jsakaluk/gclm

GCLM

GCLM

The goal of GCLM is to help you write lavaan() script for fitting generalized cross-lagged panel models (GCLM; Zyphur et al., 2019). Currently, there is only one function for fitting the base GCLM, but new functions and arguments may be added to automate the scripting of short-run Granger-Sims tests, and long-run impulse responses.

Installation

You can install the development version from GitHub with:

# install.packages("devtools")
devtools::install_github("jsakaluk/GCLM")

Example

This is a basic example which shows you how to solve a common problem:

library(GCLM)
library(lavaan)
#> This is lavaan 0.6-3
#> lavaan is BETA software! Please report any bugs.
#Save/export script for GCLM between Sacrifice and Satisfaction over 5 waves
base.model <- gclm(xvar = "sacrifice", yvar = "satisfaction", waves = 5)

The function saves a lavaan() script for you to immediately submit to its model-fitting functions (e.g., lavaan::sem()), and also outputs a .txt document of the same script to the working directory.

base.model.fit <- sem(base.model, data = dat)
summary(base.model.fit)

Eventually, new functionality may be programmed to automate the tests of short-run and long-run effects. Until then, users can manually ammend this script and save as a new lavaan() script to conduct these additional tests.

gclm's Issues

Restricted Contemporaneous Covariances

Hi, John! I'm looking at the snapshot of expanded lavaan code shown on the package's main page. In the "restrictions" section, it appears that some of the contemporaneous covariances among the impulses are fixed to zero. This seems like it would overwrite the freely estimated covariances listed earlier in the "co-movements" section. Perhaps this is a typo?