2.4 Statistical Analysis
The package SPSS 20 was used to analyze data. Conformity to normal
distribution was examined with skewness and kurtosis coefficients in
addition to the Kolmogorov-Smirnov test. The medians of groups which did
not show normal distribution were compared, and in the comparison of two
independent groups, the Mann-Whitney test was used, and in the
comparison of three or more independent groups, the Kruskal-Wallis test
was used. When the difference between groups was significant, Bonferroni
correction was used to determine the difference.
Correlation analysis was used for the relation between continuous
variables. Effect size was examined with Cohen’s d, and effect dimension
with Cohen’s r. Cohen’s d:0.20 was accepted as a small effect, d:0.50 as
a moderate effect, d:0.80 as a large effect, r:0.10 as a low effect,
r:0.30 as a moderate effect, and r≥0.50 as a high effect30. Independent variables in the linear regression
model (enter method) in which GAD-7 was accepted as a dependent variable
were ISI, OSI and CSI scores. In addition, the categoric variables of
gender, having children, marital status, and tobacco and alcohol use
were analyzed in the model as dummy variables, and as a result only the
variables of gender (reference group: female) and having children
(reference group: I have children) were found to be significant for the
linear regression model (p<0.05 and F4,307:
106.347). In order to identify the problem of multicollinearity, the
limits of VIF (Variance Inflation Factor)<10,
tolerance<2 and Durbin-Watson<2.5 were checked, and
the conformity of residuals to normal distribution was examined, taking
skewness-kurtosis coefficients of ±1 as a base 31. A
statistical significance level of p<0.05 was taken as
significant.