ESEM (feature request) by LSLindeloo · Pull Request #137 · jasp-stats/jaspSem

LSLindeloo · 2023-01-18T12:58:26Z

LSLindeloo · 2023-01-18T13:02:19Z

What do you think about this analysis, would you recommend fundamental changes, e.g. to the qml layout?

The analysis runs fine in R, however when i run the analysis in JASP, the "engine crashes", do you know what might be the problem?

Thanks in advance!

codecov-commenter · 2026-03-01T08:34:11Z

Codecov Report

❌ Patch coverage is 0% with 658 lines in your changes missing coverage. Please review.
✅ Project coverage is 55.95%. Comparing base (8e4cbc4) to head (2131aae).

Files with missing lines	Patch %	Lines
R/esem.R	0.00%	658 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #137      +/-   ##
==========================================
- Coverage   61.67%   55.95%   -5.72%     
==========================================
  Files          13       14       +1     
  Lines        6437     7095     +658     
==========================================
  Hits         3970     3970              
- Misses       2467     3125     +658

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

Copilot

Pull request overview

Adds a new ESEM analysis to the SEM module (new QML UI + R backend) to address feature request jasp-issues#1866.

Changes:

Register new analysis entry in inst/Description.qml.
Add new UI definition inst/qml/ESEM.qml with model/EFA-block inputs and estimation/output options.
Add new backend implementation R/esem.R and export ESEM in NAMESPACE.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 17 comments.

File	Description
inst/qml/ESEM.qml	New ESEM UI (models + EFA blocks, estimation/output, multigroup controls).
inst/Description.qml	Registers “Exploratory Structural Equation Modeling” analysis pointing to `ESEM.qml` / `ESEM`.
R/esem.R	New ESEM analysis backend (model building, lavaan fitting, tables/plots).
NAMESPACE	Exports `ESEM`.

Copilot · 2026-03-01T08:37:51Z

R/esem.R

+                              "dependentCorrelation", "threshold", "scalingParameter", "efaConstrained", "standardizedVariable", "naAction", "estimator", "modelTest",
+                              "errorCalculationMethod", "informationMatrix", "emulation", "group", "equalLoading", "equalIntercept",
+                              "equalResidual", "equalResidualCovariance", "equalMean", "equalThreshold", "equalRegression",
+                              "equalVariance", "equalLatentCovariance", "dataType", "sampleSize", "freeParameters", "rotation"))


Dependency list uses equalVariance, but the QML option is equalLatentVariance. This will prevent correct cache invalidation when the latent-variance equality constraint is toggled.

Suggested change

"equalVariance", "equalLatentCovariance", "dataType", "sampleSize", "freeParameters", "rotation"))

"equalLatentVariance", "equalLatentCovariance", "dataType", "sampleSize", "freeParameters", "rotation"))

Copilot · 2026-03-01T08:37:51Z

R/esem.R

+    options[["equalRegression"]],
+    options[["equalResidual"]],
+    options[["equalResidualCovariance"]],
+    options[["equalVariance"]],


equality_constraints reads options[["equalVariance"]], but the QML checkbox is equalLatentVariance. This means the latent-variance constraint won’t be applied (and can error if the option key is missing).

Suggested change

options[["equalVariance"]],

isTRUE(options[["equalVariance"]]) || isTRUE(options[["equalLatentVariance"]]),

Copilot · 2026-03-01T08:37:51Z

R/esem.R

+  if (options[["group"]] != "") {
+    lavopts[["group"]] <- .v(options[["group"]])
+  }
+
+  # sampling weights
+  if (options[["samplingWeights"]] != "") {
+    lavopts[["sampling.weights"]] <- .v(options[["samplingWeights"]])
+  }


.v() is called here but is never defined in this file/package scope. Use the raw option strings (as SEM does) or define .v <- I explicitly before use.

Copilot · 2026-03-01T08:37:52Z

R/esem.R

+  # edit options[["models"]]
+  options[["models"]] <- lapply(options[["models"]], .esemSyntax)
+  return(options)


options[["models"]] <- lapply(..., .esemSyntax) can inject NULL entries because .esemSyntax() uses bare return() in several branches. That will make later code (which indexes options[["models"]][[i]]) fail. Consider filtering/removing empty models explicitly (like .semPrepOpts() does) instead of returning NULL from the mapper.

Copilot · 2026-03-01T08:37:52Z

inst/Description.qml

+	Analysis 
+	{
+		title:	qsTr("Exploratory Structural Equation Modeling")
+		qml:	"ESEM.qml"
+		func:	"ESEM"


New analysis added, but there are no testthat tests covering it (this repo has comprehensive tests for SEM-family analyses, e.g. tests/testthat/test-sem.R). Please add at least a smoke test that runs ESEM and checks key outputs are produced.

Copilot · 2026-03-01T08:37:53Z

R/esem.R

+      modelContainer$setError(paste0("Error in model \"", options[["models"]][[i]][["name"]], "\" - ",
+                                     .decodeVarsInMessage(names(dataset), errmsg)))


This error text is user-visible but isn’t wrapped in gettext()/gettextf(), so it won’t be translated. Please wrap the full message (including the model name) for i18n consistency with the rest of the module.

Suggested change

modelContainer$setError(paste0("Error in model \"", options[["models"]][[i]][["name"]], "\" - ",

.decodeVarsInMessage(names(dataset), errmsg)))

fullErrMsg <- gettextf(

"Error in model \"%1$s\" - %2$s",

options[["models"]][[i]][["name"]],

.decodeVarsInMessage(names(dataset), errmsg)

)

modelContainer$setError(fullErrMsg)

Copilot · 2026-03-01T08:37:53Z

R/esem.R

+.esemComputeResults <- function(modelContainer, dataset, options) {
+  #' create result list from options
+  # find reusable results
+  if (!options[["estimator"]] %in% c("default", "ml", "mlr", "mlf") && options[["naAction"]] == "fiml") return()


Returning NULL here can break callers like .esemFitTab() which assume a list of fits. Prefer setting a user-visible error on modelContainer and returning, or make callers handle NULL safely.

Suggested change

if (!options[["estimator"]] %in% c("default", "ml", "mlr", "mlf") && options[["naAction"]] == "fiml") return()

if (!options[["estimator"]] %in% c("default", "ml", "mlr", "mlf") && options[["naAction"]] == "fiml") {

modelContainer$setError(gettext("Full information maximum likelihood (FIML) is only available for ML-based estimators."))

return()

}

Copilot · 2026-03-01T08:37:53Z

R/esem.R

+                       overtitle = gettextf("%s%% Confidence Interval", options$ciLevel * 100))
+
+
+  efatab$addFootnote(message = gettextf("Applied rotation method is <i>%s</i>. Different EFA blocks are rotated seperately.", options[["rotation"]]))


Spelling: "seperately" → "separately".

Suggested change

efatab$addFootnote(message = gettextf("Applied rotation method is <i>%s</i>. Different EFA blocks are rotated seperately.", options[["rotation"]]))

efatab$addFootnote(message = gettextf("Applied rotation method is <i>%s</i>. Different EFA blocks are rotated separately.", options[["rotation"]]))

Copilot · 2026-03-01T08:37:54Z

R/esem.R

+.esemEditOptions <- function(dataset, options) {
+  indicators <- unlist(unique(lapply(options[["models"]], function(x) {
+    parsed <- lavaan::lavParseModelString(x[["syntax"]][["model"]], TRUE)
+    return(unique(parsed[parsed$op == "=~",]$rhs))
+  })))


Auto-detecting ordinal indicators here only parses the CFA/regression syntax and ignores indicators that are only included via EFA blocks. That can lead to choosing an ML-type estimator even when EFA indicators are ordinal. Consider adding EFA block variables to indicators before checking is.ordered.

Copilot · 2026-03-01T08:37:54Z

R/esem.R

+  cfaAndRegressionSyntax <- model[["syntax"]][["model"]]
+  if(cfaAndRegressionSyntax == "") return()
+
+  efaBlocks <- model[["efaBlock"]]
+  efaSyntax <- ""
+  for (i in 1:length(efaBlocks)) {
+    if (length(efaBlocks[[i]][["variables"]]) < 1) return()


.esemSyntax() returns NULL if the CFA/regression syntax is empty, which prevents EFA-only models and also creates NULL entries in options[["models"]]. Consider allowing empty CFA/regression syntax (generate model from EFA blocks only) and only skipping specific empty EFA blocks rather than aborting the whole model.

draft version ESEM

6a7c0f9

juliuspfadt self-requested a review July 1, 2025 09:42

Merge branch 'master' into ESEM

2131aae

juliuspfadt requested a review from Copilot March 1, 2026 08:28

Copilot started reviewing on behalf of juliuspfadt March 1, 2026 08:28 View session

Copilot AI reviewed Mar 1, 2026

View reviewed changes

	"equalVariance", "equalLatentCovariance", "dataType", "sampleSize", "freeParameters", "rotation"))
	"equalLatentVariance", "equalLatentCovariance", "dataType", "sampleSize", "freeParameters", "rotation"))

	options[["equalVariance"]],
	isTRUE(options[["equalVariance"]]) \|\| isTRUE(options[["equalLatentVariance"]]),

		modelContainer$setError(paste0("Error in model \"", options[["models"]][[i]][["name"]], "\" - ",
		.decodeVarsInMessage(names(dataset), errmsg)))

		overtitle = gettextf("%s%% Confidence Interval", options$ciLevel * 100))


		efatab$addFootnote(message = gettextf("Applied rotation method is <i>%s</i>. Different EFA blocks are rotated seperately.", options[["rotation"]]))

Conversation

LSLindeloo commented Jan 18, 2023

Uh oh!

LSLindeloo commented Jan 18, 2023

Uh oh!

codecov-commenter commented Mar 1, 2026

Codecov Report

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Mar 1, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 1, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 1, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 1, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 1, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 1, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 1, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 1, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 1, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 1, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants