compRelSEM() with ord.scale = TRUE throws 'object isShared not found' error

[louisechristinerohde]

Bug Report

Package version: semTools 0.5.8
Version 2026.01.1+403 (2026.01.1+403)

Description

When calling compRelSEM() with ord.scale = TRUE on a fitted WLSMV CFA model
with ordinal indicators, the function throws the following error:

Error in ifelse(isShared, "(cluster means of) observed variables:\n\t", :
object 'isShared' not found

Reproducible example

The error occurs when:

• A unidimensional CFA model is fitted using WLSMV estimator with ordered indicators
• Items have an unequal number of observed response categories
  (e.g., one item has observed levels 2-6 while others have 1-6 on the same scale)
• compRelSEM() is called on the fitted model with ord.scale = TRUE

Expected behavior

compRelSEM() should return McDonald's omega for ordinal CFA models
without throwing an error, and ideally support percentile bootstrapping
with polychoric correlation matrices for ordinal data.

Additional context

The reliability() function works as a workaround but is deprecated.
Bootstrapped CIs via omega() in the psych package also fail with
polychoric correlations when items have unequal observed response categories.

[TDJorgensen]

This looks like a duplicate of #154
Can you install the development version to verify whether this is already fixed?

remotes::install_github("simsem/semTools/semTools")

[louisechristinerohde]

I installed the development version (0.5.8.902) but the error persists

[TDJorgensen]

Then please provide a reprex. I cannot reproduce the problem with this reprex, adapted from the ?compRelSEM help-page example:

data(HolzingerSwineford1939)
HS9 <- HolzingerSwineford1939[ , paste0("x",1:9)]
HS <- cbind(HS9,
            y7 = cut(HS9$x7, breaks = 4), # 4 categories
            y8 = cut(HS9$x8, breaks = 5), # 5 categories
            y9 = cut(HS9$x9, breaks = 6)) # 6 categories

HS.model <- ' speed   =~ y7 + y8 + y9 '

fit  <- cfa(HS.model, data = HS, ordered = TRUE, std.lv = TRUE)
compRelSEM(fit)

Output:

$speed

Composite `speed` is composed of observed variables:
	y7, y8, y9
True-score variance is represented by common factor(s):
	speed
Total variance of composite `speed` determined from the unrestricted model.
The proportion attributable to "true" scores is its model-based estimate of reliability ("omega"):

[1] 0.641

Also check your sessionInfo() output to verify the latest version is installed (0.5-8.902).

[georgemsavva]

For information, I came here with the same issue with the current CRAN version 0.5-8 but it was fixed by installing the development version 0.5-8.902.

Thanks