Cessation during the age 20- to 28-year interval was defined using the same criteria, except that cessation was for at least twelve months at age 28. While a substantial duration of cessation would be ideal (e.g., at least twelve months) for the age 20 outcome, only a small number of individuals had quit for a year because only 2 years had passed since the age 18 assessment. selleck chemical Demographic characteristics by smoking outcomes are shown in Table 1. Table 1. Demographic Characteristics by Smoking Outcomes Analytic Strategy We used logistic regression models to examine to what extent avoidant coping predicts the absolute probability that an individual would make escalate or quit smoking. The first two models expressed the absolute probability of escalating to daily smoking during the period ages 18�C20 (first model) and 20�C28 (second model) as a function of scoring high on avoidant coping at age 18.

The second set of two models expressed the absolute probability of quitting smoking during the periods 18�C20 (first model) and 20�C28 (second model) as a function of the same variable. All probabilities ranged from 0 to 1 and were generated with Stata’s prvalue function, which uses the delta method to transform the logistic regression model coefficients of interest into probabilities while holding other adjustment covariates at their mean values. See our prior papers for more information about these probability models (e.g., Bricker et al., 2009). In addition to the probability models, we also provide readers the traditionally reported odds ratios from logistic regression models.

With each model, a Wald test for heterogeneity was used to assess the influence of avoidant coping at the quartiles of this measure. A Wald test for interaction was also used to examine the moderating effect of life stress. A linear test for trend was conducted by including an ordered variable (e.g., avoidant coping or stress) in the regression model as a continuous covariate to demonstrate if any monotonic relationship existed between the degree of avoidant coping and the probability of smoking escalation and cessation. Covariates All models adjusted for gender, parents�� highest level of education (less than high school [HS] and greater than or equal to HS), and the condition (i.e., control vs. experimental). In addition, all models accounted for intraclass correlations due to clustering within a school district by using Stata’s cluster variance estimation option. Per HSPP trial design specifications, all 40 districts had one HS per district (Peterson et al., 2000). Therefore, clustering by school district was analytically the same as that by HS. All statistical analyses were conducted Brefeldin_A with Stata Statistical Software (version 10.0).