Factor Analysis
The data (n = 360) were examined for normality (by inspect- ing skewness and kurtosis), linearity (by inspecting scattering plots), and the presence of outliers, and visually inspected using histogram graphs. No outliers were detected in any of the variables, and the standard deviations were within an acceptable range of less than three deviations. Most of the statistics were within acceptable ranges (e.g., skewness and kurtosis ± 2.00) and normally distributed. Three variables (or items) relating to emotional value reported kurtosis sta- tistics of 2.02, suggesting slightly beyond normal distribu- tion for this construct. This issue of nonnormality is addressed in the discussion of the partial least squares (PLS) analysis. Thus, the data were suitable for factor analysis.
As Table 2 shows, the factor analysis (n = 360) revealed that the items in the questionnaire were valid and reliable measures of each construct. Factor loadings ranged from .74 to .98 and the variance explained ranged from .69 to .85, indicating that the items were valid measures on the constructs, consistent with Hair et al.’s (1995) recommen- dations. The scales were also reliable composite measures of the constructs, reporting Cronbach’s alpha levels between .88 and .96, consistent with Malhotra’s (2010) recommendations.