Score one ascending ladder by consistency maximization — .score

The Kirby-style scoring core shared by the MCQ scorers (overall + magnitude ladders) and the PDQ block scorer. Extracted verbatim from score_one_mcq()'s magnitude loop; reproduces Gray et al.'s (2016) published PDQ lookup tables for all 1024 response patterns per block.

Usage

.score_ladder(resp, vals, edge, top = c("geomean", "edge"))

Arguments

resp: 0/1/NA responses in ladder order (1 = chose the larger delayed/risky option). Any NA makes all three outputs NA – impute upstream.
vals: Ascending indifference values (k or h), one per item; must be the same length as resp.
edge: Value appended past the steepest item (the overall-ladder edge constant, or vals[length(vals)] for the repeat-last convention).
top: Value for the top switch point (smaller/guaranteed option on every item): "geomean" (default) = geometric mean of the last item and edge; "edge" = edge itself. Identical under the repeat-last edge.

Value

list(value, consistency, proportion).