On the Efficiency of Grouped Regression Estimators Based on Indian Consumption Data

For one reason or the other, in the regression analysis of cross-section data in general and of consumer-budget data in particular, one has to deal with grouped observations rather than the original ungrouped observations. In other words, this means that the original observations are first grouped into an appropriate number of classes formed on the basis of some desirable criterion, then class means are obtained for the variables—explained as well as explanatory-entering the regression model, and the regression analysis is applied to these class means.