Identifying misclassification in youth self-reported smoking status: testing different consent processes of...

In Florida, since 1998, identical survey items have been used to measure youth smoking status for the CDC sponsored state school-based survey and the tobacco control program evaluation telephone survey. The two surveys should parallel one another to track tobacco use. Tobacco items collected in the two surveys closely paralleled one another until recently. Since 2008, data show dramatically divergent youth smoking estimates (e.g., relative differences as high as 50%), which cannot be explained by differences in survey and sampling design. As a first step in detecting misclassification of smoking status, we examined the feasibility of asking youth to self-report their smoking behavior and collect a biological sample, with the expectation that some youth will misreport their smoking status.

Using a cross-sectional population level telephone survey, youth were randomly assigned to one of three groups to test mode effects of collecting biological data with self-reported survey data (n = 303).

It showed two groups of youth (those who are not asked for a biological and those asked for a biological with an indirect explanation of its use) had similar response rates and self-reported smoking status, while the third group (biological request with a direct explanation of its use) had a substantially lower response rate and self-reported smoking status.

The data show youth who are given an indirect explanation of how biological data are to be used were as likely to self-report their smoking status as youth who were not asked to provide a biological sample.