Zürich - Mainz - Leipzig

Asking sensitive questions: Possibilities and limits of randomized response and other techniques in different survey modes

University of Leipzig (D), University of Mainz (D) and ETH Zürich (CH)

Management

Prof. Dr. Andreas Diekmann (Zürich)
Prof. Dr. Karl-Dieter Opp (Leipzig)
Prof. Dr. Peter Preisendörfer (Mainz)
Prof. Dr. Thomas Voss (Leipzig)

Researchers

Marc Höglinger (Zürich)
Dr. Ivar Krumpal (Leipzig)
M.A. Felix Wolter (Mainz)

Scheduled duration

2008 - 2012

Research project

Many surveys include so-called "sensitive questions", i.e. questions that pertain to private or illegal behavior or socially frowned-upon attitudes. Respondents misreport on sensitive topics, such as drug use, delinquency, sexual activities and racial attitudes, in order to avoid painful feelings of embarrassment in the interview situation or repercussions from third parties beyond the interview setting. As a consequence of underreporting socially undesirable activities (and attitudes), survey estimates are often erroneous, more specifically distorted by bias. Former research shows that features of the survey design (e.g. question technique, question mode) may have an impact on the level of measurement error generated by socially desirable answers.

The goal of the project is to develop, compare and evaluate different techniques and strategies of asking sensitive questions in three different survey modes: face-to-face, telephone and online-surveys.

During the first two project years, we compared different variants of the randomized response technique (RRT) with other question techniques. We evaluated the validity of the data obtained with the different technique/mode-combinations in terms of a reduction of non-response and response bias. One challenge was to develop appropriate RRT formats for each survey mode. (1) The subproject in Zürich (CH) developed and tested suitable implementations for online surveys. (2) The subproject in Leipzig (D) conducted an experimental survey in order to evaluate possibilities and limits of RRT in the telephone survey mode. (3) The subproject in Mainz (D) conducted a validation study, comparing data from face-to-face interviews with administrative records.

During the second funding period (2010 - 2011), we intend to continue our research along four main lines: (1) Theoretical aspects of the interviewer-respondent interaction and statistical considerations regarding the trade-off between anonymity and statistical power are to be worked out to further refine and test different question techniques. This also includes the development of software for the analysis of interviewer (context) effects in RRT-data (Zürich). (2) The face-to-face study and external validation are to be continued and extended (Mainz). (3) Mode effects are to be investigated by means of a sensitive-questions survey using the access panel developed by the research group lead by Uwe Engel (also part of the DFG priority programme 1292) (Leipzig, Mainz, Zürich). (4) An experimental survey investigates the viability of the Item Count Technique (ICT) in the telephone survey mode (Leipzig).

Cooperation

Access Panel and Mixed-Mode Internet Survey
Prof. Dr. Uwe Engel (Bremen)
Prof. Joop Hox, PhD (Utrecht)
Prof. Edith de Leeuw, PhD (Utrecht)
Annette Scherpenzeel, PhD (Tilburg)

Cooperation beyond the Programme

Project specific cooperation (persons involved in the project without financial support from the DFG):
Prof. Dr. Ben Jann (University of Bern)
Dr. Wolfgang Langer (University of Halle-Wittenberg)
M.Sc. Heiko Rauhut (ETH Zürich)
Dr. Juergen Schiener (University of Mainz)