Rates between stages of the justice pipeline

Cases moving from one stage to the next (charges to convictions, admissions to releases) take time, so a rate of this period's outflow over this period's inflow mixes cohorts. jpipe() divides the numerator summed over a window by the denominator summed over the same window moved back by the lag between the stages.

Usage

jpipe(num, den, win = 12, lag = 0, grp = NULL)

Arguments

num

The counts of the later stage, in time order: numbers from 0.

den

The counts of the earlier stage: numbers from 0, one per num.

win

The window, in periods: a whole number from 1.

lag

The lag between the stages, in periods: a whole number from 0.

grp

NULL, or the group of each period (each group in time order): labels.

Value

A data frame of class jpipe with one row per period, the groups in label order: group, index (the period within its group), num and den (the window sums, NA until a full window), rate.

Examples

jpipe(num = c(5, 7, 6, 9, 8, 10), den = c(20, 22, 19, 25, 24, 26),
      win = 3, lag = 1)
#> jpipe: lagged rates between stages
#>  group index num den      rate
#>    all     1  NA  NA        NA
#>    all     2  NA  NA        NA
#>    all     3  18  NA        NA
#>    all     4  22  61 0.3606557
#>    all     5  23  66 0.3484848
#>    all     6  27  68 0.3970588