Build an unexpected-after-adjustment screening report

Usage

unexpected_after_bias_table(
  fit,
  bias_results,
  diagnostics = NULL,
  abs_z_min = 2,
  prob_max = 0.3,
  top_n = 100,
  rule = c("either", "both")
)

Arguments

fit: Output from fit_mfrm().
bias_results: Output from estimate_bias().
diagnostics: Optional output from diagnose_mfrm() for baseline comparison.
abs_z_min: Absolute standardized-residual cutoff.
prob_max: Maximum observed-category probability cutoff.
top_n: Maximum number of rows to return.
rule: Flagging rule: "either" or "both".

Value

A named list with:

table: unexpected responses after bias adjustment
summary: one-row summary (includes baseline-vs-after counts)
thresholds: applied thresholds
facets: analyzed bias facet pair

Details

This helper recomputes expected values and residuals after interaction adjustments from estimate_bias() have been introduced.

summary(t10) is supported through summary(). plot(t10) is dispatched through plot() for class mfrm_unexpected_after_bias (type = "scatter", "severity", "comparison").

Interpreting output

summary: before/after unexpected counts and reduction metrics.
table: residual unexpected responses after bias adjustment.
thresholds: screening settings used in this comparison.

Large reductions indicate bias terms explain part of prior unexpectedness; persistent unexpected rows indicate remaining model-data mismatch.

Typical workflow

Run unexpected_response_table() as baseline.
Estimate bias via estimate_bias().
Run unexpected_after_bias_table(...) and compare reductions.

Further guidance

For a plot-selection guide and a longer walkthrough, see mfrmr_visual_diagnostics and vignette("mfrmr-visual-diagnostics", package = "mfrmr").

Output columns

The table data.frame has the same structure as unexpected_response_table() output, with an additional BiasAdjustment column showing the bias correction applied to each observation's expected value.

The summary data.frame contains:

TotalObservations: Total observations analyzed.
BaselineUnexpectedN: Unexpected count before bias adjustment.
AfterBiasUnexpectedN: Unexpected count after adjustment.
ReducedBy, ReducedPercent: Reduction in unexpected count.

Examples

toy <- load_mfrmr_data("example_bias")
fit <- fit_mfrm(toy, "Person", c("Rater", "Criterion"), "Score", method = "JML", maxit = 25)
diag <- diagnose_mfrm(fit, residual_pca = "none")
bias <- estimate_bias(fit, diag, facet_a = "Rater", facet_b = "Criterion", max_iter = 2)
t10 <- unexpected_after_bias_table(fit, bias, diagnostics = diag, top_n = 20)
summary(t10)
#> mfrmr Unexpected-after-Bias Summary 
#>   Class: mfrm_unexpected_after_bias
#>   Components (4): table, summary, thresholds, facets
#> 
#> After-bias threshold summary
#>  TotalObservations UnexpectedN UnexpectedPercent LowProbabilityN LargeResidualN
#>                384          20             5.208              20              9
#>    Rule AbsZThreshold ProbThreshold BaselineUnexpectedN AfterBiasUnexpectedN
#>  either             2           0.3                  88                   20
#>  ReducedBy ReducedPercent
#>         68         77.273
#> 
#> After-bias flagged rows: table
#>  Row Person Rater    Criterion Weight Score Observed Expected Residual
#>  343   P043   R01     Accuracy      1     1        1    3.191   -2.191
#>  136   P017   R04     Accuracy      1     3        3    1.451    1.549
#>  279   P035   R02     Accuracy      1     2        2    3.501   -1.501
#>  254   P032   R03     Language      1     4        4    2.216    1.784
#>   31   P004   R02     Accuracy      1     3        3    1.496    1.504
#>  131   P017   R02 Organization      1     1        1    2.717   -1.717
#>  110   P014   R03     Language      1     2        2    3.397   -1.397
#>  215   P027   R01     Accuracy      1     2        2    3.363   -1.363
#>  135   P017   R02     Accuracy      1     4        4    2.471    1.529
#>  269   P034   R02     Language      1     1        1    2.500   -1.500
#>  StdResidual ObsProb MostLikely MostLikelyProb CategoryGap Surprise
#>       -3.148   0.008          3          0.505           2    2.072
#>        2.630   0.046          1          0.598           2    1.335
#>       -2.491   0.052          4          0.556           2    1.284
#>        2.343   0.037          2          0.486           2    1.427
#>        2.467   0.056          1          0.563           2    1.251
#>       -2.262   0.048          3          0.487           2    1.320
#>       -2.179   0.077          4          0.480           2    1.114
#>       -2.091   0.086          4          0.456           2    1.065
#>        1.990   0.077          2          0.419           2    1.114
#>       -1.954   0.087          3          0.422           2    1.061
#>             Direction FlagLowProbability FlagLargeResidual Severity
#>   Lower than expected               TRUE              TRUE    6.220
#>  Higher than expected               TRUE              TRUE    4.964
#>   Lower than expected               TRUE              TRUE    4.775
#>  Higher than expected               TRUE              TRUE    4.771
#>  Higher than expected               TRUE              TRUE    4.718
#>   Lower than expected               TRUE              TRUE    4.581
#>   Lower than expected               TRUE              TRUE    4.292
#>   Lower than expected               TRUE              TRUE    4.155
#>  Higher than expected               TRUE             FALSE    4.105
#>   Lower than expected               TRUE             FALSE    4.014
#>  BiasAdjustment
#>           0.776
#>          -1.103
#>           0.246
#>           0.155
#>           0.246
#>          -0.023
#>           0.155
#>           0.776
#>           0.246
#>          -0.209
#> 
#> Settings
#>    Setting  Value
#>  abs_z_min      2
#>   prob_max    0.3
#>       rule either
#> 
#> Notes
#>  - Unexpected-response summary after interaction adjustment.Bias interaction: Rater x Criterion x c("Rater", "Criterion") x 2 x pairwise
p_t10 <- plot(t10, draw = FALSE)
class(p_t10)
#> [1] "mfrm_plot_data" "list"