A Function That Fits and Examines Competing Dyadic Uni-construct Models

This function takes the outputted object from scrapeVarCross() along with the corresponding dataset and automatically tests competing uni-construct dyadic models for the latent variable under consideration. It inspects four possible model variants:

• Unidimensional (via scriptUni)

• Correlated Factors (via scriptCor)

• Hierarchical (via scriptHier)

• Bifactor (via scriptBifac)

Usage

getUniConstructMods(
  dvn,
  dat,
  indexes = c("npar", "df", "chisq", "cfi", "rmsea", "tli", "aic", "bic"),
  modelCompare = TRUE,
  genTEFI = TRUE,
  ...
)

Arguments

dvn

Input dvn list from scrapeVarCross().

dat

Input data frame containing the dataset for model estimation.

indexes

Input character vector specifying which index(es) to return. Default is c("npar", "df", "chisq", "cfi", "rmsea", "tli", "aic", "bic"). Note: Must be valid options from lavaan::fitMeasures().

modelCompare

A logical input indicating whether to compare fit statistics (i.e., chi-squared, AIC, and BIC) of the fitted models. The default is TRUE. Note: A scaled chi-squared difference test will automatically be applied (via lavaan::lavTestLRT()) if a robust estimator variant (e.g., estimator = "mlr") is used.

genTEFI

A logical input indicating whether to compute the Generalized Total Entropy Fit Index (genTEFI) (see Golino et al., 2024). Default is TRUE.

...

Additional arguments to be passed to lavaan::cfa(), allowing users to customize model estimation settings. By default, the models will be fit with robust maximum-likelihood (MLR) estimation—estimator = "mlr" —and missing data will handled using full-information maximum likelihood (FIML)—missing = "fiml".

Value

A list containing up to three components:

• Global Fit: A tibble with the desired index(es) for each fitted model (requested via the indexes argument).

• Model Comparisons: A tibble summarizing the differences in fit statistics between the fitted models. (if modelCompare = TRUE).

• genTEFI: A tibble of the genTEFI (if genTEFI = TRUE).

Examples

dvn <- scrapeVarCross(
  commitmentQ,
  x_order = "spi",
  x_stem = "sat.g",
  x_delim1 = ".",
  x_delim2="_",
  distinguish_1="1",
  distinguish_2="2"
  )
  
  getUniConstructMods(
  dvn,
  commitmentQ
  )
#> $`Global Fit`
#> # A tibble: 4 × 9
#>   Model               npar    df chisq   cfi  rmsea   tli   aic   bic
#>   <chr>              <dbl> <dbl> <dbl> <dbl>  <dbl> <dbl> <dbl> <dbl>
#> 1 Bifactor              45    20  41.3 0.983 0.0951 0.962 3953. 4077.
#> 2 Hierarchical          36    29  49.1 0.984 0.0767 0.975 3942. 4042.
#> 3 Correlated Factors    36    29  49.1 0.984 0.0767 0.975 3942. 4042.
#> 4 Unidimensional        35    30 235.  0.839 0.240  0.758 4126. 4223.
#> 
#> $`Model Comparisons`
#> # A tibble: 3 × 7
#>   Comparison                 chisq_diff df_diff   pvalue sig   aic_diff bic_diff
#>   <chr>                           <dbl>   <int>    <dbl> <chr>    <dbl>    <dbl>
#> 1 Hierarchical vs. Bifactor        5.14       9  8.22e-1 NA       -10.2    -35.2
#> 2 Correlated factors vs. Hi…       0          0 NA       NA         0        0  
#> 3 Unidimensional vs. Correl…      24.9        1  5.96e-7 ***      184.     181. 
#> 
#> $genTEFI
#> # A tibble: 1 × 3
#>   VN.Entropy.Fit Lower.Order.VN Higher.Order.VN
#>            <dbl>          <dbl>           <dbl>
#> 1          -15.2          -5.75           -9.42
#> 
  
  getUniConstructMods(
  dvn,
  commitmentQ,
  indexes = c("npar", "df", "bic"),
  modelCompare = FALSE,
  genTEFI = TRUE
  )
#> $`Global Fit`
#> # A tibble: 4 × 4
#>   Model               npar    df   bic
#>   <chr>              <dbl> <dbl> <dbl>
#> 1 Bifactor              45    20 4077.
#> 2 Hierarchical          36    29 4042.
#> 3 Correlated Factors    36    29 4042.
#> 4 Unidimensional        35    30 4223.
#> 
#> $genTEFI
#> # A tibble: 1 × 3
#>   VN.Entropy.Fit Lower.Order.VN Higher.Order.VN
#>            <dbl>          <dbl>           <dbl>
#> 1          -15.2          -5.75           -9.42
#> 
  
  getUniConstructMods(
  dvn,
  commitmentQ,
  indexes = c("npar", "df", "bic"),
  modelCompare = FALSE,
  genTEFI = TRUE,
  estimator = "ml"
  )
#> $`Global Fit`
#> # A tibble: 4 × 4
#>   Model               npar    df   bic
#>   <chr>              <dbl> <dbl> <dbl>
#> 1 Bifactor              45    20 4077.
#> 2 Hierarchical          36    29 4042.
#> 3 Correlated Factors    36    29 4042.
#> 4 Unidimensional        35    30 4223.
#> 
#> $genTEFI
#> # A tibble: 1 × 3
#>   VN.Entropy.Fit Lower.Order.VN Higher.Order.VN
#>            <dbl>          <dbl>           <dbl>
#> 1          -15.2          -5.75           -9.42
#>