Assessing Drug Attribute Importance

This study examined the convergent validity between importance measures derived from conjoint analysis and from process-tracing techniques under differing levels of situational involvement and task complexity. A randomized design was conducted in which a sample of 48 physicians indicated preferences for hypothetical drugs. The Mouselab program was used to monitor subjects' search order, proportion of information searched, and proportion of time spent. Spearman's coefficient was used to assess the level of agreement between importance measures. Results indicated a low level of agreement between conjoint utility weights and processtracing measures. The correlations were not affected by changes in situational involvement but were affected by changes in task complexity. Higher correlations were found in the highcomplexity task than in the low-complexity task. The low convergent validity may be attributed to the different processes involved in information acquisition and integration.