Stratify a rate schedule by additional columns, applying proportional changes to rates by stratum.
Arguments
- rates
A data frame with a numeric
ratecolumn, such as a mortality, fertility, search, conversion, dissolution or transition rate table.- by
A named list. Each name is a column to add; each element is a named numeric vector of finite, nonnegative factors, one per category. The final rate will be determined by taking the original rate times the factor provided.
Details
Any of the input rate processes such as fertility, mortality, marriage, transition, dissolution, and conversion can be stratified by this function. Stratified rates can also be further stratified by this function.
Columns in by are either built-in strata (sex, union_type,
previous_union, origin, destination) or group dimensions declared with
set_groups(): a dimension's name (such as race) for person-level rates,
or <name>_w and <name>_h (the wife's and husband's categories) for
dissolution and conversion rates. Band, rate and calendar columns cannot be
added. Process setters reject built-in strata their process does not support.
If by includes multiple columns, every combination of categories gets the
product of its factors.
Examples
# stratify fertility rates by group and partnership status
usa_2000_fertility |> calculate_tfr()
#> # A tibble: 1 × 3
#> age_start age_end tfr
#> <int> <int> <dbl>
#> 1 15 45 2.05
fertility_rates <- stratify_rates(
usa_2000_fertility,
by = list(
group = c(A = 1.1, B = 1),
# this is not really a great idea for union type as within union
# type fertility rates have a totally different age profile, but
# ok for testing purposes
union_type = c("unpartnered" = 0.3, "partnered" = 1.3)
)
)
fertility_rates |> calculate_tfr()
#> # A tibble: 4 × 5
#> group union_type age_start age_end tfr
#> <chr> <chr> <int> <int> <dbl>
#> 1 A unpartnered 15 45 0.676
#> 2 A partnered 15 45 2.93
#> 3 B unpartnered 15 45 0.615
#> 4 B partnered 15 45 2.66
# although this works, a proportional hazards assumption by union type
# is unrealistic because the within union type age profile of fertility is
# very different from the overall age profile. So for this case, its better
# to use already stratified rates by union type and then further stratify
# by group
fertility_rates <- stratify_rates(
usa_2000_fertility_union,
by = list(group = c(A = 1.1, B = 1))
)
fertility_rates |> calculate_tfr()
#> # A tibble: 4 × 5
#> union_type group age_start age_end tfr
#> <chr> <chr> <int> <int> <dbl>
#> 1 partnered A 15 45 4.61
#> 2 unpartnered A 15 45 1.29
#> 3 partnered B 15 45 4.20
#> 4 unpartnered B 15 45 1.17
