HTMT confidence intervals

[Carsten S]

Hello,

up to now, I have used semTools::htmt() to get heterotrait-monotrait results.

Can anybody point me to a way to receive htmt values with confidence intervals?

Thank you!

[Christian Arnold]

https://groups.google.com/g/lavaan/c/0RSsh4M6zQg/m/03uaQYpyAQAJ

[Carsten S]

Hi Christian,

is there an introduction to the function?

Or do you have an example for using it towards the CI of HTMT?

Thanks 

Carsten

[Christian Arnold]

Hi Carsten,

The first post explains how to bootstrap HTMT with x.boot. Since you don't need to fit a lavaan object to compute HTMT, you can also do something like this (quick and dirty - check code):

library(semTools)

HS.model <- 'visual  =~ x1 + x2 + x3
             textual =~ x4 + x5 + x6
             speed   =~ x7 + x8 + x9'

data <- HolzingerSwineford1939

# Bootstrap draws
R <- 100

# Run the bootstrap
boot <- sapply(1 : R, function(x, data) {
  d <- data[sample(1 : nrow(data), replace = TRUE, nrow(data)),]
  t <- htmt(HS.model, d)
  t[lower.tri(t)]
}, data)

# Get CI
sapply(1 : nrow(boot), function(x, t0, boot) {
  c(
    coef = t0[lower.tri(t0)][x],
    quantile(boot[x,], c(0.025, 0.975))
  )
}, htmt(HS.model, data), boot) 


HTH

Christian

[Carsten S]

Thank you very much. Works like a charm.

However, I also have one single item measurement. This is not covered
by the htmt() function.
Any suggestions on how to best work around this?

Thank you!

> --
> You received this message because you are subscribed to a topic in the Google Groups "lavaan" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/topic/lavaan/Uizg37S3dYc/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to lavaan+un...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/lavaan/fba67f23-0349-41d3-bc4f-a5d5b39d547dn%40googlegroups.com.

[Christian Arnold]

Hi Carsten,

Well, the documentation says quite clearly: "lavaan model.syntax of a confirmatory factor analysis model where at least two factors are required for indicators measuring the same construct" (https://rdrr.io/cran/semTools/man/htmt.html). Why don't you switch to the discriminantValidity function (https://rdrr.io/cran/semTools/man/discriminantValidity.html)?

There is literature that very clearly emphasizes the advantage of this method for SEM compared to HTMT: https://journals.sagepub.com/doi/10.1177/1094428120968614

HTH

Christian