Only available in German

Let’s Talk Money: Reducing Income Nonresponse in Surveys

18 November 2026
Working Paper 285
Katharina Allinger, Melanie Koch
LinkedIn Facebook X E-Mail
Copy link

Summary

Survey participants frequently refuse to answer questions about their income. This is problematic as income is an important variable in social surveys and nonresponse can complicate and potentially bias results. We conduct a randomized experiment in four Central and Eastern European countries. In the survey respondents receive a question on the exact income amount followed by a categorical question if they refuse to answer. The experiment shows that a simplification of the follow-up question – reducing the number of brackets from more than twenty to three – raises response rates substantially.

Highlights

Simpler questions reduce income nonresponse
Our experiment shows that replacing a long list of more than twenty income brackets with only three broad income categories significantly and substantially reduces income nonresponse in a survey of individuals. The treatment is effective for all four countries examined.

Refusal more common than lack of knowledge
In the CESEE countries studied, people who do not answer the question on exact income overwhelmingly give “no answer” rather than saying they “don’t know”. Such refusals might be an indicator of privacy concerns when declaring one’s income, and our treatment effectively reduces nonresponse in this case.

No positive spillover to personal income question
Simpler, less intrusive questions may increase trust in the survey and interviewer and thus potentially reduce nonresponse to subsequent questions. However, our treatment does not improve the respondents’ willingness to answer a subsequent question on exact personal income.