Examination of Census Omission and Erroneous Enumeration Based

Attached is the second of two reports on results of the study based on the Ethnographic Evaluation of the Behavioral Causes of Census Undercount for the 1990 Decennial Census. This report focuses on the results of an exploratory data analysis in which logistic linear regression models were fit to data to assess the relative values of explanatory variables on the prediction of census omissions and erroneous enumerations. 1 I. INTRODUCTION This paper reports results of an analysis of data from the Ethnographic Evaluation of the Behavioral Causes of Census Undercount for the 1990 Decennial Census (referred to as the 1990 Ethnographic Evaluation henceforth). The purpose of the analysis was to search for factors related to the two components of census coverage errors, omission and erroneous enumeration of persons, in the 1990 census. Earlier papers (de la Puente, 1992 and 1993b) on the 1990 Ethnographic Evaluation focused on census omissions and examined the effects of demographic variables on the outcome of census enumeration. This paper broadened the scope of the analyses on two fronts. First, not only census omissions but also erroneous enumerations were examined. Second, in addition to demographic variables, factors related to the social aspects of the sample areas were included in the analyses. The results of the analyses in this paper confirmed many of the results of the earlier studies on the census coverage errors but also shed some new light on the possible effects that social and demographic factors might have had on the outcome of the census enumeration.