
Summarize an mfrm_bias object in a user-friendly format
Source: R/api-methods.R
summary.mfrm_bias.RdSummarize an mfrm_bias object in a user-friendly format
Usage
# S3 method for class 'mfrm_bias'
summary(object, digits = 3, top_n = 10, p_cut = 0.05, ...)Arguments
- object
Output from
estimate_bias().- digits
Number of digits for printed numeric values.
- top_n
Number of strongest bias rows to keep.
- p_cut
Tail-area cutoff used for counting screen-positive rows.
- ...
Reserved for generic compatibility.
Value
An object of class summary.mfrm_bias with:
overview: interaction facets/order, cell counts, and effect-size profilechi_sq: fixed-effect chi-square blockfinal_iteration: end-of-iteration status rowtop_rows: highest-|t|interaction rowsnotes: short interpretation notes
Details
This method returns a compact interaction-bias summary:
interaction facets/order and analyzed cell counts
effect-size profile (
|bias|mean/max, screen-positive cell count)fixed-effect chi-square block
iteration-end convergence indicators
top rows ranked by absolute t
Interpreting output
overview: interaction order, analyzed cells, and effect-size profile.chi_sq: fixed-effect test block.final_iteration: end-of-loop status from the bias routine.top_rows: strongest bias contrasts by|t|;GPCMsummaries also retain the profile-likelihood review columns when present.
Typical workflow
Estimate interactions with
estimate_bias().Check
summary(bias)for screen-positive and unstable cells.Use
bias_interaction_report()or plot(x, type = "bias") for details.
Examples
toy <- load_mfrmr_data("example_bias")
toy <- toy[toy$Person %in% unique(toy$Person)[1:8], ]
fit <- fit_mfrm(toy, "Person", c("Rater", "Criterion"), "Score", method = "JML", maxit = 30)
#> Warning: Category support is retained but requires review: at least one fitted or local scope contains an empty or singleton category/transition cell. The fit may be inspected, but category-information strength has not been certified; inspect `fit$data_review$category_support` before inference.
diag <- diagnose_mfrm(fit, residual_pca = "none")
bias <- estimate_bias(fit, diag, facet_a = "Rater", facet_b = "Criterion", max_iter = 1)
summary(bias)
#> Many-Facet Measurement Bias Summary
#> Interaction facets: Rater x Criterion | Cells: 16
#> Order: 2 | Mode: pairwise
#> Mean |Bias|: 0.304 | Max |Bias|: 0.909 | Screen-positive (p <= 0.050): 0
#> Bonferroni screen-positive: 0 | Holm screen-positive: 0 (cut = 0.050, m = 16)
#>
#> Fixed-effect chi-square
#> FixedChiSq FixedDF FixedProb InferenceTier SupportsFormalInference
#> 3.521 16 1 screening FALSE
#> FormalInferenceEligible PrimaryReportingEligible ReportingUse
#> FALSE FALSE screening_only
#> TestBasis Cells AvailableCells
#> conditional plug-in zero-reference screen 16 16
#> Status
#> available_screening
#> Reason
#> Conditional all-zero screen; not a formal joint interaction test.
#> InteractionFacets InteractionOrder InteractionMode
#> Rater x Criterion 2 pairwise
#>
#> Final iteration status
#> Iteration MaxScoreResidual MaxScoreResidualPct MaxScoreResidualCategories
#> 1 0 0 NA
#> MaxLogitChange BiasCells
#> -0.909 16
#>
#> Top |t| bias rows
#> Pair Rater Criterion Bias Size S.E. t Prob.
#> R04 | Accuracy R04 Accuracy -0.909 0.929 -0.979 0.431
#> R04 | Content R04 Content 0.759 0.982 0.773 0.520
#> R04 | Organization R04 Organization 0.759 0.982 0.773 0.520
#> R01 | Organization R01 Organization -0.384 0.624 -0.615 0.572
#> R02 | Content R02 Content -0.473 0.783 -0.604 0.588
#> R04 | Language R04 Language -0.470 0.929 -0.506 0.663
#> R02 | Accuracy R02 Accuracy 0.248 0.735 0.337 0.758
#> R01 | Language R01 Language 0.203 0.624 0.325 0.761
#> R01 | Accuracy R01 Accuracy 0.163 0.643 0.253 0.813
#> R03 | Accuracy R03 Accuracy 0.134 0.823 0.163 0.881
#> Obs-Exp Average d.f. DFUnrounded
#> -0.348 2 2
#> 0.265 2 2
#> 0.265 2 2
#> -0.192 4 4
#> -0.204 3 3
#> -0.182 2 2
#> 0.112 3 3
#> 0.105 4 4
#> 0.080 4 4
#> 0.049 3 3
#> ProbabilityReference
#> Two-sided Student t at continuous count-minus-one df; calibration held fixed
#> Two-sided Student t at continuous count-minus-one df; calibration held fixed
#> Two-sided Student t at continuous count-minus-one df; calibration held fixed
#> Two-sided Student t at continuous count-minus-one df; calibration held fixed
#> Two-sided Student t at continuous count-minus-one df; calibration held fixed
#> Two-sided Student t at continuous count-minus-one df; calibration held fixed
#> Two-sided Student t at continuous count-minus-one df; calibration held fixed
#> Two-sided Student t at continuous count-minus-one df; calibration held fixed
#> Two-sided Student t at continuous count-minus-one df; calibration held fixed
#> Two-sided Student t at continuous count-minus-one df; calibration held fixed
#> ProbabilityStatus
#> available_conditional
#> available_conditional
#> available_conditional
#> available_conditional
#> available_conditional
#> available_conditional
#> available_conditional
#> available_conditional
#> available_conditional
#> available_conditional
#> ProbabilityReason
#> The tail conditions on the fitted calibration; count-minus-one is a screening approximation, not a calibrated test.
#> The tail conditions on the fitted calibration; count-minus-one is a screening approximation, not a calibrated test.
#> The tail conditions on the fitted calibration; count-minus-one is a screening approximation, not a calibrated test.
#> The tail conditions on the fitted calibration; count-minus-one is a screening approximation, not a calibrated test.
#> The tail conditions on the fitted calibration; count-minus-one is a screening approximation, not a calibrated test.
#> The tail conditions on the fitted calibration; count-minus-one is a screening approximation, not a calibrated test.
#> The tail conditions on the fitted calibration; count-minus-one is a screening approximation, not a calibrated test.
#> The tail conditions on the fitted calibration; count-minus-one is a screening approximation, not a calibrated test.
#> The tail conditions on the fitted calibration; count-minus-one is a screening approximation, not a calibrated test.
#> The tail conditions on the fitted calibration; count-minus-one is a screening approximation, not a calibrated test.
#> BiasStatus ScreenStatus
#> interior available_conditional
#> interior available_conditional
#> interior available_conditional
#> interior available_conditional
#> interior available_conditional
#> interior available_conditional
#> interior available_conditional
#> interior available_conditional
#> interior available_conditional
#> interior available_conditional
#> ScreenReason
#> Conditional information uses all cell rows; calibration and Person uncertainty are excluded.
#> Conditional information uses all cell rows; calibration and Person uncertainty are excluded.
#> Conditional information uses all cell rows; calibration and Person uncertainty are excluded.
#> Conditional information uses all cell rows; calibration and Person uncertainty are excluded.
#> Conditional information uses all cell rows; calibration and Person uncertainty are excluded.
#> Conditional information uses all cell rows; calibration and Person uncertainty are excluded.
#> Conditional information uses all cell rows; calibration and Person uncertainty are excluded.
#> Conditional information uses all cell rows; calibration and Person uncertainty are excluded.
#> Conditional information uses all cell rows; calibration and Person uncertainty are excluded.
#> Conditional information uses all cell rows; calibration and Person uncertainty are excluded.
#> FitStatus
#> available_descriptive
#> available_descriptive
#> available_descriptive
#> available_descriptive
#> available_descriptive
#> available_descriptive
#> available_descriptive
#> available_descriptive
#> available_descriptive
#> available_descriptive
#> FitReason
#> Descriptive mean squares; their calculation does not establish inferential validity.
#> Descriptive mean squares; their calculation does not establish inferential validity.
#> Descriptive mean squares; their calculation does not establish inferential validity.
#> Descriptive mean squares; their calculation does not establish inferential validity.
#> Descriptive mean squares; their calculation does not establish inferential validity.
#> Descriptive mean squares; their calculation does not establish inferential validity.
#> Descriptive mean squares; their calculation does not establish inferential validity.
#> Descriptive mean squares; their calculation does not establish inferential validity.
#> Descriptive mean squares; their calculation does not establish inferential validity.
#> Descriptive mean squares; their calculation does not establish inferential validity.
#> AbsT
#> 0.979
#> 0.773
#> 0.773
#> 0.615
#> 0.604
#> 0.506
#> 0.337
#> 0.325
#> 0.253
#> 0.163
#>
#> Notes
#> - Conditional plug-in screening holds calibration and Person estimates fixed. Bonferroni/Holm summaries retain all reported cells in this pair's family; they do not establish formal family-wise error control or collection-wide multiplicity control.
#> - Bias iteration may not have fully stabilized (BiasCells > 0 at final step).