RESULTS
Measurement Operationalization
The primary concern relating to the measures of study constructs was multicollinearity stemming from high
correlations between OSQ items and CS items (Taylor & Baker, 1994). Thus, a multicollinearity check was
conducted by examining the correlation matrices. While the correlations between two CS items (CS2 and CS3) and
OSQ items ranged from ±.67 to ±.72, which is below the problematic level of 0.80 (Hair et al., 1998), the first
satisfaction item (CS1) strongly correlated with the whole range of OSQ items (.82, .86, and .89), showing the
presence of a multicollinearity problem. These values are greater than their correlations with other satisfaction items
Do'stlaringiz bilan baham: |