Compute displacement diagnostics for facet levels

Usage

displacement_table(
  fit,
  diagnostics = NULL,
  facets = NULL,
  anchored_only = FALSE,
  abs_displacement_warn = 0.5,
  abs_t_warn = 2,
  top_n = NULL
)

Arguments

fit: Output from fit_mfrm().
diagnostics: Optional output from diagnose_mfrm().
facets: Optional subset of facets.
anchored_only: If TRUE, keep only directly/group anchored levels.
abs_displacement_warn: Absolute displacement warning threshold.
abs_t_warn: Absolute displacement t-value warning threshold.
top_n: Optional maximum number of rows to keep after sorting.

Value

A named list with:

table: displacement diagnostics by level
summary: one-row summary
thresholds: applied thresholds

Details

Displacement is computed as a one-step Newton update: sum(residual) / sum(information) for each facet level. This approximates how much a level would move if constraints were relaxed.

Interpreting output

table: level-wise displacement and flag indicators.
summary: count/share of flagged levels.
thresholds: displacement and t-value cutoffs.

Large absolute displacement in anchored levels suggests potential instability in anchor assumptions.

Typical workflow

Run displacement_table(fit, anchored_only = TRUE) for anchor checks.
Inspect summary(disp) then detailed rows.
Visualize with plot_displacement().

Output columns

The table data.frame contains:

Facet, Level: Facet name and element label.
Displacement: One-step Newton displacement estimate (logits).
DisplacementSE: Standard error of the displacement.
DisplacementT: Displacement / SE ratio.
Estimate, SE: Current measure estimate and its standard error.
N: Number of observations involving this level.
AnchorValue, AnchorStatus, AnchorType: Anchor metadata.
Flag: Logical; TRUE when displacement exceeds thresholds.

Examples

# \donttest{
toy <- load_mfrmr_data("example_core")
fit <- fit_mfrm(toy, "Person", c("Rater", "Criterion"), "Score", method = "JML", maxit = 30)
#> Warning: Optimization convergence review did not produce an inference-ready numerical solution (code = 1, status = iteration_limit). Optimizer reached the iteration limit before the terminal gradient became small enough for review-only acceptance. Inspect the model specification, data support, and starting values. Do not interpret estimates until the review is resolved.
disp <- displacement_table(fit, anchored_only = FALSE)
summary(disp)
#> mfrmr Displacement Summary 
#>   Class: mfrm_displacement
#>   Components: 3
#> 
#> Displacement summary
#>  Levels AnchoredLevels FlaggedLevels FlaggedAnchoredLevels MaxAbsDisplacement
#>      56              0             0                     0                  0
#>  MaxAbsDisplacementT AbsDisplacementThreshold AbsTThreshold
#>                0.001                      0.5             2
#> 
#> Displacement rows: table
#>   Facet        Level WeightedN ResidualSum Information Displacement
#>  Person <suppressed>        16      -0.002       9.452            0
#>  Person <suppressed>        16      -0.002       9.452            0
#>  Person <suppressed>        16      -0.002      10.333            0
#>  Person <suppressed>        16       0.001       6.760            0
#>  Person <suppressed>        16       0.001       4.966            0
#>  Person <suppressed>        16      -0.001       3.461            0
#>  Person <suppressed>        16       0.001       9.182            0
#>  Person <suppressed>        16      -0.001       4.840            0
#>  Person <suppressed>        16      -0.001       9.205            0
#>  Person <suppressed>        16      -0.001       9.205            0
#>  DisplacementSE DisplacementT Estimate    SE  N AnchorValue AnchorStatus
#>           0.325        -0.001    0.684 0.325 16          NA             
#>           0.325        -0.001    0.684 0.325 16          NA             
#>           0.311        -0.001   -0.015 0.311 16          NA             
#>           0.385         0.000   -1.665 0.385 16          NA             
#>           0.449         0.000   -2.178 0.449 16          NA             
#>           0.538         0.000    2.684 0.538 16          NA             
#>           0.330         0.000   -0.918 0.330 16          NA             
#>           0.455         0.000    2.200 0.455 16          NA             
#>           0.330         0.000    0.791 0.330 16          NA             
#>           0.330         0.000    0.791 0.330 16          NA             
#>  AnchorType ReleasedEstimate AnchorGap FlagDisplacement FlagT  Flag
#>        Free            0.684        NA            FALSE FALSE FALSE
#>        Free            0.684        NA            FALSE FALSE FALSE
#>        Free           -0.016        NA            FALSE FALSE FALSE
#>        Free           -1.665        NA            FALSE FALSE FALSE
#>        Free           -2.178        NA            FALSE FALSE FALSE
#>        Free            2.684        NA            FALSE FALSE FALSE
#>        Free           -0.918        NA            FALSE FALSE FALSE
#>        Free            2.199        NA            FALSE FALSE FALSE
#>        Free            0.791        NA            FALSE FALSE FALSE
#>        Free            0.791        NA            FALSE FALSE FALSE
#> 
#> Settings
#>                Setting Value
#>  abs_displacement_warn   0.5
#>             abs_t_warn     2
#> 
#> Notes
#>  - Displacement summary for anchor drift and baseline drift checks.
#>  - Person identifiers are suppressed in this summary. Use `include_person =
#>    TRUE` only under appropriate privacy controls.
p_disp <- plot(disp, draw = FALSE)
p_disp$data$plot
#> [1] "lollipop"
# }