Table 1. Composite Reliability and Average Variance Extracted
Problems of missing data are often magnified in structural equation modeling, and so adequate missing-data computation is of particular importance (Ullman and Bentler 2004). Thus the multiple imputation option with Expected Maximization (EM) algorithm included in LISREL 8.72 was used for imputation of missing values. The technical details of this method are presented in Schafer (1997). Table 2 shows the fit indices for the measurement model, which included the five factors measuring attitudes towards different aspects of mobile advertising. Overall, the fit indices for the measurement model indicate that the scale structures fit the data acceptably.