Analyses and Results
The model in the current study consists of multiple variables whose relationships are not well-defined and form various inner models. Besides, in the current study’s measurement model, there could be random error caused by respondent fatigue and systematic error caused by variance attributable to the measurement method. Hence, to establish the extent to which the proposed model represents an acceptable approximation of actual relationship, analysis of the collected data was performed using structural equation modeling (SEM) with IBM SPSS AMOS v20.0 and various fit indices were examined.
We first tested our unconstrained measurement model. The CFA showed that this measurement model was a reasonable fit for the data with χ2 = 185.5, RMSEA = 0.049, CFI = 0.963, NFI = 0.956. Our sample being a smaller one we used CFI and NFI as they are less sensitive to sample size (Hair et al, 1998; Ho 2006).
Table III: Confirmatory Factor Analysis for validity and reliability of construct measures