Try to fix ordered = TRUE

[sfcheung]

Dear Prof. Epskamp,

May I propose some changes to semPlot?

A post in the lavaan Google Group (https://groups.google.com/g/lavaan/c/xnOqMUbUYfY/m/aZUQo1MYAgAJ) led me to issue #40 .

The error occurred with ordered = TRUE. As one of the users pointed out, it is related to lavaan.R and lavaanNames().

I am interested in this topic because I may also need to plot this kind of model. I took a look and it seems that you have already been working on this issue, which is related to the conditional.x option. The current error occurred due to two other matrices, ImpCovs and ObsCovs, which do not match the names retrieved "ov" and need to use "ov.model", when conditional.x is TRUE.

This is an illustration of the error:

library(lavaan)
#> This is lavaan 0.6-19
#> lavaan is FREE software! Please report any bugs.
# library(semPlot)

mod_pop <-
"
m ~ .3 * x
y ~ .4 * m
"
dat <- simulateData(model = mod_pop,
                    sample.nobs = 1000,
                    seed = 2345)
dat$y <- ifelse(dat$y > mean(dat$y), 1, 0)
dat$m <- ifelse(dat$m > mean(dat$m), 1, 0)
head(dat)
#>   m y          x
#> 1 0 0 -1.2856876
#> 2 1 1 -0.5067523
#> 3 1 0  1.1251033
#> 4 1 1  0.3062505
#> 5 1 0 -1.6463903
#> 6 0 1 -0.9033056

mod <-
"
m ~ x
y ~ m
"
fit <- sem(mod,
           data = dat,
           ordered = c("y", "m"))

devtools::load_all("C:/GitHub/semPlot")
#> ℹ Loading semPlot
p0 <- semPlotModel(fit)
#> Error in dimnames(x) <- dn: length of 'dimnames' [2] not equal to array extent

^{Created on 2024-11-15 with reprex v2.1.1}

I tried to fix this problem based on what you have done:

sfcheung@1ee4f3c

It seems to work:

library(lavaan)
#> This is lavaan 0.6-19
#> lavaan is FREE software! Please report any bugs.
# library(semPlot)

mod_pop <-
"
m ~ .3 * x
y ~ .4 * m
"
dat <- simulateData(model = mod_pop,
                    sample.nobs = 1000,
                    seed = 2345)
dat$y <- ifelse(dat$y > mean(dat$y), 1, 0)
dat$m <- ifelse(dat$m > mean(dat$m), 1, 0)
head(dat)
#>   m y          x
#> 1 0 0 -1.2856876
#> 2 1 1 -0.5067523
#> 3 1 0  1.1251033
#> 4 1 1  0.3062505
#> 5 1 0 -1.6463903
#> 6 0 1 -0.9033056

mod <-
"
m ~ x
y ~ m
"
fit <- sem(mod,
           data = dat,
           ordered = c("y", "m"))

devtools::load_all("C:/GitHub/semPlot")
#> ℹ Loading semPlot
p0 <- semPlotModel(fit)
semPaths(p0)

^{Created on 2024-11-15 with reprex v2.1.1}

I am not sure whether these changes will break other things. If it will, I can try to see how to fix them.

Feel free to reject this PR if it does not fit your plan.

Thanks again for developing this package for SEM users.

-- Shu Fai

[juliuspfadt]

we have also come across this error in JASP (https://github.com/jasp-stats/INTERNAL-jasp/issues/2724) Would be nice if this was merged soon @SachaEpskamp.