Detecting Fraudulent Interviewers by Improved Clustering Methods – The Case of Falsifications of Answers to Parts of a Questionnaire

What is it about?

Fabricated interviews contaminate survey data and might result in wrong conclusions for relevant issues. A method is presented which identifies suspicious cases based on the collected data. It exploits that real and fabricated data differ in some properties which are not related to the content of the questionnaire. It is discussed how well the method works if interviewers fabricate only a part of the data in an interview.

Featured Image

Why is it important?

Using fabricated data might render worthless any result based on survey data. Given the high impact such results might have on decisions in politics, economics, health etc., it is of outmost importance to ensure high data quality. Applying a data driven screening method as the one presented in this paper might be one instrument to be used to this end.