In the last decade, many institutions from around the world have come to rely considerably upon laboratories employing cognitive interviewing techniques to evaluate survey questions and questionnaires (see, e.g., DeMaio et al., 1996; Akkerboom and Dehue, 1997; Gower and Haarasma, 1997). Two of the most widely described cognitive interviewing techniques are the concurrent and retrospective interviews (see, e.g., Ericcson and Simon, 1980; Forsyth and Lessler, 1991 ). In concurrent interviews, subjects are asked to verbalize the process they go through to answer a question as they progress through a questionnaire. In retrospective interviews, subjects are asked at the end of the interview to provide thought processes to questions they answered earlier in the questionnaire. Ericcson and Simon (1980 and 1984) argue that the concurrent technique is not advisable under certain conditions. One example they give is when the task requires subjects to verbalize visual information. With the recent advent of the significance of the visual component of self-administered questionnaires (Jenkins and Dillman, 1995 and 1997 ), it follows that the concurrent technique, at least by itself, may not be as ideal a choice for evaluating self-administered questionnaires as it is for interviewer-administered questionnaires. In this paper, we describe the preliminary results of a small-scale experiment designed to evaluate the concurrent and retrospective techniques with three self-administered decennial short forms. We conclude with a discussion of the results and their implications for the future.
