4.3. Main study: factor structure validation
The main study was conducted using the same sampling frame with the pre-test, excluding consumers who had participated in the pre-test from the pool. A total of 9921 emails were sent out, 1587 clicked from email offer, and 1200 completed the survey. A demographic profile of the respondents indicated that approximately 74% of the respondents were female and 87% were Caucasian. They represented all income categories with $50,000–59,999 as the median income. The sample was divided into two sub-samples, calibration sample and holdout sample, via random sample selection. The calibration data were used for running confirmatory factor analysis (CFA) and testing validities, and were compared with the holdout data to ensure the generalizability of the results.
Because the nine first-order constructs can be theoretically positioned as utilitarian shopping benefits, hedonic shopping benefits, or shopping costs, a higher order factor analysis was performed to determine whether the nine constructs could be modelled as three correlated second-order factors. CFA using LISREL 8.80 with the maximum likelihood method and covariance matrices produced supportive results for the nine first-order and three second-order factors of SVO. For the model fit indexes, we used chi-square, non-normed fit index (NNFI), and root mean square error of approximation (RMSEA), as these indices provide sufficient unique value, degree of freedom, comparative fit information for model evaluation. The model fit was good, supporting the use of indexing the three second-order factors for measuring SVO at the general level. The correlation coefficients among the three factors were between utilitarian and hedonic shopping benefit between utilitarian shopping benefit and shopping cost; and .667 between hedonic shopping benefit and shopping cost.