Impact of survey design features on prevalence estimates of sensitive characteristics

点击翻译按钮获取中文摘要

Due to many sources of sampling and non-sampling error, survey estimates may deviate from the population parameter. The impact of nonsampling errors is usually more considerable in surveys of sensitive characteristics. Using prevalence estimates of LGBTIQ populations, a systematic meta-analysis after an extensive literature search following the PRISMA guidelines of published survey estimates will be conducted. The effects of survey design features on the resulting estimates will be studied with a meta-regression. Many different design options will be coded, for example, question form, position in the questionnaire, presence of an interviewer, type of sampling, nonresponse and sponsor. The project will inform the interpretation of prevalence estimates of sensitive characteristics in general and will yield recommendations for the design of future surveys.