Human Development Index and Human Poverty Index for Indian states, 2005: Multivariate Statistical Analysis of basic indicators

Abstract: Human development index (HDI) is a summary measure used around the world that indicates whether a country is developed, still developing, or undeveloped. HDI incorporates the main factors of human life such as health, education and income. Deprivation in these areas of human life is measured by calculating human poverty index (HPI). The purpose of this thesis is to find the human development index and human poverty index for all states of India and perform multivariate statistical analysis on the indicators used in the calculation of HDI and HPI. We calculate HDI and HPI for all 33 states of India and observe variation among states according to HDI and HPI. Ranking these 33 states helps us to compare the states according to both HDI and HPI. Multivariate techniques are used to analyze the main indicators used in the calculation of HDI and HPI. Performing factor analysis separately for the indicators used in the calculation of HDI and those of HPI, we find that two factors explain 85% and 66% variation in the data respectively. We observe that indicators used in HDI calculation are strongly correlated to each other but we don’t found any statistically significant correlation among the indicators used in the calculation of HPI. Using cluster analysis for 7 indicators from HDI and HPI, we divide all the 33 states into two clusters size of 13 and 20 each. We observe that 12 states in cluster 1 are among the 14 top ranks states according to HDI. We observe that most of the states are highly developed and rich in 1st cluster. The ANOVA table for these 7 indicators shows that the indicators from HPI are not statistically significant. After removing these indicators from our analysis we obtain two clusters again of same size and having same states as in our first cluster analysis. We conclude that to divide these 33 states into two clusters we only need the indicators from HDI.