Survey modifications could impact respondent burden. Increased burden could potentially lead to refusals in the following waves or inadequate answers to questions, which could in turn induce bias affecting overall data quality. Census has studied survey participant impressions of data security and privacy. Burden measurement would allow us to identify where interventions may be needed to offset the impact of respondents’ perception of burden and mitigate burden-induced bias on data quality. During the 2012 and 2017 Consumer Expenditure Surveys (CE) Quarterly Interview, answers on perceived burden were collected at the end of the final interview wave. In this presentation, we will show constructing a composite burden index score using a multivariate technique , and investigating the performance of a single burden question compared to a composite burden index using indirect indicators of data quality such as the number of unknowns and refusals. We also studied respondent burden proxy indicators by using the nonparametric recursive partitioning model under a complex survey design. We will also present the results from the 2017 CE newly revised respondent burden questions.