# 2. PLS power analysis

## Purpose

What power is expected for a Quality → Loyalty path of 0.30 at planned sample sizes of 80 and 120?

## Practice study

- File: [customer-experience-pls.csv](../samples/customer-experience-pls.csv)
- Prepared project: [customer-experience-pls-power.qpls](../projects/customer-experience-pls-power.qpls)
- Observations: 400
- Data type: deterministic synthetic instructional data
- Complexity: a focused two-construct prospective-power design with four reflective indicators per construct

> **Installed 2.62.8 qualification note:** the method accepts this model and completes the bounded Monte Carlo/bootstrap workload, but QuickPLS then rejects its own completed result during canonical publication with “A completed result-backed run or validated dataset-only power result is required.” The prepared project is therefore an unqualified teaching template. Result screenshots and close/reopen evidence will remain pending until a future application version resolves that publication defect.

## Exact procedure

1. Start QuickPLS and choose **Open Project**. Select [customer-experience-pls-power.qpls](../projects/customer-experience-pls-power.qpls). This website-derived teaching project contains the exact dataset, a focused analysis-ready model, bindings, and a saved layout.

2. Open **Data** and confirm **customer-experience-pls.csv** contains 400 observations. Return to **Model**. The linked CSV remains available separately for inspection and re-import practice.

   ![PLS power analysis sample data](../screenshots/pls-power-analysis/01-data-ready.png)

3. Confirm that the prepared model contains Quality (quality_1–quality_4) and Loyalty (loyalty_1–loyalty_4), both reflective, with one saved Quality → Loyalty structural path.

4. The prepared project already contains a saved, neatly arranged model layout. Do not choose **Arrange** or otherwise change the canvas before this tutorial calculation. If you later edit the model, arrange it once and save the edited project before calculating.

   ![PLS power analysis prepared model or variable roles](../screenshots/pls-power-analysis/02-arranged-model.png)

5. Choose **Validate** and confirm there are no blockers. Do not save the unchanged prepared project; save only after you intentionally edit it.

6. Use **Calculate → PLS-SEM Sample Size and Power**.

   ![PLS power analysis calculation setup](../screenshots/pls-power-analysis/03-calculation-setup.png)

7. Complete the method-specific settings:

- Enter `quality_to_loyalty_power` as the scenario identity.
- Select **Quality** as predictor and **Loyalty** as outcome.
- Use population path `0.30`, sample-size grid `80,120`, two-sided alpha `0.05`, target power `0.80`, 95% Wilson confidence, and a deterministic seed.
- For a bounded instructional check, 100 Monte Carlo replicates and 99 case-bootstrap replicates are sufficient to exercise the workflow. Use a larger justified design for substantive planning.

8. Read **Readiness**, leave **Open Results when finished** selected, and choose **Start calculation** once. In QuickPLS 2.62.8 the calculation completes, but result publication enters **Project reconciliation required** because canonical validation rejects the completed result. Do not treat this as a scientific result.

9. After the application publication defect is corrected in a future qualified release, the expected Results workflow is to inspect these outputs in order:

1. Minimum sample-size decision
2. Power curve
3. Simulation accounting and convergence receipt

   ![PLS power analysis primary result](../screenshots/pls-power-analysis/04-results-primary.png)

   ![PLS power analysis secondary result](../screenshots/pls-power-analysis/05-results-secondary.png)

10. Report, export, and close/reopen verification are not claimed for 2.62.8 because no validated result is published.

## Reading guidance

- The data are deterministic synthetic teaching data. They are suitable for reproducing the workflow, not for substantive publication claims.
- The 80/120 grid is intentionally small and the replicate counts are intentionally bounded for workflow qualification; neither is a universal sample-size recommendation.