[Bug]: SEM mediation bug

[Joshnneal]

JASP Version

0.95.1

Commit ID

No response

JASP Module

SEM

What analysis are you seeing the problem on?

Mediation

What OS are you seeing the problem on?

macOS Silicon

Bug Description

Attempted to bootstrap a mediation analysis for my dataset (attached) utilizing the following options:
jaspSem::MediationAnalysis(
data = NULL,
version = "0.95",
bootstrapCiType = "percentileBiasCorrected",
bootstrapSamples = 1000,
ciLevel = 0.95,
confounds = ~ ANT_Correct_Total + Age + Sex + Medication,
emulation = "lavaan",
errorCalculationMethod = "bootstrap",
estimator = "default",
mediators = ~ R/S_LC_right,
naAction = "fiml",
outcomes = ~ ANT_RT_Alerting_Effect,
pathCoefficient = TRUE,
pathPlot = FALSE,
pathPlotLegend = FALSE,
pathPlotParameter = FALSE,
plotHeight = 320,
plotWidth = 480,
predictors = ~ IRR,
rSquared = FALSE,
residualCovariance = TRUE,
standardizedEstimate = TRUE,
standardizedEstimateType = "all",
syntax = TRUE,
totalIndirectEffect = TRUE)

Then received the following error message:
Error in if (var(t) = 0) {: missing value where TRUE/FALSE needed

Stack trace
tryCatchOne(expr, names, parentenv, handlers[[1]])

doTryCatch(return(expr), name, parentenv, handler)

withCallingHandlers(expr = analysis(jaspResults = jaspResults, dataset = dataset, options = options), error = .addStackTrace)

analysis(jaspResults = jaspResults, dataset = dataset, options = options)

.medParTable(modelContainer, dataset, options, ready)

lavaan::parameterestimates(fit, level = options[['ciLevel']], boot.ci.type = bootstrapCiType)

I am concerned with what is going on, as well as if this reflects issues with the non-bootstrapped estimates of the indirect effects.

BS_3.6_OALC.csv

Expected Behaviour

Ideally, bootstrapping for the model intended.

Steps to Reproduce

1. Enter the mediation analysis
2. Set the variables as noted above
3. Select Standardized Estimates
4. Select the Bootstrapping Option

Log (if any)

No response

More Debug Information

-------- Application Info --------
JASP Version: JASP 0.95.1 (Apple Silicon)
Build Branch: development
Build Date: Aug 20 2025 12:37:17 (Netherlands)
Last Commit: 180d14c18d13c107ee7c16b0993ff7e1cc02680b

-------- Basic Info --------
Operating System: macOS Sequoia (15.6)
Product Version: 15.6
Kernel Type: darwin
Kernel Version: 24.6.0
Architecture: arm64
Install Path: /Applications/JASP.app/Contents/MacOS
Platfotm Name: cocoa
System Local: en_US

Final Checklist

• I have included a screenshot showcasing the issue, if possible.
• I have included a JASP file (zipped) or data file that causes the crash/bug, if applicable.
• I have accurately described the bug, and steps to reproduce it.

[tomtomme]

@Joshnneal
thx for the report.
I cannot reproduce your problem exactly. I tried your R-Syntax above, but with that, there would be no mediator in the model. So I had to put it manually in but then got a different error message than yours. Then I suspected a problem with the dash "/" in the variable name and removed it to no avail.
Some tinkering later I realized that the model runs fine with bootstrapping, but not with "standardized estimates".
Also when I go for standardized estimates without bootstrap, I get:

@juliuspfadt
See these errors:

BS_3.6_OALC.jasp.zip

any idea?

[Joshnneal]

Thanks for the testing. Both of the variables identified in the error message above are coded binary (0 or 1), yet have to be manually switched to ordinal from nominal. Doing that, when I turn off standardized estimates JASP fails to calculate variance or significance levels:

And still fails when attempting to bootstrap:

For comparison, here is what is being modeled with standardized estimates, without bootstrapping:

[juliuspfadt]

@Joshnneal thanks for creating the issue. there were multiple things not going well in the R code, especially in terms of standardization and error handling. I fixed them, should be in the next release, and earlier in the nightly versions. Keep track of the linked PR for checking the progress.