Fit the same MFRM to every completed rating data set

Fit a separate MFRM to each completed version of the ratings reviewed by review_mfrm_imputations(). All fits use the same model and measurement scale so eligible estimates can be combined with pool_mfrm_imputed(). Failed fits and their messages are retained for review.

Usage

fit_mfrm_imputed(x, model = c("RSM", "PCM"), step_facet = NULL, ...)

# S3 method for class 'mfrm_imputed_fits'
print(x, ...)

# S3 method for class 'mfrm_imputed_fits'
summary(object, ...)

Arguments

x

An review_mfrm_imputations() object.

model

"RSM" or "PCM".

step_facet

Required for PCM: the facet with separate step parameters.

...

Shared fit_mfrm() arguments, such as quad_points, anchors, facet_interactions or optimizer controls. They must be named. Data columns, category scale and MML identification are set by this workflow. Observation weights, latent regression, shrinkage, checkpointing and adaptive integration are not supported by this workflow.

object

An object returned by fit_mfrm_imputed().

Value

An mfrm_imputed_fits object with the imputations review, all fits (including NULL for failures), analysis_summary recording every fit's status, error and warnings, and common settings. Use pool_mfrm_imputed() for eligible non-person facet targets.

Details

Every fit uses the same original category ladder, facet levels, anchors and other supplied constraints, with fixed-standard-normal person MML identification. keep_original = TRUE prevents per-completion category collapsing. An unsupported category contrast, insufficient observed information or failed optimization is retained as a failed or ineligible analysis; subsequent pooling requires every imputation to qualify.

A common coordinate system does not establish model adequacy. Review the imputation model, rating design and numerical integration. The fit objects retain conditional person scores for individual review; those EAPs and posterior SDs are not ordinary complete-data parameter estimates and standard errors for Rubin pooling.

See also

review_mfrm_imputations(), pool_mfrm_imputed()