Inspect MAPT Cuts and Fit Diagnostics

Show the main MAPT result without printing all fitted models and input rows. The main cuts use the ratings from all raters together. They are not averages of separately estimated rater cuts.

Usage

# S3 method for class 'cutsIDMMAPT'
summary(object, per_rater = FALSE, ...)

# S3 method for class 'cutsIDMMAPT'
print(x, digits = 3, ...)

# S3 method for class 'summary.cutsIDMMAPT'
print(x, digits = 3, ...)

Arguments

object: A result from computeCutsIDMMAPT.
x: A MAPT result or its summary, as appropriate.
per_rater: Also include individual rater cuts and their complete diagnostic tables in the summary? Defaults to FALSE. This does not fit new models: individual results must have been computed with computeCutsIDMMAPT(..., per_rater = TRUE).
digits: Number of decimal places shown in numeric table columns, from 0 to 15. Stored values retain their full precision.
...: Additional arguments. The result's print method forwards these to the summary print method; otherwise they are currently unused.

Details

Typing the result's name or calling print(result) displays pooled cuts, their status, and whether they fall outside the observed difficulty range. It also reports missing ratings removed during preparation.

A status of "ok" means a usable cut was obtained. A different status means that the cut is unavailable; see computeCutsIDMMAPT for the individual failure reasons. extrapolated = TRUE means that the fitted curve reaches probability 0.5 outside the observed item difficulty range. Such a cut is still returned, but is supported by extrapolation of the model.

The issues table contains every pooled or available individual fit with an unusable cut, an extrapolated cut, or a recorded numerical warning. The printed view shows at most the first 20 flagged fits; the table itself is complete. Pooled rows have scope = "pooled" and a missing person; individual rows have scope = "rater" and the rater ID. The display also flags pooled cuts that are not strictly increasing, or explains why their order cannot be checked.

Use summary(result, per_rater = TRUE) to compare individual rater cuts. These are auxiliary results and do not replace the pooled estimate. Models and input ratings remain available in the original result through result$models, result$models_per_person, and result$data_used. Summarising does not refit or modify the result.

Value

summary() returns an object of class summary.cutsIDMMAPT:

cuts_summary, cut_diagnostics: The original pooled cut table and complete pooled diagnostics.
issues: All flagged fits, with columns scope, person, cut_name, status, extrapolated, and warning.
n_ratings, n_items, n_raters: Counts of retained rating observations, distinct items, and distinct raters in data_used. Raters with only missing ratings are excluded from these counts, but can appear with no_data individual-fit diagnostics.
n_dropped: Number of missing ratings removed.
cuts_ordered: Whether all pooled cuts are strictly increasing in boundary order; NA if a pooled cut is unavailable.
per_rater_available, per_rater: Whether individual results were computed and whether they were requested in this summary.
cuts_per_person, diagnostics_per_person: Individual results when requested and available, otherwise NULL.
method: The method description from the original result.

Both print methods return their input invisibly.

Examples

dat <- data.frame(
  est = 1:12,
  RaterA = c(1, 2, 1, 1, 2, 1, 2, 1, 2, 2, 1, 2),
  RaterB = c(1, 1, 2, 1, 1, 2, 1, 2, 1, 2, 2, 2)
)
cuts <- computeCutsIDMMAPT(dat, boundaries = 1.5)
cuts
#> MAPT pooled binary logistic cuts (no rating trimming)
#> 24 ratings from 2 raters on 12 items; 0 missing ratings dropped.
#> 
#> Pooled cuts (all raters fitted together):
#>  cut_name cut status extrapolated
#>     cut12 6.5     ok        FALSE
#> 
#> No flagged fits.
#> Extrapolated = cut outside the observed item difficulty range.
#> Individual cuts: summary(x, per_rater = TRUE).
#> Models and ratings remain in the original result: $models, $models_per_person, $data_used.
s <- summary(cuts, per_rater = TRUE)
s
#> MAPT pooled binary logistic cuts (no rating trimming)
#> 24 ratings from 2 raters on 12 items; 0 missing ratings dropped.
#> 
#> Pooled cuts (all raters fitted together):
#>  cut_name cut status extrapolated
#>     cut12 6.5     ok        FALSE
#> 
#> No flagged fits.
#> Extrapolated = cut outside the observed item difficulty range.
#> 
#> Individual rater cuts (comparison results):
#>  person cut12
#>  RaterA   6.5
#>  RaterB   6.5
#> Models and ratings remain in the original result: $models, $models_per_person, $data_used.
s$cuts_summary
#> # A tibble: 1 × 1
#>   cut12
#>   <dbl>
#> 1   6.5
s$issues
#> # A tibble: 0 × 6
#> # ℹ 6 variables: scope <chr>, person <chr>, cut_name <chr>, status <chr>,
#> #   extrapolated <lgl>, warning <chr>
s$diagnostics_per_person
#> # A tibble: 2 × 17
#>   person cut_name boundary   cut     n n_lower n_upper min_est max_est intercept
#>   <chr>  <chr>       <dbl> <dbl> <int>   <int>   <int>   <int>   <int>     <dbl>
#> 1 RaterA cut12         1.5   6.5    12       6       6       1      12     -1.15
#> 2 RaterB cut12         1.5   6.5    12       6       6       1      12     -2.46
#> # ℹ 7 more variables: slope <dbl>, est_center <dbl>, est_scale <dbl>,
#> #   converged <lgl>, extrapolated <lgl>, status <chr>, warning <chr>
# Fitted models and the actual ratings stay in the original object.
stats::coef(cuts$models[[1]])
#>   (Intercept)    difficulty 
#> -1.359079e-16  9.386582e-01 
head(cuts$data_used)
#> # A tibble: 6 × 5
#>   item_id person   est stage_label stage_raw
#>   <chr>   <chr>  <int> <chr>           <dbl>
#> 1 1       RaterA     1 1                   1
#> 2 1       RaterB     1 1                   1
#> 3 2       RaterB     2 1                   1
#> 4 2       RaterA     2 2                   2
#> 5 3       RaterA     3 1                   1
#> 6 3       RaterB     3 2                   2

Usage

Arguments

Details

Value

See also

Examples