Improving adolescent tobacco surveillance: Prevalence and correlates of aberrant responding in the National Youth Tobacco Survey.

Abstract

Youth tobacco surveillance depends on the National Youth Tobacco Survey (NYTS), yet self-reported data are vulnerable to aberrant responding (careless or inconsistent responses). This study identifies atypical response patterns in NYTS data and evaluates their implications for harm perception estimates. We analyzed 2017-2019 NYTS data collected in the United States (N = 52,406) using four person-fit statistics (lzpoly, gpoly, gnormed.poly, U3poly) to assess the plausibility of response patterns on harm perception items. Cutoffs were generated through bootstrap resampling. Respondents flagged by two or more indices were classified as aberrant. Weighted multivariable logistic regression examined demographic and behavioral correlates. Aberrant responses were identified in 11-12% of adolescents. Higher odds were observed among youth who currently smoked (OR = 1.52, 95% CI = 1.30, 1.78), used e-cigarettes (OR = 1.41, 95% CI = 1.25, 1.59), or had ever tried cigarettes (OR = 1.26, 95% CI = 1.13, 1.41). Males (OR = 1.30, 95% CI = 1.21, 1.40) and Black adolescents (OR = 2.05, 95% CI = 1.84, 2.28) also showed higher odds. Including aberrant responses attenuated population estimates of perceived harm, resulting in modest downward bias. Approximately one in ten NYTS respondents exhibit aberrant response patterns that can subtly bias harm perception estimates. Person-fit analysis offers a practical diagnostic tool to enhance the validity of youth tobacco surveillance.