Testing for block effects and misspecification in regession models based on survey data

Authors:

Abstract
Much applied work in econometrics and other social sciences involves regression analysis using survey data. King and Evans (1986) and Bhatti (1991) considered an equicorrelated error term model and proposed one-sided locally best invariant and point optimal invariant tests for this model, respectively. We consider more general situations and construct tests for equicorrelation over different blocks and even over different subblocks using King and Wu's (1990) locally most mean powerful invariant (LMMPI) test. Further, a procedure of diagnostic checking for model adequacy based on testing simultaneously for omitted regressors and equicorrelation in the errors is presented. Monte Carlo studies are used to investigate the performance of our tests. Finally, we apply these tests to Bangladesh data.

This publication has 18 references indexed in Scilit: