Undercoverage bias in telephone surveys in Europe: The Italian case

In Europe, list frame sampling methods are often adopted in computer-assisted telephone interviewing surveys. When sampling frames (i.e., white pages) are incomplete, the quality of the resulting survey data may be seriously jeopardised. In 2012, only 50% of Italians are included in the sampling frame; the remaining are excluded, as they do not own a landline telephone (32%) or their landline telephone number is unlisted (18%). This paper explores differences in the socioeconomic and demographic characteristics of different sub-samples of respondents, investigates the impact of telephone under-coverage on a set of items belonging to different survey domains, and assesses the effectiveness of current weighting procedures. We find that under-coverage leads to bias and that a “one size fits all” weighting strategy can hardly be considered as an accurate method to correct for noncoverage bias