Examining differences in cigarette smoking prevalence among young adults across national surveillance surveys

Peter Messeri, Jennifer Cantrell, Paul Mowery, Morgane Bennett, Elizabeth Hair, Donna Vallone

Research output: Contribution to journalArticlepeer-review


Accurate smoking prevalence data is critical for monitoring, surveillance, and evaluation. However, estimates of prevalence vary across surveys due to various factors. This study examines smoking prevalence estimates for 18–21 year olds across six U.S. national telephone, online and in-person surveys for the years 2013 and 2014. Estimates of ever smoking ranged from 35% to 55%. Current smoking ranged from 16% to 30%. Across the three modalities, household surveys were found to yield the highest estimates of smoking prevalence among 18 to 21 year olds while online surveys yielded the lowest estimates, and this was consistent when stratifying by gender and race/ethnicity. Assessments of the joint effect of gender, race/ethnicity, educational attainment and survey mode indicated that the relative differences in the likelihood of smoking were consistent across modes for gender and education groups. However, the relative likelihood of smoking among minority groups compared with non-Hispanic Whites varied across modes. Gender and racial/ethnic distributions for most surveys significantly differed from the U.S. Census. Over and underrepresentation of certain demographic subpopulations, variations in survey question wording, and social desirability effects may explain modality differences in smoking estimates observed in this study. Further research is needed to evaluate the effect of survey mode on variation in smoking prevalence estimates across national surveys, particularly for young adult populations.

Original languageEnglish (US)
Article numbere0225312
JournalPloS one
Issue number12
StatePublished - Dec 1 2019

ASJC Scopus subject areas

  • General


Dive into the research topics of 'Examining differences in cigarette smoking prevalence among young adults across national surveillance surveys'. Together they form a unique fingerprint.

Cite this