Synthetic many-facet rating examples — mfrmr_example

Compact synthetic many-facet datasets sized to exercise the documented workflow. Their dimensions are not evidence of sample-size adequacy for an applied study.

Format

A data.frame with 6 columns:

Study: Example dataset label ("OperationalExample", "ExampleCore", or "ExampleBias").
Person: Person/respondent identifier.
Rater: Rater identifier.
Criterion: Criterion facet label.
Score: Observed category score on a four-category scale (1–4).
Group: Balanced grouping label ("A" / "B"). It is neutral in the operational and core examples; the bias example has the planted group structure described below.

Source

Synthetic documentation data generated from rating-scale Rasch facet designs with fixed seeds. Generator scripts are maintained under data-raw/ for this release in the public source repository. These are synthetic examples, not empirical records.

Details

Available data objects:

mfrmr_example_operational
mfrmr_example_operational_design (documented separately below)
mfrmr_example_core
mfrmr_example_bias

mfrmr_example_operational is the primary applied teaching example. It has a connected but incomplete two-rater assignment, moderately unequal rater workloads, and six planned criterion-level omissions. Scores are sampled directly from stated RSM category probabilities. The six unobserved ratings are absent rows in the long data rather than NA scores. Each group contains 24 persons, and three omissions in each group leave 141 observed rows for Group A and 141 for Group B. The balanced Group variable has no effect in the generating model; random observed differences may still occur.

mfrmr_example_core is an idealized complete crossing generated from a single latent trait plus rater and criterion main effects. It is useful as a fast deterministic example, but it is not representative of routine incomplete operational assignment.

mfrmr_example_bias instead uses a balanced partial two-rater assignment. Group A and B latent means are -0.1 and 0.1 logits, respectively, with a common SD of 0.9. It also contains:

a planted Group x Criterion effect (Group B is advantaged by 1.2 logits on Language)
a planted Rater x Criterion interaction (R04 x Accuracy lowers the linear predictor by 1.2 logits)

This lets differential-functioning and bias-analysis help pages demonstrate non-null findings.

Data dimensions

Dataset	Rows	Persons	Raters	Criteria	Groups
example_operational	282	48	6	3	2
example_core	768	48	4	4	2
example_bias	384	48	4	4	2

Suggested usage

Use mfrmr_example_operational for the beginner data-to-report workflow and for inspecting incomplete but connected assignment.
Use mfrmr_example_core for fast, idealized checks and examples that specifically require complete crossing.
Use mfrmr_example_bias for analyze_dff(), analyze_dif(), dif_interaction_table(), plot_dif_heatmap(), and estimate_bias().

All three objects can be loaded either with load_mfrmr_data() or directly with data(), for example data("mfrmr_example_operational", package = "mfrmr").

Examples

data("mfrmr_example_operational", package = "mfrmr")
table(mfrmr_example_operational$Score)
#> 
#>  1  2  3  4 
#> 62 96 78 46 
table(mfrmr_example_operational$Group)
#> 
#>   A   B 
#> 141 141