Quality of 1980 census data for American Indians

Abstract The count of American Indians in the 1980 census was over 70 per cent larger than the 1970 census count. An assessment of the demographic basis for this change shows that the cohorts from ages 10 to 74 in 1980 increased by substantial amounts, reaching 35 per cent for many ages. Increases of this nature in the absence of immigration are demographically impossible—an indication that the changes in response patterns identified by Passel (1976) between the 1970 and 1960 censuses persisted in 1980, possibly at even greater levels. In addition to presenting demographic analyses of the American Indian data at the national level, this paper includes an analysis of geographic variation of implied birth, death, and migration rates at the state level. States which historically have had large American Indian populations in general had high birth and death rates with reasonable migration rates. Many other states, however, had anomalously low birth and death rates with extraordinarily high implied migration rates. This pattern suggests that the changes in response may have occurred primarily in the latter areas. Several other anomalies in the American Indian data are reported here. The sample figure for American Indians from the 1980 census exceeds the complete count by more than 8 per cent at the national level. Increases occurred in most states, but the amount of increase varied substantially. Also, the increases tend to occur outside American Indian areas (reservations, tribal trust lands, etc.). Such differences are, for the most part, outside of expected sample variability. Furthermore, differences in the number of persons reporting American Indian race and American Indian ancestry were substantial. Some of the causes of these differences are investigated here. The paper concludes with an overall assessment of the quality and utility of these census data for various types of analysis.