Conferences

In a recent paper, Durrant, Staetsky and Steele (2008) notes that although survey experts have long recognized |the importance of the interviewers in considering surveys errors, little is known about the effects of interviewer characteristics on survey response (or nonresponse). They developed a multilevel logistic cross-classified model that allowed for testing both the effects of household characteristics and interviewer characteristics, as well as distinguishing between interviewer and area effects. This paper follows their lead and focus on the interplay between demographic characteristics of interviewer and of the potential respondents in the Norwegian sample of round 3 of the European Social Survey.

The study is based on the contact data from the round 3 combined with register data on the gross selected sample, and some interviewer characteristics from the administrative records of Statistics Norway. The respondents were selected from the population register by means of simple random sampling. Combined with register data, this allows as to improve upon the analysis of Durrant et al. by using characteristics of the individually sampled persons (sex, age, and education) rather than household characteristics. The information on the interviewers is, however, restricted to a few demographic variables (sex, age, education, and experience) available from the register of Statistics Norway. However, from the contact data, we will develop measures of contact patters that may vary among the interviewers.