Do Different Respondent Selection Methods Produce Different Survey Estimates? A Simulation of Within-Household Selection Methods for 2013 Turkey DHS

Saraç M., Koç İ.

Pamukkale University Journal of Social Sciences Institute, no.51, pp.257-274, 2022 (Peer-Reviewed Journal)


Respondent selection procedures are utilized to choose a particular respondent in surveys. In the 2013 Turkey Demographic and Health Survey (2013 Turkey DHS), all women who meet the eligibility criteria were interviewed according to standard regulations of the International DHS Program. Simulating the application of different methods, this study mainly asks “How would main indicators produced in 2013 Turkey DHS differ with the implementation of various respondent choice strategies?”. The selection procedures include two probabilistic methods, two quasi-probabilistic methods, and four non-probabilistic methods. This study advanced that although interviewing all eligible women in 2013 Turkey DHS appears to be reasonable, providing estimates that are close to survey results, is also conceivable through the execution of selection methods. When the mean of total deviations is considered with their confidence intervals, probabilistic and quasi-probabilistic selection techniques have produced much better estimates with minimal deviations from the confidence intervals of the population values.