Report on the Effects of the Violation of Assumption on Regression Estimation of Census Coverage Area

Written by:
RR87-04

Introduction

This report takes the regression model and formulation of the basic undercount model presented by Freedman and Navidi (1984) as a starting point. The authors specify seven assumptions underlying the undercount estimation procedure of Ericksen and Kadane (1985). They then discuss five potential violations of the assumptions. These violations deal with the synthetic assumption, bias of the direct estimates, omitted explanatory variables, measurement error of explanatory variables and estimation of variances (independence of errors).

Related Information


Page Last Revised - October 28, 2021