Predictability of reasons for refusal in telephone surveys

Refusals are one of the key problems in surveys. Data from panel surveys can help to predict reasons for refusal, as information on respondents is available from previous waves. If specific reasons can be anticipated measures can be taken to cope with them such as interviewer tailoring for those participants with a higher likelihood to use these reasons. This strategy implies that the reason which specific respondents are likely to use can be predicted. We study effects on the refusal reason that was given by the respondents for the first time in the Swiss Household Panel from different domains, including sociodemographic features, social inclusion aspects, answer quality, and interviewer assessment of respondent behavior, question understanding, and future participation. We find that ‘No interest’ is uttered more frequently by young respondents with lower education and little interest in politics, and by those exhibiting a high variation of answers on subjective questions and those rated particularly difficult by interviewers. ‘No time’ reasons are more likely mentioned by the employed; ‘Age’ or ‘health reasons’ more often by older people, and generally by people with bad health. ‘Not willing to fix a date and time for an appointment’ is preferred by young foreigners not speaking one of the national languages as a mother tongue, the socially inactive, and especially those who tend to satisfice. ‘Other reasons’ apply to those with a smaller likelihood to continue to participate. ‘Family reasons’ are more difficult to assess; also family reasons are the only ones which are not mentioned during an early wave.