Skip to content

Refactor code for compositeRVCOV() #134

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Kss2k merged 4 commits into from
Mar 17, 2026
Merged

Refactor code for compositeRVCOV() #134

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
546c2c6
Refactor code for `compositeRVCOV()`
Kss2k Mar 17, 2026
9f7125e
Fix codex comments
Kss2k Mar 17, 2026
19c0bf1
Don't use sign(I)
Kss2k Mar 17, 2026
fbd7de1
Check non-zero values, not length
Kss2k Mar 17, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
61 changes: 24 additions & 37 deletions R/reliablity_single_item.R
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -598,47 +598,34 @@ calcAVE <- function(lambda.std, lV) {
}


getCompositeDenominator <- function(lV, parTable, cfa, scale.corrected) {
measr <- parTable[parTable$lhs == lV & parTable$op == "=~", ]
inds <- unique(measr$rhs)
mats <- lavaan::lavInspect(cfa, what = "coef")
if (scale.corrected) {
sum(as.vector(mats$lambda[inds, lV]))
} else {
length(inds) # rowMeans: divide sum by k
}
}
getCompositeRVCOV <- function(lVs, parTable, cfa, scale.corrected) {
matrices <- lavaan::lavInspect(cfa, what = "coef")
lambda <- matrices$lambda
theta <- matrices$theta

inds <- rownames(lambda)[apply(lambda, MARGIN = 1L, FUN = \(x) any(x != 0))]
lambda <- lambda[inds, lVs, drop = FALSE]
theta <- theta[inds, inds, drop = FALSE]

getCompositeRVCOV <- function(lVs, parTable, cfa, scale.corrected) {
# Build Theta blocks and denominators
mats <- lavaan::lavInspect(cfa, what = "coef")
thetaF <- mats$theta
I <- lambda
I[I!=0] <- 1

# item lists per LV
item_list <- lapply(lVs, function(lv)
unique(parTable[parTable$lhs == lv & parTable$op == "=~", "rhs"])
)
names(item_list) <- lVs
T <- t(I) %*% theta %*% I

# denominators D_f
denom <- vapply(lVs, function(lv)
getCompositeDenominator(lv, parTable, cfa, scale.corrected),
numeric(1)
)
FUN <- if (scale.corrected) sum else \(x) sum(x != 0)

p <- length(lVs)
resCov <- matrix(0, p, p, dimnames = list(lVs, lVs))

for (i in seq_len(p)) {
Ii <- item_list[[i]]
for (j in i:p) {
Ij <- item_list[[j]]
block_sum <- sum(thetaF[Ii, Ij, drop = FALSE])
val <- block_sum / (denom[i] * denom[j])
resCov[i, j] <- val
resCov[j, i] <- val
}
l <- apply(lambda, MARGIN = 2L, FUN = FUN)
l.inv <- 1 / l
l.inv[l <= 0] <- 1L
Copy link
Copy Markdown

@chatgpt-codex-connector chatgpt-codex-connector bot Mar 17, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Preserve denominator sign when inverting loadings

Overwriting l.inv for all l <= 0 silently changes the scaling for any factor whose corrected denominator is negative (or exactly zero), so L.inv %*% T %*% L.inv no longer matches the intended per-factor normalization and can emit incorrect ~~ terms in generated syntax for models with negative/canceling loadings. The previous implementation divided by each factor’s actual denominator, so this guard should not collapse all non-positive values to 1.

Useful? React with 👍 / 👎.


if (length(l.inv) > 1) {
L.inv <- diag(l.inv)
dimnames(L.inv) <- list(names(l.inv), names(l.inv))

} else {
L.inv <- matrix(l.inv, nrow = 1, ncol = 1)
dimnames(L.inv) <- list(names(l.inv), names(l.inv))
}
resCov

L.inv %*% T %*% L.inv
}
Loading