Abstract:
|
Attrition or dropout is a common phenomenon in surveys and questionnaires. One theory is that attrition occurs in three distinct types of phases: a curiosity plateau, a dropout phase, and then stable participation. We generalize this theory to either phases of stable participation or phases of dropout and aim to empirically estimate phase transitions with user-specified thresholds, using a generalized linear mixed model (GLMM), and applying discrete time survival analysis (DTSA). We simulate survey responses for 200 participants in a ten item questionnaire under three attrition patterns. Using these methods, we test the null hypothesis of no distinct attrition phases (linear attrition) against the alternative hypothesis that distinct attrition phases exist.
|