This paper examined the stability of response style by SEM. In previously study, the response pattern can't be explained by response style because the homogeneous content of items are administrated cross time. The heterogeneous items were used in this study which can distinguish between common variance due to style and content. The four response styles are examined, positive response, negative response, midpoint response, and extreme response. The interesting is the two order model which modeled the time invariant response style factors as second order factors, and the time-specific response style are captured in first order factors. Then the model data fit was evaluated and the demographic factors affecting the response style were discussed.
1) what's the difference between the two order model and using the measurement invariance, such as structure invariance and strict factorial invariance? How about to use factorial invariance to examine the Wave 1 group and Wave 2 group?