Publication type
Survey Futures Working Paper Series
Series Number
26
Series
Survey Futures Working Paper Series
Authors
Publication date
July 31, 2026
Summary:
As many social surveys transition from interviewer-administered to self-completion data collection, there is increasing debate about the most appropriate methods for selecting individuals within households when only address-based sampling frames are available. A pertinent consideration is the extent to which selecting multiple respondents per household influences the precision of survey estimates through its effects on within-household response clustering and sampling variance. For opinion, values, and attitudinal surveys, there is little empirical evidence on the extent of intra-household correlations and their implications for sampling variance and statistical efficiency, limiting evidence-based sample design choices.
This paper addresses this gap using data from the 2021 British Social Attitudes survey. It quantifies the magnitude of within-household response clustering across attitudinal measures, examines its contribution to sampling variance, and compares the statistical implications of selecting one versus two respondents per household. Findings show that although selecting two respondents introduces additional sampling variance through household-level clustering, this is offset by variation in the contribution of unequal selection probabilities to the overall design effect, resulting in modest differences in overall statistical efficiency between the two approaches. These results provide evidence to inform sample design decisions about within household respondent selection in self-completion household surveys using address-based sampling frames.
Subjects
Paper download#589120