Tuesday 14th July      Wednesday 15th July      Thursday 16th July      Friday 17th July     

Wednesday 15th July, 09:00 - 10:30 Room: O-206

Multi-actor surveys
Convenor	Mrs Inge Pasteels (University of Antwerp, Belgium )

Session Details

For this session researchers are invited to submit papers considering a particular survey methodological topic within the context of multi-actor surveys. Recently multi-actor surveys came up as a new type of surveys besides individual and household surveys. In surveys with a multi-actor design, several individuals who are related to each other by precisely defined social ties, are involved. All individuals together with the well-defined social relationships and social roles constitute the multi-actor unit. Several types of multi-actor surveys can be distinguished according to the sampling unit. If the sampling unit corresponds with only one directly selected individual around which the remaining multi-actor pattern will be built, survey data are considered as being “singular multi-actor data”. If several persons are directly sampled, survey data are “multiple multi-actor data”. In case of two directly sampled individuals, we refer to this type of survey data as “dyadic multi-actor data”. Furthermore, several settings can be approached by a multi-actor survey, e.g. families with members living in different households, an educational system with pupils, teachers and parents as the main actors or labor force settings with employers and employees. Papers in this session can deal with a wide range of survey methodological issues as sampling, fieldwork processes and fieldwork monitoring, item and unit nonresponse, weighting and imputation, interviewer effects and interviewer training, mixed mode designs,…. but highlighting specificities of well-known techniques, procedures, terminology,… given the multi-actor survey design has to be the main goal of the paper.

Paper Details

1. The effect of interviewers' motivation and attitudes on respondents' consent to contact secondary respondents in a multi-actor survey
Dr Jette Schröder (GESIS - Leibniz Institute for the Social Sciences)
Dr Claudia Schmiedeberg (Ludwig Maximilian University, Munich)
Dr Laura Castiglioni (Ludwig Maximilian University, Munich)

We use the German Family Panel (pairfam) to investigate the effects of interviewers' characteristics and attitudes on respondents' consent to sending their parents and step-parents a mail questionnaire. During the fifth wave of the panel (2012), an interviewer survey was conducted, covering, among other topics, the interviewers' motivation for their work as well as their attitudes towards certain aspects of the survey process. Using multi-level models, we estimate the magnitude of interviewer effects on consent and investigate in how far differences in the interviewers' motivation and attitudes can explain these effects.

2. Surveying child activities – A case of a multi-actor approach within in the National Educational Panel Study (NEPS)
Dr Thomas Bäumer (LIfBi)
Mr Tobias Linberg (LIfBi)

Children's involvement in ten activities were surveyed from the perspectives of educators on the one hand and of parents on the other hand. Moreover, the same items were used as a measurement of activities offered at the Kindergartens the children attended. Educators rated these activities using identical response scale. With these three different ratings, a variety of relations between multiple actors and/or multiple level of analysis can be analyzed. In the paper, we will discuss these different relations with regard to possible advantages as well as well as challenges for survey design, sampling strategies, data management and analysis.

3. Who’s willing to talk about divorce? Adjustment of nonresponse bias among divorcees using survey data with a multi-actor design.
Mrs Inge Pasteels (University of Antwerp)

We’ll explore how characteristics of both ex-partners of a dissolved marriage and characteristics describing the previous relationship determine the likelihood to participate in a survey about divorce. Especially response behavior on the dyadic level will be considered. Explanatory variables for selective dyadic nonresponse will be used as auxiliary information in order to adjust the nonresponse bias at the dyadic level. We use the data from the “Divorce in Flanders (DiF)” project as an example of a survey in which this innovative multi-actor design was implemented.