The impact of estimation method and population adjustment on Canadian life table estimates.

Abridged life tables centred on 1991 were produced from the 1991 Canadian census, net census undercoverage estimates, and death data from 1990 to 1992. The sensitivity of life table values to differing methods of estimation and population estimates was investigated. The results from four methods by Greville, Chiang, and Keyfitz were compared, and population estimates, both adjusted and unadjusted for net census undercoverage, were used to test the effects of method and type of population estimates on life table values. The results indicate that the method used to derive the estimates had much less influence on the life table values than did the choice of population estimate. The change in life expectancy at birth due to the method of calculation chosen was at most 15 days, whereas the change due to the population estimate chosen was about 73 days. Since there are age, sex and provincial variations in net undercoverage rates, life expectancies differed accordingly.