Regency Ballroom-Monroe
Response Burden: What Predicts it, and Who Is Burdened Out? (303114)
Scott Fricker, Bureau of Labor StatisticsShirley Tsai, Bureau of Labor Statistics
*Ting Yan, Westat
Keywords: response burden, perception of survey, task difficulty, motivation, survey request
Concerns about the burden surveys place on respondents have a long history in the survey field. A review of the burden literature reveals burden is measured in many ways. Too often, researchers employ as a proxy measure of burden properties of surveys/tasks that are believed to impose response burden, respondents’ attitudes and beliefs toward surveys, and respondents’ behaviors such as willingness to be interviewed again. These very different measurements of response burden reflect both the lack of and need for a well-developed conceptual framework on burden.
In this paper, we posit a path model that explicitly looks into the direct and indirect effects of survey features, respondent characteristics, and respondents’ perceptions of the survey on burden in hopes of shedding light on which factors (or combination of factors) are most likely to result in response burden.
Data from the Consumer Expenditure Quarterly Interview Survey were analyzed using structural equation modeling. We found that low motivation, difficult task, and negative impressions of survey all lead to higher level of perceived burden. Contrary to views commonly held in the survey field, the usual-suspect causes of burden such as challenging survey request had no significant overall effects on burden. We also found that some relationships between factors and their indicators are not equal across respondents who were attempted in different modes. After releasing constraints on those relationships to achieve partial metric invariance, we found equivalence of relationships among latent factors across modes of data collection.
Our findings support the notion of burden as a subjective multidimensional phenomenon and have important practical implications. Factors found to be related to burden and their indicators should be closely examined during questionnaire development and evaluation so as to produce survey questions minimizing response burden.