Crystal Smit

Evaluation of the Motivation Process 5 109 higher) versus negative responses (score of 2 or lower) on these measurements. To analyze the quantitative data associated with the second and third aim, a series of paired sample t -tests were conducted to examine changes before and after the training in the influence agents’ intrinsic motivation, and SSB and water consumption (second aim); and their peers’ intrinsic motivation, perceived social support, and perceived social norms (third aim). It should be noted that although the data fail to meet the assumptions of normality, the paired sample t -test was nevertheless chosen over the customary Wilcoxon signed-rank (nonparametric) test due to the findings supporting its application in small samples involving non- normal distributions, and/or ordinal data (Meek, Ozgur, & Dunning, 2007). Table 5.2 presents the descriptive statistics of these quantitative measurements. A content analysis was performed on the open-ended (qualitative) data related to the research goals. First, the primary researcher openly coded the open-ended responses to compile the categories, and afterwards, a second researcher coded the responses using the compiled categories. The responses of the influence agents related to their experiences with the training were classified based on the techniques implemented in the training. The influence agents’ responses concerning how they motivated their peers were classified based on whether or not they had set a good example themselves (i.e., modelling; Cruwys et al., 2015), talked about water and its benefits (Rogers, 2010), and/or had offered social support (Kiernan et al., 2012). The Krippendorff’s alpha test was used to estimate the interrater reliability between the two coders (Hayes & Krippendorff, 2007). The interrater reliability ranged from acceptable to good (Krippendorff’s alpha ranged between α = .77 and α = 1.00). Finally, the percentages of influence agents in the compiled categories were reported. Additional analyses (i.e., Pearson’s correlations) were performed to explore the effect of the training on the changes in intrinsic motivation, social support, and perceived social norms for different demographic variables of the peers (i.e., sex, grade level, and family affluence). The interindividual change score between the two assessments of the measurements were included as a change variable in the correlation analyses.