返回Google圖書搜尋

A Regression Method for Estimating the Completeness of Registration of White Births for States

出版	U.S. Census Bureau [custodian]
URL	http://books.google.com.hk/books?id=0Ow3wAEACAAJ&hl=&source=gbs_api

註釋The results of the 1940, 1950, and 1964-68 Birth Registration Tests regarding the completeness of registration of white births for states and the interpolated values for other years are not completely accurate. Analysis of the variation in the test results both within and between regions and the results of research on the coverage of the white population of states in the 1970 census suggest that the completeness of registration in some states may be overstated or understated in the tests, particularly in 1940, by as much as 1 - 3 percent. This paper describes a method for evaluating the results of the Birth Registration Tests for States and the corresponding interpolated estimates of birth registration completeness for white births since 1935 in each of the States, while retaining the national results. For each of the three test years, a regression equation is developed relating the percent of registered white births occurring in hospitals and several dummy variables (describing characteristics of states and their registration programs) to the completeness of registration of white births. The equations provide a sufficiently good fit to the data to suggest that the regression estimates may be more accurate than the test results and that the regression coefficients may be fruitfully employed in interpolation schemes for other years. The results of applying the regression estimates to determine the corrected number of white births for states over the period 1935 to 1970 are described; other applications and possible extensions of the research are discussed.