A. Reliability and Validity of Indicators
1) Reliability Test
Measured before part of the research questionnaire using the revised total correlation coefficient (CITC) measurement to purify and Cronbach ’ s Alpha coefficient and standard deviation to test the reliability of the questionnaire. Correct project reliability test screening of the project criteria two, must both set up before you can delete this item: after correction total correlation coefficient (every study design and variable measured score associated with the remaining items score between coefficients) of less than 0.3; remove this item to increase alpha value, which can enhance the overall reliability (line Lu Dai, 2002)
a) CITC total correlation coefficient and reliability of brand experience
Obviously could be seen from table 2, the revised total correlation coefficient (CITC) is greater than 0.3, and delete any item, did not improve overall reliability. Thus, 25 measurement of brand experience projects do not meet deletion standards, all preserved.
b) CITC total correlation coefficient and reliability of brand loyalty
Obviously could be seen from table 3, as amended, the total correlation coefficient (CITC) is greater than 0.3, and delete any item, do not raise the overall reliability. The brand loyalty measurement of 4 items do not meet delete criteria, all preserved.
2) Validity Test
In this study we took confirmatory factor analysis in statistic analysis to measure tht validity of the indexes, tested by structural equation modeling. The results of the test are as follows:
a) Validity Test of the brand experience
The conclusion could be reached from Figure 2, the metrics in their corresponding variables on the standardization factor loading is greater than 0.5, that achieved statistically significant level, indicating that the convergent validity of study on this scale was better.
B) Validity Test of the brand loyalty
The conclusion could be reached from Figure 3, the metrics in their corresponding variables on the standardization factor loading is greater than 0.5, that achieved statistically significant level, indicating that the convergent validity of study on this scale as better.