lsem.permutationTest: add sufficient_statistics to arglist

[martinabader]

Hello Alexander,

I ran into a problem with the lsem.permutationTest function.

Here is a reproducible example:

library(MASS)
library(sirt)

sigma <- matrix(.5, nrow = 4, ncol = 4)
diag(sigma) <- 1

colnames(sigma) <- rownames(sigma) <- c('V1', 'V2', 'V3', 'V4')

mu = c(1, 2, 3, 4)

df <- as.data.frame(mvrnorm(n = 1000, Sigma = sigma, mu = mu))
df$moderator <- sample(c(1,2,3), size = 1000, replace = T)

mod <- 'g =~ V1 + V2 + V3 + V4'

lsem.object <- lsem.estimate(df, 
                             moderator = 'moderator', 
                             moderator.grid = 1:3, 
                             lavmodel = mod, 
                             sufficient_statistics = FALSE, 
                             estimator = 'MLM')

lsem.permutation <- lsem.permutationTest(lsem.object, B = 3)

This gives the error Error in res0$parameters : $ operator is invalid for atomic vectors

When looking into the source code I found that the problem arises because the lsem.permutationTest function does not consider the sufficient_statistics argument of the lsem.object when estimating LSEMs with the permuted data sets. However, because I want to use the MLM estimator, lavaan threw an error when using the default of sufficient_statistics = TRUE.

In the pull request, I added sufficient_statistics to arglist which fixed the error.

[alexanderrobitzsch]

Thank you! I will include the change in the next update.

[alexanderrobitzsch]

included.