strengejacke
commented
3 years ago I added the "definition" parameters now as well:
library(lavaan)
#> This is lavaan 0.6-7
#> lavaan is BETA software! Please report any bugs.
library(parameters)
set.seed(1234)
X <- rnorm(100)
M <- 0.5*X + rnorm(100)
Y <- 0.7*M + rnorm(100)
Data <- data.frame(X = X, Y = Y, M = M)
model <- ' # direct effect
Y ~ c*X
# mediator
M ~ a*X
Y ~ b*M
# indirect effect (a*b)
ab := a*b
# total effect
total := c + (a*b)
'
fit <- sem(model, data = Data, test = "Satorra-Bentler")
summary(fit)
#> lavaan 0.6-7 ended normally after 12 iterations
#>
#> Estimator ML
#> Optimization method NLMINB
#> Number of free parameters 5
#>
#> Number of observations 100
#>
#> Model Test User Model:
#> Standard Robust
#> Test Statistic 0.000 0.000
#> Degrees of freedom 0 0
#>
#> Parameter Estimates:
#>
#> Standard errors Standard
#> Information Expected
#> Information saturated (h1) model Structured
#>
#> Regressions:
#> Estimate Std.Err z-value P(>|z|)
#> Y ~
#> X (c) 0.036 0.104 0.348 0.728
#> M ~
#> X (a) 0.474 0.103 4.613 0.000
#> Y ~
#> M (b) 0.788 0.092 8.539 0.000
#>
#> Variances:
#> Estimate Std.Err z-value P(>|z|)
#> .Y 0.898 0.127 7.071 0.000
#> .M 1.054 0.149 7.071 0.000
#>
#> Defined Parameters:
#> Estimate Std.Err z-value P(>|z|)
#> ab 0.374 0.092 4.059 0.000
#> total 0.410 0.125 3.287 0.001
model_parameters(fit, standardize = TRUE)
#> # Regression type
#>
#> Link | Coefficient | SE | 95% CI | p
#> ---------------------------------------------------
#> Y ~ X | 0.03 | 0.08 | [-0.13, 0.18] | 0.728
#> M ~ X | 0.42 | 0.08 | [ 0.26, 0.57] | < .001
#> Y ~ M | 0.68 | 0.06 | [ 0.56, 0.80] | < .001
#>
#> # Defined type
#>
#> Link | Coefficient | SE | 95% CI | p
#> -------------------------------------------------------------
#> ab := a*b | 0.28 | 0.06 | [0.16, 0.41] | < .001
#> total := c+(a*b) | 0.31 | 0.09 | [0.14, 0.48] | < .001Created on 2020-08-26 by the reprex package (v0.3.0)
Not sure, though, if the output still lacks some useful information? Is there more you would add to the output?
First of all, this is more of a feature request, not a exaclty a blocker for acceptance.
In the JOSS Paper lavaan::sem() is listed in the image of modeling methods supported by the parameters-package. The vignette https://stat.ethz.ch/CRAN/web/packages/parameters/vignettes/efa_cfa.html offers an explanation how parameters() works for EFA and CFA, but keeps quiet of lavaan::sem(). Structural equation models have an enormous amount of modeling possibilities. For example, it is possible to estimate indirect effects in partial mediation structures. For this, the effects have to be defined by the ':=' operator. At the moment, parameters does not extract those.
I think, having the possibility to include those effects in the parameters-package would be a useful addition. Concerning the image in the JOSS-paper with the supported models, i would suggest to empasize the support of lavaan::cfa() instead of lavaan::sem() at the moment, because the vignette is pretty silent about lavaan::sem() and some output ist not yet available in the parameters package.