The impact of nonresponse in different survey stages on the precision of prevalence estimates for multi-stage survey studies

Abstract

Objective: While it is known that nonresponse might produce biased results and impair the precision of results in survey research studies, the pattern of the impact on the precision of estimates due to the nonresponse in different survey stages is historically overlooked. Having this type of information is essential when creating recruitment plans. This study proposes to examine and compare the effect of nonresponse in different stages on the precision of prevalence estimates in multi-stage survey studies. Based on data from a state level survey, a simulation approach was used to generate datasets with different nonresponse rates in three stages. The margin of error was then compared between the datasets with nonresponse at three different survey stages for 12 outcomes. Results: At the same nonresponse rate, the mean margin of error was greater for the data with nonresponse at higher stages. Additionally, as the nonresponse rate increased, precision was more inflated within the data with higher stage nonresponse. This suggests that the effort used to recruit the primary sampling units is more crucial to improve the precision of estimates in multi-stage survey studies.