First, instructions can sacrifice speed for both completeness of the reports and naturalness of the primary process. Note that this may not be compatible with typical experimental instructions to minimize both errors and RT. Second, subjects can be instructed to preserve naturalness over completeness.
common practice to recommend nondirective prompting when subjects lapse into silence. However, because prompts run counter to the emphasis on nonreactivity over completeness,
they should be minimized in most situations. Instead, adequate training in verbalization prior to data collection will usually be preferred.
In spite of the substantial reactivity we have observed and the absence of a fully adequate theory of protocol generation, we do not conclude that concurrent verbal protocols are invalid and should be avoided. All methods risk some invalidity and trade off costs for benefits. On the basis of our own experience with verbal protocols and other process-tracing data (e.g., eye movements and manual responses), we believe that nothing can match the process- ing insights provided by a verbal protocol. Given their unique benefits, the challenge is to identify and reduce causes of their invalidity.
No comments:
Post a Comment