Abstract

Strengthening collaborative partnerships between National Statistical Offices (NSOs) and society requires a comprehensive understanding of response burden. While traditional methodology focuses on interview duration as a proxy for respondent burden, this creates a “blind spot” regarding the social cost imposed on different population groups and various cultures.
To address this gap, this study conducts a multidimensional analysis on response burden. Using data from the National Socio-Economic Survey (Susenas), this research integrates paradata (enumeration length, number of visits, and time of interview), respondent characteristics, and direct measurements of perceived respondent burden. Furthermore, this study deepens the analysis by incorporating cultural timing factors, such as prayer periods, domestic activity peaks, and rest times, to examine how survey timing interacts with household circumstances. This study establishes some models and utilizes a range of statistics (frequentist and Bayesian) and machine learning methods. We compared the performances of different models and methods to show the validity of the findings.
The preliminary findings indicate that "burden response" is highly context-dependent. Interview duration alone does not fully capture response burden, and cultural timing and respondent characteristics contribute to differential burden experiences, as well as differences between urban and rural. In different groups of the population, response burden is driven by different types of burden experiences. Based on the result, this study proposes Adaptive Survey Logistics to maintain a symbiotic partnership.
This study contributes to developing methodology on measuring and modelling response burden, providing a broader understanding of response burden, and managing response burden to build strong, collaborative partnerships to achieve shared goals between NSOs and society.