peterdalle/semproducible

Generates R code for your lavaan models automatically, which helps with reproducible open science

Semproducible reproduce a lavaan model

Semproducible is an R package that makes your latent variable models in lavaan reproducible. Semproducible generates all the necessary data and R code from your existing lavaan model.

This package makes it possible to reproduce a model without disclosing the raw data (which might be sensitive).

You can use semproducible in two ways:

1. Generate reproducible code from a lavaan model
2. Generate reproducible code from a data frame

Benefits

• Only one line of code to reproduce your model
• The generated code fit a journal appendix
• Researchers can reproduce your model (and try alternative model specifications)

Install

devtools::install_github("peterdalle/semproducible")

Usage

semproducible(x, formula)

Where x is either a lavaan model or a data frame.

And formula is the lavaan formula syntax (e.g., ind60 =~ x1 + x2 + x3).

Examples

1. Generate code from a lavaan model

Let's say you have the following lavaan model, and you want to make it reproducible:

library(lavaan)

# Example: http://lavaan.ugent.be/tutorial/sem.html
formula <- "# latent variables
              ind60 =~ x1 + x2 + x3
              dem60 =~ y1 + y2 + y3 + y4
              dem65 =~ y5 + y6 + y7 + y8
            # regressions
              dem60 ~ ind60
              dem65 ~ ind60 + dem60
            # residual covariances
              y1 ~~ y5
              y2 ~~ y4 + y6
              y3 ~~ y7
              y4 ~~ y8
              y6 ~~ y8"

fit <- sem(formula, data=PoliticalDemocracy)

The only thing you need to do is to pass fit and formula into semproducible:

library(semproducible)

semproducible(fit, formula)

2. Generate code from a data frame

If you have a data frame with many variables, you can reproduce the entire data frame and let other researchers choose variables to model.

library(semproducible)
library(tidyverse)

iris %>% 
  select(Sepal.Length, Sepal.Width, Petal.Length, Petal.Width) %>% 
  semproducible(formula = "Sepal.Length ~ Sepal.Width + Petal.Length")

3. Save the generated code to a file

code <- semproducible(fit, formula)

save_code(code, "filename.R")

What the generated code looks like

If you run example 2 above, the generated code will look like this:

library(tibble)
library(lavaan)

# Number of observations.
observations <- 150

# Covariance matrix.
cov_mat <- tribble(~Sepal.Length, ~Sepal.Width, ~Petal.Length, ~Petal.Width,
              0.685693512304251, -0.0424340044742729, 1.27431543624161, 0.516270693512304,               
              -0.0424340044742729, 0.189979418344519, -0.329656375838926, -0.12163937360179,               
              1.27431543624161, -0.329656375838926, 3.11627785234899, 1.29560939597315,               
              0.516270693512304, -0.12163937360179, 1.29560939597315, 0.581006263982103)

# Convert data frame to matrix (that lavaan can handle).
cov_mat <- as.matrix(cov_mat)

# Rows should have names too.
rownames(cov_mat) <- colnames(cov_mat)

# SEM model in lavaan syntax.
formula <- 'Sepal.Length ~ Sepal.Width + Petal.Length'

# Fit SEM model.
fit <- lavaan::sem(formula, sample.cov = cov_mat, sample.nobs = observations)

# Show results.
summary(fit)

And if you then run the generated code above, you will notice that it has been successfully reproduced.

Questions

• Do I need to specify all columns?
• How do I control the width of the code?
• Can I use semproducible in a tidyverse pipeline?
• How do I know that my generated code is correct?
• Why doesn't my generated code produce the exact same result?

Documentation

Load semproducible and then run ?semproducible in the R console to view the full documentation.

Support

Report problems or request a new feature by submitting a new issue.

Contribute

You can help with:

• Test semproducible in your project, and report a bug
• Read the documentation and verify that it corresponds to actual behavior

License

MIT