Test for Discretized Normality in Ordinal Data

Tests whether multivariate ordinal data may stem from discretizing a multivariate normal distribution. The test is described by Foldnes and Grønneberg (2019) . In addition, an adjusted polychoric correlation estimator is provided that takes marginal knowledge into account, as described by Grønneberg and Foldnes (2022) .

discnorm

This package contains an implementation of a the bootstrap test for underlying non-normality proposed by Foldnes and Gronneberg (Structural Equation Modeling, 2019). Also contains an adjusted polychoric estimator proposed by Gronneberg and Foldnes (Psychological Methods, 2022).

How to install

You can install:

• the stable release on CRAN:

  install.packages("discnorm")
  

• the latest development version:

  devtools::install_github("njaalf/discnorm")
  

Package overview

The package offers function bootTest() which tests an ordinal data frame for underlying normality. A function catLSadj() is provided that computes the adjusted polychoric correlations based on user-provided non-normal marginals.

References

Njål Foldnes & Steffen Grønneberg (2019) Pernicious Polychorics: The Impact and Detection of Underlying Non-normality, Structural Equation Modeling: A Multidisciplinary Journal, DOI: 10.1080/10705511.2019.1673168

Steffen Grønneberg & Njål Foldnes (2022) Factor Analyzing Ordinal Items Requires Substantive Knowledge of Response Marginals, Psychological Methods, DOI: 10.1037/met0000495