Each analysis included only participants with complete data for the five primary variables. SS was missing for 4.2% of participants, compared with 3.6% for negative affect, 5% for risk perceptions, 0.3% for past 30-day smoking, and 0.2% for lifetime smoking. All five variables were present for 90.3% (n = 1,524) of participants, nevertheless excluding those who endorsed use of a fictitious substance or provided inconsistent data. To estimate an overall response rate (i.e., proportion of potential participants who provided usable data), we multiplied this figure by the 92% nonrefusal rate for all versions of the survey, yielding an estimated total response rate of 83.1%. All hypothesis tests used an alpha level of .05. Results Preliminary Analyses We first examined relationships between smoking outcomes and demographic variables.

Logistic regression showed that males were more likely to endorse past 30-day (odds ratio [OR] = 1.46 [95% CI 1.13, 1.89], p = .004) and lifetime smoking (OR = 1.37 [1.10, 1.70], p = .004), and that both 30-day (OR = 1.48 [1.31, 1.67], p < .001) and lifetime smoking (OR = 1.42 [1.28, 1.57], p < .001) increased with grade. Race/ethnicity was not associated with 30-day smoking but was related to lifetime smoking (OR = 1.09 [1.00, 1.20], p = .049), such that Asian Americans were less likely than others to endorse ever having smoked. Descriptive statistics for SS, risk perceptions, and negative affect are shown in Table 1. Male participants reported higher levels of SS but lower levels of both risk perceptions and negative affect.

Freshmen had lower levels of SS compared with juniors and seniors and lower levels of negative affect compared with sophomores and juniors. There were no grade differences in risk perceptions. Asian Americans reported lower levels of SS compared with other participants, whereas Hispanic students perceived smoking as least dangerous. There was no association between race/ethnicity and negative affect. Because of the associations between demographic variables and our primary variables, hypothesis tests using the full sample included interactions between SS and sex, grade, and race-ethnicity. Nonsignificant interactions were removed and the model was re-fit. Table 1. Associations Between Demographic Factors, Sensation Seeking, Mediators, and Outcomes SS and Smoking Logistic regression indicated that the interactions between SS and grade and race/ethnicity were not significant.

However, the sensation seeking �� sex interaction was significant for both 30-day (OR = 0.89 [0.82, 0.97], p = .008] and lifetime (OR = 0.92 [0.86, 0.99], p = .027) smoking. After stratifying Brefeldin_A the sample by sex to interpret these interactions, we found that SS was significantly associated with smoking for both sexes and that the association tended to be stronger for males (30 day: OR = 1.39 [1.30, 1.49], p < .001; lifetime: OR = 1.36 [1.28, 1.44], p < .