Tuesday, August 23, 2011

Survivorship probabilities

Is that low or high? Some time ago, a family friend attended a 50th high school reunion and remarked to me that a little over 100 or 10% of the graduating class of 1,000 had passed away. In formal demography terms, that's a survivorship probability ratio for that cohort of ℓ(72) /ℓ(22) = 1 - 0.1 = 0.9 or so. What do you think, low or high?

Being well versed in dissembling, I said I thought that sounded about right. I think the family friend thought it was too low, that the Grim Reaper had claimed more than the usual number of victims in this cohort.

There are a number of elements that certainly matter here: the sex ratio, the racial and ethnic composition, and so on, all the covariates that we know affect mortality and tend to vary across geographic regions.

But a broad look can be retrieved from cohort life tables published by Social Security. There, these survivorship ratios for the 1940 and 1950 male and female birth cohorts are 0.73 (1940 men) and 0.83 (1940 women), and 0.76 (1950 men) and 0.84 (1950 women).

So the survivorship ratio was pretty high for that birth cohort. Part of it is surely the protective effect of the high school degree, but one can also infer something about the socioeconomic status and racial/ethnic composition of this cohort.