the status of causes of within-session nonverbal performance. In an experimental analysis, it is important to treat such reports as instances, not causes, of behavior.
