Case flow through a court — dktflw • crimlab

Cases received and disposed by period: the clearance rate (disposed over received, above 1 when the caseload shrinks) with its exact interval, the net, the clearance over a rolling window, and with the pending caseload its change and the periods it would take to clear at the recent pace. With the length of each period (span) the filing rate gives, by Little's law, the mean time to disposition the pending caseload implies, to set beside a published mean (mtime): their ratio far from 1 means the published mean leaves out the long-pending cases. Published duration bands (shares of cases in bands starting at floors) bound the mean from below.

Usage

dktflw(
  recv = NULL,
  disp = NULL,
  pend = NULL,
  prd = NULL,
  win = 12,
  level = 0.95,
  span = NULL,
  mtime = NULL,
  shares = NULL,
  floors = NULL,
  mix = NULL,
  scores = NULL
)

Arguments

recv, disp: NULL, or the cases received and disposed in each period: numbers from 0.
pend: NULL, or the cases pending at the end of each period.
prd: NULL (1, 2, ...) or the label of each period.
win: The rolling window in periods: a whole number from 1.
level: The level of the intervals: one number between 0 and 1.
span: NULL, or the days in each period (one number or one per period); needs pend.
mtime: NULL, or the published mean days to disposition of each period; needs span.
shares, floors: NULL, or the share of cases in each duration band (a matrix, one row per period) and each band's lower edge.
mix: NULL, or the outcome counts: a numeric matrix of periods by outcomes (named by its dimnames).
scores: NULL (1, 2, ...) or the period scores of the trend test.

Value

An object of class dktflw, a list: flow (with recv: period, recv, disp, clear, lo, hi, net, rclear, rlo, rhi, then pend, pchg, toclear, or cumnet, and with span frate, impl, with mtime ratio, ltlbk, with shares floorbd); mix (a list of shares, homog and trend); and level.

Details

With mix, the counts of each outcome (conviction, withdrawal, acquittal) by period: each period's shares with Goodman's simultaneous intervals, the chi-square test that the mix is the same in every period with Cramer's V, and the Cochran-Armitage trend in each share over the periods (as stats::prop.trend.test()) with Holm's adjustment.

References

Goodman, L. A. (1965). On simultaneous confidence intervals for multinomial proportions. Technometrics, 7(2), 247-254. doi:10.1080/00401706.1965.10490252

Examples

dktflw(recv = c(120, 130, 125, 140), disp = c(110, 135, 120, 150),
       pend = c(400, 395, 400, 390), win = 2, span = 91)
#> dktflw: case flow
#>  period recv disp     clear        lo       hi net   rclear       rlo      rhi
#>       1  120  110 0.9166667 0.7011985 1.197368  10       NA        NA       NA
#>       2  130  135 1.0384615 0.8101053 1.331593  -5 0.980000 0.8183090 1.173541
#>       3  125  120 0.9600000 0.7409969 1.243285   5 1.000000 0.8373357 1.194264
#>       4  140  150 1.0714286 0.8453024 1.358733 -10 1.018868 0.8568263 1.211642
#>  pend pchg  toclear    frate     impl
#>   400   NA 3.636364 1.318681 303.3333
#>   395   -5 3.224490 1.428571 276.5000
#>   400    5 3.137255 1.373626 291.2000
#>   390  -10 2.888889 1.538462 253.5000
m <- matrix(c(50, 55, 60, 30, 28, 25, 20, 17, 15), 3,
            dimnames = list(2021:2023, c("guilty", "withdrawn", "other")))
dktflw(mix = m)$mix$trend
#>     outcome      stat         p      holm change
#> 1    guilty 2.0202020 0.1552185 0.4656555   0.10
#> 2 withdrawn 0.6246183 0.4293363 0.7006036  -0.05
#> 3     other 0.8723635 0.3503018 0.7006036  -0.05