This paper investigates whether answer behavior that points at increased risk of measurement error is consistent across surveys, i.e. whether it is a trait of respondents.
Measurement errors are a common problem in data collection, both in surveys and in administrative data, These errors follow from deficiencies in the answer or registration process. Usually, measurement errors cannot be measured directly, but clues can be found in answer behavior. An important question is whether deviant behavior is circumstantial or a trait of certain respondents. Insight into such potential traits is essential in design of instruments.
