Spatial regression models are standard tools for analyzing data with spatial correlation. These models are broadly used in the social sciences for predicting the socio-economic factors. In this paper, we discuss about various spatial regression models and explain the concepts based on real data to demonstrate how to obtain and interpret relevant results. We describe prediction efficiencies of various predictors relative to the efficient minimum mean square error predictor in spatial models containing spatial lags in both the dependent variable and the error term. We consider Multiple Linear Regression Model (MLRM), Spatial Autoregressive Models (SAR), Spatial Autoregressive in the Error-term Model (SEM) and Spatial Durbin Models (SDMs) to estimate the literacy progress in the districts in Odisha as a result of changing socio-economic factors over time. The goodness of fit of the different models are compared along series of hypotheses about the performance of the specifications considering spatial relationships among the observations. The spatial analysis proved the existence of positive spatial autocorrelation and persistence of disparities in literacy attainment level across the regions during the analyzed period. The results of econometric analysis confirmed the expected positive impact of economic growth on literacy progress level as well as the necessity to incorporate the spatial dimension into the model.