Original practice answer: brand-response-group-comparisons

Read https://quickpls.com/research/papers/brand-response-group-comparisons

Synthetic teaching CSV, CC0. Not the journal dataset. These files are readable instructions and results, not an importable QuickPLS project.

Repeat the reference with R and lavaan 0.7.3, jsonlite and digest:
Rscript reference-calculations.R . NEW-OUTPUT-FOLDER
Use a new directory. The script produces a new reference calculation, not a desktop-app run. See checks.json for exact verification scope. All estimates and failures remain in the reference record.


ADDITIONAL WORKED ANALYSIS
Group contrast: requires R, lavaan 0.7.3, jsonlite and digest. Run: Rscript group-contrast-reference.R . NEW-CONTRAST-OUTPUT
This separately repeats the full-metric stratified bootstrap and joint-model raw path contrast. It writes the exact sampled row indices and every usable contrast.
