How do researchers account for underreporting in surveys?
Advancing measurement and analysis of sensitive behaviors
Challenge
Studies have shown consistently that survey respondents overreport socially desirable behaviors and underreport socially undesirable behaviors.
It is not clear, however, who is most prone to underreporting a specific socially undesirable behavior and whether the same people underreport to multiple socially undesirable behaviors.
Solution
We first used latent class models to identify respondents to a national study as underreporters of a socially undesirable behavior.
We then used statistical models to compare them to those classified as truthful reporters of the same socially undesirable behavior in terms of demographic characteristics, health behaviors, and attitudes.
The Results
Through this analysis, we identified specific characteristics of respondents who underreported socially undesirable behaviors. The data showed that different people underreported to different socially undesirable behaviors.
The results of these analyses help to expand the knowledge base and inform policymakers about sensitive behaviors as they work to develop programs to enhance health and reduce illness.