Alaska 1990s Race "Bridged'' Smooth Series: 1990-2000
The adoption of the multirace standard for race reporting by the Office of Management and Budget has created statistical problems for the reporting of rates of incidence by race.
This has created two problems. First, it does not allow the production of rates after 2000 that are race compatible with those for the 1990s. Second, since educational statistics and vital statistics among others have not changed over to a multirace definition for incidence, rates produced with multirace denominators are inaccurate for 2000 forward until the data collection catches up. In addition, there is a need to correct the age/race/sex structure of population estimates produced during the 1990s so they produce a seamless series. The series presented here attempts to address these problems.
The National Center for Health Statistics in conjunction with the Census Bureau has produced a similar series but it has several shortcomings. First it doesn’t reflect population estimates produced by the Alaska Department of Labor and Workforce Development during the 1990s. Second, it uses national survey results to allocate the responses to the 2000 census race definitions back to 1990-comparable race groups. Third, it does not correct the 2000 counts for the age imputation errors for persons under 18 in Alaska.
To define a comparable 1990 race definition for 2000 race we used the ``Equal Proportion Bridge Series'' Race and Ethnicity estimates that our office produced in 2001, which were reported in the October 2001 Alaska Economic Trends. To adjust the age and sex of the population, we applied a formula referred to as Das Gupta 6 to the age cohorts. It takes the difference between an April 1, 2000, estimate and the April 1, 2000, census and distributed the difference back through the decade with diminishing weights. The formula is as follows:
P’y =(Cc / Ec)t(y)/n * Py
P’y is the intercensal estimate for year y
Ccis the decennial census number, year c
Ec is the postcensal estimate for the Census Year
n is the number of years in the time period between censuses (10)
y (note that typically the estimate is for July 1, and the census is for April 1, so this quantity will end with .25, eg. For year 2 this will be 2.25).
Py is the original postcensal estimate for year y. (Note that Ec=P10)