Pipeline stage: score test takers — score

Estimates each test taker's ability from the calibrated model with mirt::fscores() and checks the scores (thresholds from mv_rules()):

Usage

score_persons(p, method = c("EAP", "MAP", "ML", "WLE"))

Arguments

p: a calibrated pipeline.
method: scoring method passed to mirt::fscores(): "EAP", "MAP", "ML" or "WLE".

Value

The pipeline with p$scores, a tibble with one row per test taker (id, theta, se, and group if the data have one).

Details

score_reliability (review): the empirical reliability of the scores is below reliability (the blueprint's target, if one was set).
extreme_scores (info): maximum-likelihood estimates that are infinite (all or none correct) or standard errors above max_se.

Examples

# \donttest{
p <- calibrate(add_responses(mv_pipeline(), expand.table(LSAT7)), model = "2PL")
p <- score_persons(p)
#> Warning: Scoring is running with 3 review open flag(s) from earlier stages:
#>   #1 [review] local_dependence: Item.1 & Item.3 (Q3 = -0.21)
#>   #2 [review] item_misfit: Item.2 (S-X2 = 14.5, adjusted p = .004)
#>   #3 [review] low_reliability: test (marginal reliability = 0.44)
#> Flags only warn because strict = FALSE. See flags(p).
head(p$scores)
#> # A tibble: 6 × 3
#>   id    theta    se
#>   <chr> <dbl> <dbl>
#> 1 1     -1.87 0.693
#> 2 2     -1.87 0.693
#> 3 3     -1.87 0.693
#> 4 4     -1.87 0.693
#> 5 5     -1.87 0.693
#> 6 6     -1.87 0.693
# }