Clustering Türkiye's Provinces According to Education Level Indicators with Two-Stage Cluster Analysis

Abstract Purpose: This study aims to cluster similar provinces of Turkey according to (i) below-high school and (ii) high school and above-high school education level indicators using cluster analysis methods. Design/Methodology: The annual most recent data for 81 provinces were taken from the education statistics for the year 2022 published on the official website of the Turkish Statistical Institute. A two-stage cluster analysis method was used to identify similar provinces. Findings: According to indicators of education level below the high school level, the first cluster consists of 17 provinces (Adıyaman, Ağrı, Bingöl, Bitlis, Diyarbakır, Gaziantep, Hakkâri, Kars, Mardin, Muş, Siirt, Şanlıurfa, Van, Batman, Şırnak, Iğdır and Kilis) and the second cluster consists of 64 provinces (other remaining provinces). The 17 provinces in the first cluster are eastern provinces where education needs to be increased. According to the education level indicators at high school and high school level, the first cluster consists of 6 provinces (Ankara, Çanakkale, Eskisehir, Isparta, Istanbul, and Izmir), and the second cluster consists of 75 provinces (the remaining other provinces). The six provinces in the first cluster are western provinces where education is advanced. Limitations: The limitation of the study is that cluster analyses were conducted for Turkey's 81 provinces and 2022 cross-sectional data. Originality/Value: The use of the two-stage cluster analysis method, which allows the number of clusters to be automatically determined according to AIC and BIC criteria, unlike the Ward and K-means methods used in cluster analysis, contributes to the original value of the research. Considering that the increase in education benefits people and society, it points out the importance of the study in providing meaningful and valuable information by creating awareness among individuals and policymakers in the relevant provinces by identifying similar provinces with clustering methods.