Decision tree-Based classification model for identification of effective leadership indicators

Abstract

This study was aimed at identifying effective leadership abilities as appreciated by soldiers in the Lithuanian armed forces. Leader behavior was measured using an adapted version of the Leader Behavior Description Questionnaire (LBDQ), which was originally developed by Andrew W. Halpin from Ohio State University. Data were collected from soldiers holding different ranks and doing professional military service in all units of the Lithuanian armed forces and were analyzed using the IBM SPSS version 20 software application. For our data analysis, the Chi-square Automatic Interaction Detector (CHAID) decision tree growing method was used with three class dependent variables. The CHAID algorithm helped in specifying the best splits for each of twelve potential predictors and then select the predictors whose splits presented the most serious differences in the sub-populations of the sample. In the Chi-squared significance test, the lowest p-value was achieved. The model structures obtained after analysis are presented.