2019-2020 NHIS data


For the longitudinal NHIS data of 2019-2020, could there be non-random attrition problem in 2020 data because of COVID-19 even with the weighting? For example, people who were more affected psychologically by COVID-19 were less likely to participate in 2020 survey, which may not be considered in weighting…?


Yes, this is always a concern with sampled survey data. Although the weights are adjusted to try to account for some sources of nonrandom attrition, there are always going to be some unobserved factors that cause the sample to deviate from true representativeness of the population. I’ll refer you to the IPUMS NHIS User Note on COVID 19 and the linked references therein for more information.

1 Like