factor. Moreover, the factor loading ranging from 0.703 – 0.876 the highest factor
Mahasarakham University
69
loading is in continuous improvement mindset and the lowest factor loading is in audit
innovation learning. Further, construct validity of this research was tapped by items in
the measure, as theorized. Moreover, the factor loadings of each item are greater than
the 0.40 cut-off and are statistically significant that recommended by Nunnally and
Bernstein (1994). Consequently, the construct validity is obtained. Additionally, the
Cronbach’s alpha coefficients for all variables are expressed between 0.707 – 0.810 the
highest coefficient is in audit profession well-roundedness and the lowest coefficient is
in audit judgment, which is greater than 0.70 as recommended by Hair and others
(2006) and Cronbach (1951). That is, internal consistency of the measures used in this
study is considered good for all constructs. As a result, the reliability of all variables is
obtained.
Statistics
The Ordinary Least Squares (OLS) regression analysis is used to test
hypotheses relationship to meet the objective. The statistic techniques include factor
analysis which is exploited to ensure the validity, variance inflation factor (VIFs),
correlation analysis. But before hypotheses testing, all of raw data are checked,
encoded, and recorded in a data file. Then, the basis assumption of regression analysis is
tested. This process involves checking the normality, heteroscedasticity,
autocorrelation, and linearity. Moreover, outliner problem is concerned.
Factor Analysis. This research uses factor analysis to ensure two statistic
objectives. First, convergent and discriminant validity are examined. Second, factor
score of all variables is calculated in order to avoid from the multicollinearity problem
and assessed by Ordinary Least Squares (OLS) regression analysis.
Variance inflation factor. To identify multicollinearity problem, this research
conducts a variance inflation factor (VIF) and a tolerance value as indicators to indicate
a high degree of multicollinearity among the independent variables. VIF is directly
related to the tolerance value. A tolerance value is greater than 0.10 and VIF is less than