empirical_reliability: Empirical Reliability Estimates

In autoFC: Automatic Toolkit for Construction, Optimization, Scoring and Simulation of Forced-Choice Tests

Empirical Reliability Estimates

Description

Calculates empirical reliability for IRT-based trait scores using the formula provided by Brown & Maydeu-Olivares (2018).

Usage

empirical_reliability(dataset, score_names, se_names = NULL)

Arguments

dataset	A data frame containing the trait estimates and standard errors.
score_names	Character vector. The names of the columns specifying trait scores.
se_names	Optional character vector. The names of the columns specifying trait standard errors. If NULL, the function automatically searches for columns named "[score_name]_SE".

Details

For trait scores estimated using item response theory (IRT) models, a single test-level reliability coefficient (like Cronbach's Alpha) is often inappropriate because the standard error of measurement varies across the latent continuum.

Empirical reliability provides a summary estimate of how reliable the trait scores are "as a whole" by comparing the variance of the estimated scores to the average error variance.

Value

A named numeric vector containing the empirical reliability estimates for each trait.

Author(s)

Mengtong Li

References

Brown, A., & Maydeu-Olivares, A. (2018). Ordinal factor analysis of graded-preference questionnaire data. Structural Equation Modeling, 25(4), 516-529. \Sexpr[results=rd]{tools:::Rd_expr_doi("10.1080/10705511.2017.1392247")}

Examples

# Create fake scores and standard errors
fake_scores <- data.frame(
  Trait1 = rnorm(100), Trait1_SE = runif(100, 0.1, 0.3),
  Trait2 = rnorm(100), Trait2_SE = runif(100, 0.2, 0.4)
)

# Auto-detects the "_SE" columns
empirical_reliability(fake_scores, score_names = c("Trait1", "Trait2"))

autoFC documentation built on July 14, 2026, 5:07 p.m.