Importance of Survey Design for Studying the Epidemiology of Emerging Tobacco Product Use Among Youth.

Abstract

Accurate surveillance is critical for monitoring the epidemiology of emerging tobacco products in the United States, and survey science suggests that survey response format can impact prevalence estimates. We utilized data from the 2014 New Jersey Youth Tobacco Survey (n=3,909) to compare estimates of the prevalence of 4 behaviors (ever hookah use, current hookah use, ever e-cigarette use, and current e-cigarette use) among New Jersey high school students, as assessed using "check-all-that-apply" questions, with estimates measured by means of "forced-choice" questions. Measurement discrepancies were apparent for all 4 outcomes, with the forced-choice questions yielding prevalence estimates approximately twice those of the check-all-that-apply questions, and agreement was fair to moderate. The sensitivity of the check-all-that-apply questions, treating the forced-choice format as the "gold standard," ranged from 38.1% (current hookah use) to 58.3% (ever e-cigarette use), indicating substantial false-negative rates. These findings highlight the impact of question response format on prevalence estimates of emerging tobacco products among youth and suggest that estimates generated by means of check-all-that-apply questions may be biased downward. Alternative survey designs should be considered to avoid check-all-that-apply response formats, and researchers should use caution when interpreting tobacco use data obtained from check-all-that-apply formats.