Analyzing the Association of Objective Burden Measures to Perceived Burden with Regression Trees

Abstract Higher levels of perceived burden by respondents can lead to ambiguous responses a questionnaire, item nonresponse, or refusals continue participation in the survey which introduce bias and downgrade quality data. Therefore, it is important understand what might influence perception respondents. In this article, we demonstrate, using U.S. Consumer Expenditure Survey data, how regression tree models be used analyze associations between objective measures conditioning on household demographics other explanatory variables. The structure allows these easily explored. Our analysis shows relationship some after different demographic variables that relationships are quite affected respondent characteristics mode survey. Since were constructed an algorithm accounts for sample design, inferences from made about population. any insights could help guide future decisions design data collection reduce burden.