The impact of unlisted and no-landline respondents on non-coverage bias: The Italian case

To survey the general population, survey agencies often use sampling frames of landline numbers. However, these frames may exclude a relevant share of the target population. In 2012, 50% of Italian adults are excluded from the sampling frame, as they are unlisted (UN) or do not own a landline telephone (NT). Using a unique national survey with full-coverage of the Italian population, our study describes the changes in the composition of landline telephone coverage over time and investigates differences in demographic and socio-economic characteristics between (i) respondents included in and excluded from the sampling frame and (ii) NT and UN respondents. It also explores whether these differences lead to under-coverage bias and evaluates the impact of current post-adjustment strategies on the total under-coverage bias as well as on its two components separately.