A Study on Comparison of Classification and Regression Tree and Multiple Regression for Predicting of Soldiers' Depression / 정신간호학회지

ABSTRACT

PURPOSE: The purpose of this study was to explore status and level of depression, anxiety and self esteem in Korean soldiers and identify effective statistical methods that account for and predict their depression. METHODS: A cross sectional study design was employed. Data were collected from five hundred thirty four soldiers in Gang-won Province and analyzed using stepwise multiple regression and Classification and Regression Tree (CART) with SPSS/WIN 18.0 program. RESULTS: The mean scores for depression, anxiety and self-esteem were 10.7+/-9.75, 38.5+/-10.16 and 31.7+/-5.20 respectively. Around one-forth (23.6%) of participants were above mild depression level. Major variables showing significant correlations were anxiety, self-esteem, duration of military service and the number of ventilation activities. Anxiety, self-esteem and duration of military service accounted for 62.3% of the variance in depression according to multiple regression. In CART analysis, predicting factors in the high risk group were high level anxiety and uncertain plan after discharge from military. CONCLUSION: The result of this study showed that anxiety was major factor of soldiers' depression both in multiple regression and CART. Also, CART applied in this study was an effective method in screening a risk group of soldiers' depression.