Title: Compare Nested and Non-Nested Structural Equation Models
Version: 0.2.6
Date: 2026-04-23
Description: A comprehensive package for comparing multiple Structural Equation Models (SEM). Supports both nested and non-nested model comparisons, chi-square difference tests, and extraction of multiple fit indices including AIC (Akaike Information Criterion), BIC (Bayesian Information Criterion), CFI (Comparative Fit Index), TLI (Tucker-Lewis Index), RMSEA (Root Mean Square Error of Approximation), and SRMR (Standardized Root Mean Square Residual). Built on top of the 'lavaan' package for seamless SEM model comparison workflows. The Vuong test (Vuong, 1989) for non-nested models is used as the statistical test.
License: GPL (≥ 3)
Encoding: UTF-8
RoxygenNote: 7.3.3
Imports: lavaan (≥ 0.6), stats, boot, ggplot2, nonnest2, tidyr
Suggests: testthat (≥ 3.0.0)
URL: https://github.com/ssjerf-stack/modelscompete4
BugReports: https://github.com/ssjerf-stack/modelscompete4/issues
NeedsCompilation: no
Packaged: 2026-04-22 18:45:29 UTC; jerf
Author: Jerf Yeung [aut, cre]
Maintainer: Jerf Yeung <ssjerf@gmail.com>
Repository: CRAN
Date/Publication: 2026-04-24 20:10:02 UTC

modelscompete4: Advanced Model Comparison for Latent Variable Models

Description

The modelscompete4 package provides advanced tools for comparing latent variable models, including bootstrap comparisons, fit index extraction, and visualization functions.

Main functions

compare_latent_models

Compare multiple latent variable models

extract_latent_parameters

Extract parameters from lavaan models

plot_latent_comparison

Visualize model comparison results

bootstrap_lavaan_comparison

Bootstrap-based model comparison

Author(s)

Maintainer: Jerf Yeung ssjerf@gmail.com

See Also

Useful links:

Package startup message

Description

Package startup message

Usage

.onAttach(libname, pkgname)

Arguments

libname

library location

pkgname

package name

Value

No return value, called for side effects.

Package load

Description

Package load

Usage

.onLoad(libname, pkgname)

Arguments

libname

library location

pkgname

package name

Value

No return value, called for side effects.

Bootstrap Comparison for Lavaan Models

Description

Perform bootstrap-based comparison of lavaan models

Usage

bootstrap_lavaan_comparison(
  model1,
  model2,
  R = 1000,
  parallel = "no",
  ncpus = 1
)

Arguments

model1

First lavaan model

model2

Second lavaan model

R

Number of bootstrap replications (default: 1000)

parallel

Type of parallel processing (if any)

ncpus

Number of CPUs to use for parallel processing

Value

A list containing bootstrap results

Examples


library(lavaan)
model1 <- 'F1 =~ x1 + x2 + x3'
model2 <- 'F1 =~ x1 + x2 + x3 + x4'
fit1 <- cfa(model1, data = HolzingerSwineford1939)
fit2 <- cfa(model2, data = HolzingerSwineford1939)
boot_result <- bootstrap_lavaan_comparison(fit1, fit2, R = 100)
print(boot_result)

Compare Latent Variable Models

Description

Compare Latent Variable Models

Usage

compare_latent_models(
  ...,
  nested = FALSE,
  fit_measures = c("chisq", "df", "pvalue", "cfi", "tli", "rmsea", "srmr"),
  method = "default",
  verbose = TRUE
)

Arguments

...

lavaan model objects

nested

logical, whether models are nested

fit_measures

character vector of fit measures to extract

method

comparison method

verbose

Logical; if TRUE, progress messages are printed.

Value

A latent_comparison object

Compare Multiple Nested or Non-Nested Structural Equation Models

Description

This is the core function of the modelscompete4 package. It automatically fits a list of SEM models, determines their nesting relationship, and performs the appropriate statistical comparison (chi-square difference test for nested models, Vuong test for non-nested models).

Usage

compare_models(
  model_list,
  data,
  estimator = "ML",
  se = "standard",
  bootstrap = 1000,
  parallel = "no",
  verbose = TRUE,
  ...
)

Arguments

model_list

A named list. Each element is a character string specifying the model syntax in lavaan format.

data

A data.frame containing the observed variables used in the models.

estimator

The estimator to be used (e.g., "ML"). Passed to sem.

se

Type of standard errors. Default is "standard". Use "bootstrap" for bootstrapped standard errors and confidence intervals.

bootstrap

Number of bootstrap draws if se="bootstrap". Default is 1000.

parallel

Method for parallel processing for bootstrapping ("multicore", "snow", or "no"). Recommended for large samples.

verbose

Logical; if TRUE, progress messages are printed.

...

Additional arguments passed to sem.

Value

An object of class modelscompete4. This is a list containing:

Advanced Model Comparison with Latent Variable Support

Description

Advanced Model Comparison with Latent Variable Support

Usage

compare_models_advanced_lv(
  models,
  model_names = NULL,
  model_types = NULL,
  criteria = c("AIC", "BIC", "CFI", "TLI", "RMSEA", "SRMR"),
  latent_indicators = NULL,
  bootstrap = FALSE,
  n_bootstrap = 1000
)

Arguments

models

List of model objects (lm, lavaan)

model_names

Character vector of model names

model_types

Character vector of model types ("lm", "lavaan")

criteria

Criteria to calculate

latent_indicators

List of latent variable indicators (for lavaan)

bootstrap

Logical, whether to perform bootstrapping

n_bootstrap

Number of bootstrap replications

Value

Comparison results with latent variable support

Extract Latent Variable Fit Indices

Description

Extract Latent Variable Fit Indices

Usage

extract_latent_fit(model)

Arguments

model

lavaan model object

Value

Comprehensive fit indices for latent variable model

Extract Parameters from Lavaan Models

Description

This function extracts parameters (loadings, variances, etc.) from a lavaan model. P-values are formatted appropriately (e.g., <0.001 for very small values).

Usage

extract_latent_parameters(
  model,
  type = "loadings",
  standardized = FALSE,
  digits = 3,
  ...
)

Arguments

model

A fitted lavaan model.

type

Type of parameters to extract: "loadings", "variances", or "all". Default is "loadings".

standardized

Logical; if TRUE, returns standardized estimates. Default is FALSE.

digits

Number of decimal places for p-value formatting (default=3)

...

Additional arguments passed to lavaan::parameterEstimates or lavaan::standardizedSolution.

Value

A data frame containing the extracted parameters with formatted p-values.

Examples


library(lavaan)
model <- 'F1 =~ x1 + x2 + x3'
fit <- cfa(model, data = HolzingerSwineford1939)
extract_latent_parameters(fit, type = "loadings")

Format confidence intervals

Description

This function formats confidence intervals into a readable string.

Usage

format_ci(lower, upper, digits = 3)

Arguments

lower

Lower bound of CI

upper

Upper bound of CI

digits

Number of decimal places (default=3)

Value

Formatted CI string

Format p-values for display

Description

This internal function formats p-values for display, converting very small values (e.g., < 0.001) to appropriate string representations.

Usage

format_pvalues(pvals, digits = 3)

Arguments

pvals

Vector of p-values

digits

Number of digits (default=3)

Value

Formatted p-values as character strings

Plot Latent Model Comparison Results

Description

Creates visualization of model comparison results

Usage

plot_latent_comparison(x, type = "fit", ...)

Arguments

x

An object of class 'modelscompete4' or 'latent_comparison'

type

Type of plot: 'fit' for fit indices, 'diff' for differences

...

Additional arguments passed to plotting functions

Value

A ggplot object (if ggplot2 and tidyr are available), otherwise NULL

Print method for latent_comparison objects

Description

Prints a summary of latent model comparison results.

Usage

## S3 method for class 'latent_comparison'
print(x, digits = 3, ...)

## S3 method for class 'latent_comparison'
print(x, digits = 3, ...)

Arguments

x

An object of class 'latent_comparison'

digits

Number of digits to display (default: 3)

...

Additional arguments passed to print method

Value

Invisibly returns the input object.

Invisibly returns the input object

Print Method for modelscompete4_advanced Objects

Description

Print Method for modelscompete4_advanced Objects

Usage

## S3 method for class 'modelscompete4_advanced'
print(x, ...)

Arguments

x

A modelscompete4_advanced object

...

Additional arguments passed to print

Value

Invisibly returns the input object x.