40 variables simultaneously. Tests carried out by using the computer application
program SPSS 20.0 for Windows.
1. Classical Assumptions
a. Normality Test
Normality test is intended to determine whether the used data is normally distributed or not. Normality test needs to be done to
determine the statistical tools, so that the conclusions drawn can be accounted for. There are two ways to detect whether the residuals are
normally distributed or not, namely: 1
Graph Analysis One of the easiest ways to see the normality of the
residuals is to see a histogram graph comparing the observational data with the distribution which closes to normal
distribution. More reliable method is by looking the normal probability plots comparing to the cumulative distribution of a
normal distribution. Normal distribution will form a straight diagonal line and plot the data will be compared with the
residual diagonal lines. If the residual data distribution is normal, then the line that describes the real data would follow
the diagonal line. 2
Statistical Analysis Simple statistical test can be done by looking at the value
of the kurtosis and Z-values of skewness. Another statistical test
41 that can be used to test the normality of residuals is non-
parametric statistical test of Kolmogorov-Smirnov KS, if the significance level 0.05, then the data is normally distributed
and can be performed multiple regression models. Guidelines for decision-making about the data close to or a normal
distribution by Kolmogorov Smirnov can be seen from: a
Sig. or significantly or probability 0.05, then the data distribution is not normal.
b Sig. Or significantly or probability 0.05, then the data
distribution is normal.
b. Multicollinearity Test
Multicollinearity test means between the independent variables included in the regression model has a linear relationship is
perfect or near-perfect or even a high correlation coefficient of 1. The Regression model should appear as neither perfect nor near-
perfect correlation between the independent variables. The consequence is the correlation coefficient multicollinearity particular
variable, and the error becomes very large or infinite.
Multicollinearity can also be seen from 1 the value of tolerance and the opponent 2 variation inflation factor VIF. Cut-
off value commonly used to indicate the presence of multicollinearity is tolerance value 0.10 or equal to the value of
VIF 10