How Interviewer Effects Differ in Real and Falsified Survey Data: Using Multilevel Analysis to Identify Interviewer Falsifications

Uta Landrock


In face-to-face interviews, interviewers can have an important positive influence on the quality of survey data, but they can also introduce interviewer effects. What is even more problematic is that interviewers may decide to falsify all or parts of interviews. The question that the present article seeks to answer is whether the interviewer effects found in falsified data are similar to those found in real data, or whether interviewer effects are larger and more diverse in falsified data and may thus be used as an indicator for data contamination by interviewer falsifications. To investigate this question, experimental data were used from controlled real interviews, interviews falsified by the same interviewers, and questionnaires completed by these interviewers themselves as respondents. Intraclass correlations and multilevel regression models were applied, and interviewer effects in the real survey data were compared with those in the falsified data. No evidence of interviewer effects was found in the real data. By contrast, interviewer effects were found in the falsified data. In particular, there was a significant association between the interviewers’ own responses and the falsified responses to the same questions in the questionnaire. Thus, to detect interviewer falsifications, I recommend that researchers should also get the interviewers to complete the questionnaire and check datasets or suspicious cases for interviewer effects.


interviewer, interviewer effects, interviewer falsifications, data quality, identification of falsifications, multilevel analysis

Full Text:




  • There are currently no refbacks.

Copyright (c) 2017 Uta Landrock

Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.