[jaspFrequencies] Fix crash in Bayesian Informed Multinomial Test when enabling...

[sisyphus-jasp]

Summary

Fixes: https://github.com/jasp-stats/INTERNAL-jasp/issues/3106

Root cause

• .createInformedMultSequentialAnalysisPlot() assumed sequential results always existed.
• For aggregated multinomial input (factor + counts), .computeInformedMultSequentialResults() yields no sequential rows/state; posterior-probability code then built a mismatched data.frame, causing fatal crash.

What changed

• Added guards in R/informedmultinomialtestbayesian.R inside .createInformedMultSequentialAnalysisPlot():
• If sequential state is missing/NULL/non-data.frame/empty, create the sequential plot element and set a clear error: sequential analysis needs individual-level data.
• After include/exclude model filtering, if no rows remain, create plot element and set a clear error.
• Kept normal path unchanged when valid sequential results exist.

Why

• Prevent fatal termination and convert invalid sequential-analysis states into user-facing plot errors.

Caveats / reviewer checks

• testAll() result after fix: [ FAIL 0 | WARN 9 | SKIP 3 | PASS 110 ] (matches baseline counts).
• Reproduced attached .jasp now returns status: complete; sequential plot carries error message instead of crashing.

Implementation Plan

Root cause

• In R/informedmultinomialtestbayesian.R, .createInformedMultSequentialAnalysisPlot() assumes .computeInformedMultSequentialResults() always creates jaspResults[["sequentialAnalysisResults"]] with rows.
• For aggregated multinomial input (factor + count, no individual trials), .multinomAggregateData() sets attr(dataset, "individual") <- FALSE; then .computeInformedMultSequentialResults() returns early and no sequential state is created.
• The plot builder still executes posterior-probability code and constructs data.frame(model = unique(sequentialAnalysisResults$model), ..., postProb = priorProb), causing row-length mismatch and fatal crash.

Proposed changes

• File: R/informedmultinomialtestbayesian.R
• Function: .createInformedMultSequentialAnalysisPlot()
• Add an early guard after compute step:
  • if sequential results state is missing/NULL/empty, create sequentialAnalysisPlot with proper dependencies and set a user-facing error (sequential analysis requires unaggregated individual data), then return.
• Add a second guard after model filtering (includeNullModel, includeEncompassingModel): if no models remain, create same plot with error and return.
• Keep behavior unchanged when valid sequential rows exist.

Expected test impact

• Existing tests should stay green: change only converts current crash path into explicit plot error handling.
• No snapshot auto-accept; if any snapshot differs due new error message in this path, it would be expected stale output.

Fix issue #3106: Fix crash in Bayesian Informed Multinomial Test when enabling Sequential analysis plot with Posterior probability display option.

Attachment files available in: /workspace/attachment/
Files: bayes_inform_multinom.jasp

Screenshot Analysis

screenshot-1.png:

The screenshot shows JASP in the Frequencies module, analysis: Informed Bayesian Multinomial Test (results pane title: “Informed Bayesian Multinomial Test”).

Exact visible error text (as shown):

This analysis terminated unexpectedly.

Error in data.frame(model = unique(sequentialAnalysisResults$model),
step = 0, : arguments imply differing number of rows: 0, 1

Stack trace

To receive assistance with this problem, please report the message
above at: https://jasp-stats.org/bug-reports

Note: the call stack itself is not expanded/visible in this screenshot; only the “Stack trace” row is shown.

Selected UI state / settings visible:

• Analysis panel: Informed Bayesian Multinomial Test
• Top toolbar module selected: Frequencies
• Variables list (left): Trials, Count2, Trials2
• Assigned fields:
  • Factor: Race
  • Counts: Count
• Show all options: unchecked
• Bayes Factor Comparison:
  • Null: selected
  • Encompassing: unselected
  • vs. Model 1: unselected
• Bayes Factor output type:
  • BF₁₀: selected
  • BF₀₁: unselected
  • Log(BF₁₀): unselected
• Descriptives:
  • Table: checked
  • Plot: unchecked
  • Display Counts: unselected
  • Display Proportions: selected
• Posterior plot: unchecked (Credible interval shown as 95%, disabled)
• Sequential analysis plot: checked
  • Display Bayes factors: unselected
  • Display Posterior probability: selected
• Collapsed sections: Prior Distribution, Prior Model Probability
• Order Restricted Hypotheses: expanded
  • Tabs visible: Model 1, Model 2, Model 3
  • Entered hypothesis text (Model 1): White > Black > Chinese > Indian
• Advanced: collapsed

Output visible:

• No tables/plots produced.
• Only the red error warning card in the Results pane.

Test Results

Test Run	Result
Baseline (pre-fix)	[ FAIL 0 | WARN 9 | SKIP 3 | PASS 110 ]
Post-fix	[ FAIL 0 | WARN 9 | SKIP 3 | PASS 110 ]
Upstream CI	e24f81f -- CI: passing

Automated Code Review

Approved after 1 review iteration(s).