Skip to contents

The bound a non-decision time is measured against

Source: R/ndt-seam.R
ndt_bound.Rd

Derives the upper bound wiener() and its siblings put on the non-decision time, so that a family in another package gets the same bound, the same ndt_group() behavior and the same refusals without writing them again. Pair it with ndt_bound_attach(), which puts the result on a family object.

Usage

ndt_bound(y, aterms = list(), max_ndt = NULL, what = "ndt_bound()")

Arguments

y

The response, a numeric vector of times.

aterms

The addition-term values for this response, as a density receives them. Only ndt_group is read.

max_ndt

An absolute bound in the response's own units, or NULL to take the fastest response. Above the fastest response it is refused, and it cannot be combined with ndt_group().

what

The family's name, used in the refusals so that the sentence a user reads names the family they wrote.

Value

An object of class "frmtmb_eam_ndt_bound": a list with ub, the scalar bound; floors, the per-group bounds or NULL; sizes, the trial count behind each one; pending, always FALSE here; and what.

Details

The bound is one of two things.

  • With no ndt_group() in aterms it is ONE number: max_ndt when that is given, and the fastest response otherwise. ndt is then a TIME, the link carries the bound, and a consumer's density reads dpars$ndt in the response's own units.

  • With ndt_group() it is one number PER GROUP, that group's own fastest response. ndt is then a FRACTION of the row's own bound, on a plain logit, and the density multiplies it back out by the per-row vector that arrives in the addition-term values under the reserved name ndt_floor.

max_ndt and ndt_group() together are refused: they set the same bound to different things.

Why the group's own floor and not the data set's

The information about one subject's non-decision time is that subject's own fastest response. At 30 subjects by 400 trials with a between-subject spread of 26 ms on a mean of 250 ms, 20 of the 30 subjects have a true non-decision time above the GLOBAL minimum while none is above its own, so a single bound runs the fit to the wall. wiener() and dev/ndt-findings.md carry the measurements.

See also

ndt_bound_attach() to put it on a family, ndt_bound_of() to read one back, ndt_time() to report a fitted non-decision time in the response's own units.

Examples

rt <- c(0.31, 0.42, 0.55, 0.61, 0.78)
g <- c("a", "a", "a", "b", "b")
# `[[` and not `$`: the record has five names and `bd$s` would
# partial-match `sizes`
ndt_bound(rt)[["ub"]]
#> [1] 0.31
ndt_bound(rt, list(ndt_group = ndt_bound_key(g)))[["floors"]]
#> 1627390186 1644167403 
#>       0.31       0.61