Title not available in English.

Abstract

This work tries to check consistency of different infant and adult mortality estimation results made with data from different sources, by means of empirical applications. The methods develope by Brass, Feeney and Preston-Palloni were used to estimate childhood mortality. The confrontation of Preston-Palloni's method with Brass and Feeney's is an interesting one because it dispenses the assumption of constant fertility and is adequate to open populations. Data used for the two first methods are of Brass' sort. For the Preston-Palloni method, results obtained through the application of the Own-Child method were used. The original data comes from the Demographic Census of 1970 and 1980. Results obtained by using Brass and Feeney methods were compatible, in spite of the differences in procedures. Preston-Palloni results were rather unsatisfactory, possibly due to limitations of data derived from the application of the Own-Child method. To overcome this deficiency, we propose some new questions to be incorporated to the forthcoming Demographic Censuses.Data on maternal orphanhood and of deaths ocurred in the household in the previous year, incorporated to the Demographic Censuses of 1980, allowed the estimation of adult death probabilities. The results were compatible with estimations based on data corning from the Civil Register. The importance of the information about deaths in the previos year is stressed because of its potential use as source for calculations of socio-economic differentials of adult mortality in Brazil. Coale and Demeny Regional Model Life Tables were used, the Brazil Model being the more compatible. There are indications that the Brazilian mortality pattern is caracterized by higher childhood mortality and lower adult mortality than indicated by the Coale and Demeny life tables used. More extensive application of methods is needed in order to better evaluate available sources of information.