MODELING THE RECIPROCAL OF THE SURVIVORSHIP FUNCTION

In a recent study by Mitra [1], Brass's model of linear relationship b etween the logits of the life table survivorship function l(x) of any two life tables was transformed to a similar relationship between the reciprocals of l(x). It has been shown in the current paper that such a relationship implies strong correlation between the reciprocals of l (x) and l(y) where x and y are any two ages and that this correlation increases as the age difference decreases in absolute value. Two facto rs emerged when this correlation matrix for each sex was subjected to principal components analysis. They explained 98 percent of the variat ion and as a result turned out to be excellent predictors of life tabl es in general and of comprehensive measures of mortality such as life expectancy in particular. In fact, the factor scores corresponding to the two principal components, for any life table or for any country, c an be treated as measures of mortality along two independent dimension s labeled, in this paper, as acquired-immunity and senescence.