Methodological problems when determining verbal protocol accuracy empirically

The approach normally followed to provide an empirical assessment of verbal protocol accuracy is evaluated from a methodological point of view . Three classes of problems are identified with this approach . These pertain to : selecting and eliciting recordable behaviors, converting the observations into data and interpreting the data . Possible ways to deal with the listed problems are presented . It is emphasized that, when adopting the approach, verbal protocol accuracy can only be inferred rather than directly observed . 'Bxtended version of a paper presented at the Second Workshop of Cognitive Processes in Complex Tasks, Eindhoven, 'the Netherlands, 23-25 April 1991 . Portions of this paper were first published in the author's Ph .D. thesis "The analysis of fault diagnosis tasks: Do verbal reports speak for themselves?" (1990) . 1