When not to analyze data: Decision making on missing responses in dual scaling

In exploratory data analysis, handling of missing responses requires special considerations because of the absence of a model for them. This study investigates effects of missing responses on dual scaling results, in particular a practical decision rule when to and not to impute for missing responses. More specifically, three questions are asked: when to discard a subject, when to exclude a variable, and when to stop data analysis entirely. The decision rule is based on the concept of substantial discrepancy in the scaling outcome between the ‘best’ and ‘worst’ imputations for missing responses. This problem is discussed in the context of the current knowledge on missing responses in data analysis.