A Path Model Analysis of the Associations Among Individual and Social Factors and Youth Smoking

Background & Purpose. Most adults who become regular (i.e., daily) smokers begin smoking during adolescence. Therefore, understanding the factors that affect youth smoking could assist in the creation of effective smoking cessation/prevention interventions resulting in fewer regular adult smokers. Studies of adolescents have found that individual attributes such as age and gender and social factors such as associating with peers who smoke, affect smoking rates. However, to our knowledge, these factors have not been tested conjointly to determine the pathways by which they influence smoking and whether their effects might be mediated by smoking attitudes. This study tested a path model using generalized SEM (GSEM) to assess if the effects of individual and social factors on youth smoking are mediated by smoking attitudes.

Methods. A secondary analysis of baseline data collected for the National Youth Smoking Cessation Survey (NYSCS) data set was conducted. The NYSCS collected national data on young adult (16 to 24) smoking patterns (N = 2,421) via telephone interviews conducted between 2003 and 2005. Eligibility criteria included having smoked 20+ cigarettes lifetime and 1+ cigarettes in the past month. Independent variables included implicit (i.e., number of smokers lived with, number of friends supporting quitting smoking) and explicit (‘allowed to smoke at residence’; ‘others allowed to smoke at residence’) social influences. Individual factors included demographics (i.e., sex, race, education level, age). Smoking attitudes ranged from 8 to 24 based on a 3-point scale include in the NYSCS survey.  Smoking status was assessed at the ordinal level ranging from 0 (no days smoked past 30) to 6 (daily smoking). GSEM was used to test the full path model with smoking attitude as the mediator to test for potential indirect as well as direct effects of social and individual factors on smoking status, the dependent variable.

Results. Initial analyses revealed non-significant mediating effects. Consequently, the path model simplified to an ordinal logistic regression testing only direct effects. Marginal effects for this model revealed the strongest significant differences between daily smokers and all others, warranting further simplification to a binary logistic regression that yielded the following results: Being a daily smoker in adolescence/young adulthood was associated with higher odds of the number of smokers lived with (OR=1.24, p < .001) and being 18-20 (OR=2.40, p < .001) or 21-24 years old (OR=2.84, p < .001). Those not allowed to smoke at home (OR=.36, p < .001), African-American/Blacks (OR=.59, p < .001) or Hispanics (OR=.43, p < .001), and those with a college education (OR=.40, p < .001) had lower odds of daily smoking. Others smoking at residence and friends supportive of quitting as well as gender and individual smoking attitudes were non-significant.

Conclusion & Implications. Daily smoking is qualitatively different from less frequent and/or no smoking and is associated with some but not all social and individual influences assessed. The results imply working with parents to strengthen home smoking rules, starting prevention efforts before the age of 18, and targeting white youth would have the largest impact on reducing daily smoking.