Work

Social norms and the decline in US cigarette smoking: evidence from 30 years of US representative surveys

Year: 2026

Type: article

Abstract: Introduction Changing social norms on cigarette smoking are proposed as a primary strategy to reduce smoking behaviour. Methods Using core questions in 30 years of tobacco use supplements (TUS) to the Current Population Survey (CPS), we develop a state-specific, injunctive social norm scale of Willingness to Restrict Smoking (WTRS) and test its association with smoking behaviours. Respondents (n=1 462 050) to US national and state representative TUS-CPS conducted at nine regular intervals from 1992 to 2022. Results Respondents were queried on their WTRS in 10 different settings, which varied somewhat over time. Psychometric analyses of 10 items supported a single primary dimension that accounted for 74% (SD: 7.3%) of estimated common variance. Using item response theory methods, we calibrated scores across different years and demonstrated consistency of measurement over time (differential test functioning (DTF) across adjacent years = <1% difference in scores). The resulting WTRS scale had good reliability (omega-h mean = 0.75, SD=0.04 and alpha mean =0.81, SD=0.03). Construct and concurrent validity of the scale were shown with regression analyses with outcomes of current cigarette smoking (p<0.001) and presence of home smoking restrictions (p<0.001). An 87% correlation between state-averaged smoking prevalence and WTRS score was observed over the 30-year period. The adjusted OR (AOR) of being a person who smokes for people in states with the lowest versus highest quartiles of states for the WTRS score was 0.74 (95% CI 0.69 to 0.76) which can be compared with the same OR between the lowest and highest quartiles of cigarette prices. (AOR=0.84, 95% CI 0.84 to 0.92). Conclusions This WTRS social norm scale was likely influenced by state tobacco control programmes targeting social norms and appears to be a strong indicator of declining smoking prevalence across 30 years of US data.

Source: BMJ Public Health

DOI: 10.1136/bmjph-2024-002389