A research on a comparison of discriminant analysis and CHAID

In this study, it is aimed to compare the results of Discriminant analysis and CHAID analysis, which are multivariate statistical techniques. The research was designed in the type of predictive correlational research, one of the relational research models. The data subject to the research were collected online from 416 teachers working in different branches in public schools in Antalya Province Muratpaşa District. The "Teachers' Attitude Scale Towards Teaching Profession" and "Demographic Information Form" developed by Demirel and Ünişen (2018) with the online method were presented to the participants and the data were collected through these tools by appropriate sampling method. After the data collection process, the validity and reliability checks of the scale were made. The dependent variable in the study was determined as the scores obtained from the "Teachers' Attitude Scale towards Teaching Profession", and this continuous data was categorized as "those with positive attitudes" and "those with negative attitudes" with the k-means clustering algorithm. "Teacher's professional seniority year", "teacher's monthly total income", "teacher's monthly total income", "student number of the class in which the teacher teaches", "total number of weekly lesson hours given by the teacher" , "the total number of in-service training attended by the teacher during his/her professional life", "time of working at the same school" and "distance from the teacher's house to the school" obtained from "Demographic Information Form" were included in the analysis as independent variables. Before starting the analyzes, it was checked whether the assumptions required by the Discriminat Analysis were met. After providing the assumptions, analyzes were carried out. As a result of the discriminant analysis, the independent variables that can explain the dependent variable are "teacher's monthly total income", "monthly total income of the tacher's family, "students in the class in which the teacher teaches", "total number of teaching hours per week given by the teacher", "time working in the same school". " and "the distance from the teacher's house to the school" variables, while the independent variables that can explain the dependent variable as a result of the CHAID analysis are "the teacher's monthly income, the number of students in the class where the teacher teaches, "the time of working at the same school", "the distance from the teacher's house to the school" and "professional seniority years" variables. It was observed that the independent variable that best explained the dependent variable was the variable "distance from the teacher's house to the school" in both the discriminant analysis and the CHAID analysis. When the general classification situations of the analyzes were examined, it was seen that 398 of the 416 teachers participating in the study were classified in the right group by the discriminant function, while 18 of them were classified in the wrong group, and a general classification accuracy of 95.7% was obtained. Considering the classification status of the CHAID analysis, it was seen that 403 of the 416 teachers who participated in the study were predicted in the right class, and 13 of them were predicted in the wrong class, and it was understood that the analysis reached a total classification accuracy of 96.9% as a percentage. As it can be understood from the ratios, it has been determined that the CHAID analysis has a higher correct classification rate.