# 37. Regression bootstrapping

## Purpose

How stable are the OLS and logistic coefficient estimates under case resampling?

## Practice study

- File: [business-outcomes-observed.csv](../samples/business-outcomes-observed.csv)
- Observations: 500
- Data type: deterministic synthetic instructional data
- Complexity: a realistic multivariable observed-data analysis

**Prerequisites**

- Run the corresponding point model first

## Exact procedure

1. Start QuickPLS and choose **New Project**. Enter a readable project name such as **Regression bootstrapping Tutorial** and choose a local `.qpls` destination.

2. Open **Data**, choose **Import Data**, select **business-outcomes-observed.csv**, keep the detected header row, review the preview, and choose **Import**. Confirm 500 observations are available.

   ![Regression bootstrapping imported data](../screenshots/regression-bootstrapping/01-data-ready.png)

3. This method does not require a latent-variable diagram. In its setup screen use these roles: First use the OLS roles above; optionally repeat with customer_churn as the binary outcome and the logistic predictor set.

4. Review the selected outcome, predictors, groups, or analysis variables. Remove case_id and every unused teaching/reference column from the analysis selection.

   ![Regression bootstrapping prepared model or variable roles](../screenshots/regression-bootstrapping/02-variable-roles.png)

5. Check that every required role is populated with a numeric variable and that outcome/predictor selections do not overlap.

6. Use **Calculate → Regression, choose OLS or Binary logistic, then enable Case-resampling bootstrap**.

   ![Regression bootstrapping calculation setup](../screenshots/regression-bootstrapping/03-calculation-setup.png)

7. Complete the method-specific settings:

- Use 1,000 resamples, a fixed seed, and 95% intervals

8. Read **Readiness**, leave **Open Results when finished** selected, and choose **Start calculation** once. Wait for **Completed**; do not close the project while the result is being saved.

9. In **Results**, confirm **Regression bootstrapping** is selected and inspect these outputs in order:

1. Bootstrap coefficient estimates
2. Standard errors and confidence intervals
3. Usable and failed replicate accounting

   ![Regression bootstrapping primary result](../screenshots/regression-bootstrapping/04-results-primary.png)

   ![Regression bootstrapping secondary result](../screenshots/regression-bootstrapping/05-results-secondary.png)

10. Choose **Save Report** to preserve the exact result. Use **Export** for the needed table/report format. Close and reopen the project once and confirm the saved result remains selectable.

## Reading guidance

- The data are deterministic synthetic teaching data. They are suitable for reproducing the workflow, not for substantive publication claims.