survexp
) and person years (
pyears
) functions.
"white"
and
"nonwhite"
. For the 1970 data, the nonwhite population values were not reported separately; to make the rate table a matrix, the 1980 nonwhite data values are used instead.
"white"
,
"nonwhite"
, and
"black"
. For the 1970 data, the black population values were not reported separately, so the nonwhite values are used instead.
"white"
,
"nonwhite"
, and
"black"
. For the 1960 and 1970 data, the black population values were not reported separately, so the nonwhite values are used instead.
Each of these tables contains the daily hazard rate for a matched subject from the chosen population.
The daily hazard rate is defined as
-log(1-q)/365.24
, where
q
is the 1 year probability of death as reported in the original tables.
Thus, for age 25 in 1970,
p = 1-q
is the probability that a subject who turns 25 years old in 1970 achieves their 26th birthday. The tables are recast in terms of "hazard per day" for computational convenience.
The fraction 0.24 in the denominator of the daily hazard rate is needed to account for the 24 leap years each century.
Each rate table is stored as an array with additional attributes, and can therefore be subscripted and manipulated as a standard S-PLUS array. Interpolation between calendar years is done "on the fly" by the
survexp
routine.
Some of the rate tables were augmented to contain extrapolated values for the year 2000. Many of the deficiencies in the tables (e.g., the 1970 Arizona nonwhite population) are a result of local conditions. The data probably exist, but we don't have access to it at this time.