At the end of the session, participants either received course credit (students) or randomly selected a prize new consisting of $10.00 to Walmart, $5.00 to Publix grocery, or discount coupons. Interobserver agreement Interobserver agreement (IOA) was collected to verify exhalation speed. An RA (different from the one who conducted the session) viewed the videos. IOA was calculated by taking the smaller observed speed, dividing it by the larger speed, and multiplying by 100 (Cooper, Heron, & Heward, 2007). Mean IOA was 91%. Data analysis Analyses of variance (ANOVAs) were conducted to verify that there was a significant difference between groups for number of cigarettes smoked per day and FTND scores. Repeated measures ANOVAs assessed whether there were differences in CO and exhalation speed across conditions and groups and whether there was an effect of condition order.

When main effects were found, Bonferroni post-hoc analyses were performed. Results were deemed statistically significant at p < .05. All measures of variance are SDs, unless otherwise noted. Finally, to identify an optimal CO cutoff, sensitivity and specificity were calculated at various CO levels for the slow and fast conditions. All participants in the nonsmoking, and three participants in the light smoking group who reported smoking a cigarette greater than 24 hr from the session, were classified as nonsmokers for this analysis. The remaining 17 light smokers reported smoking within the previous 24 hr and were classified as smokers (mean time since last cigarette 5.5 hr).

Because of the half-life of breath CO (approximately 2�C8 hr; Benowitz et al., 2002), a 24-hr window was adequate for classifying individuals as either a smoker or a nonsmoker. Light smokers were used for the analysis because the purpose was to identify a sensitive CO cutoff that would ensure that people who are trying to quit will be unable to meet the cutoff, even if they smoke a few cigarettes per day. Sensitivity, defined as the ability of a particular CO cutoff to accurately detect smoking when smoking has taken place (Javors et al., 2005), was calculated by assessing the correspondence between smokers and their obtained CO level at CO cutoffs ranging from 1 to 10 ppm. For example, if a participant reported smoking and blew a CO of 5 ppm, then all CO cutoff values of 1, 2, 3, or 4 ppm would accurately categorize the individual as a smoker, whereas all CO cutoffs at or above 5 ppm would incorrectly categorize the individual as a nonsmoker.

For each cutoff, the proportion of participants with CO values above that cutoff, out of all participants who reported smoking, were calculated. For example, 34 (out of 80) samples involved participants who reported smoking cigarettes and only 18 of those samples were 10 ppm or higher during the slow Dacomitinib condition (i.e., 18/34 = 0.53), whereas 34 samples were 5 ppm or lower (i.e., 34/34 = 1.00).