The task: analyse customer satisfaction and loyalty
The customer-experience study asks whether customers who report better service quality and greater trust also report higher satisfaction, and whether satisfaction is associated with loyalty. It uses survey-style questions for Quality, Trust, Satisfaction and Loyalty.
Use this page to organise your first project, then follow the linked PLS-SEM lesson for the model, calculation settings and checked answers. PLS-SEM estimates relationships between concepts measured by several indicators, such as the questions assigned to Trust. The supplied data contain 240 synthetic cases; they were created for teaching and contain no participant responses.
Complete the exercise by reopening the project, locating its saved calculation and matching the point estimates to the checked answers. If your research question instead concerns combinations of conditions or minimum requirements, consult the method chooser.
1. Create the project and import the data
A project keeps the imported data, model and saved calculations together. In customer-experience.csv, each row represents a customer and each measurement column represents a survey question.
Keep case_id as the row identifier so that you can locate a case later. It is not an indicator or predictor: its values label records rather than measure a research concept.
Create a writable project
Choose File → New Project… (Ctrl+N). Enter First customer study in Project name. Use Browse… beside Project file to choose where to save it, then click Create project. Continue in the project you have just created.
Choose the customer data
Download customer-experience.csv below. Choose File → Import Data… (Ctrl+I), then Choose data file… and select that CSV.
Review the import settings
In Review & import, check that Input represents raw data, Header row is 1, the separator is comma and Decimal convention is Dot (1.25). Treat the twelve measurement columns as Continuous. Check the column names and preview before importing.
Import and check the table
Click Import dataset and confirm Dataset imported and verified. Inspect the imported column names and confirm 240 cases. If a whole row appears in one column, check the separator; if numeric responses appear as text, check their format and decimal convention. Investigate unexpected blanks against the original file before building the model.
2. Build the specified model and calculate it
The model-building walkthrough linked below explains how to create and name each construct, assign indicators and draw the arrows. Complete those steps before opening the calculation settings. The arrows specify relationships to estimate; survey-style data alone do not establish causal effects.
These instructions apply to the customer PLS model. The separate fsQCA lesson instead defines conditions, an outcome and set memberships in a study specification.
Build the four measurement blocks
Follow the linked model-building walkthrough. Create Quality, Trust, Satisfaction and Loyalty, assign each construct its three matching indicators and set Measurement model to Reflective.
Check the structural paths
Draw Quality → Satisfaction, Trust → Satisfaction and Satisfaction → Loyalty. Compare the indicator assignments and arrow directions with the model plan in the PLS lesson.
Select the calculation and settings
Click Calculate… on the model toolbar. In Find a method, select PLS-SEM Algorithm; open Methods if the list is collapsed. Use Starting weights → Standard (+1) and Missing data → Treatment → Casewise deletion. This complete practice file should retain all 240 cases.
Start the calculation
Check Readiness and the selected dataset and model. Keep Open Results when finished selected and click Start calculation. If prompted, save the project or a writable copy and continue. Resolve any setup message before interpreting the output.
3. Locate and check the result
Convergence means that the repeated estimation steps settled within the stopping criterion. Investigate an unfinished or nonconverged calculation before interpreting its estimates.
After checking the calculation, complete the PLS lesson's measurement assessment before interpreting the structural relationships. A coefficient describes the direction and size of a relationship in this model. Its uncertainty requires a separate bootstrap calculation; the checked point estimates do not include confidence intervals or p-values.
Select the completed result
In Results, use the Result dropdown to select the completed calculation. Confirm the intended data and model, the 240-case count and convergence.
Find the model and tables
Open Overview → Graphical Output. Then use Estimates → Outer loadings, Effects → Direct effects and Estimates → R-squared to find the named indicators, paths and outcomes. Search result outputs can help locate a table.
Compare the checked answers
Open the customer-model point estimates linked below. Compare the named paths and outcomes at the stated precision, using table values rather than numbers read from an image. If values differ, check the case count, indicator assignments, reflective measurement, arrows and calculation settings before changing observations.
4. Save a report you can trace to the analysis
Save the project with File → Save (Ctrl+S). With the completed calculation selected, choose File → Save Report…. Enter a Saved report name, check Result source, and click Save report. Follow the report guide below to export a table and reopen the saved work. Confirm that the report and exported table refer to the same analysis.
Keep the data version, analysed case count, indicator assignments, paths and settings with the output. If you revise the model or data, save the revision separately and calculate it again. Preserve the earlier result so that you can explain the difference.
Write a short account of the question, model and estimated relationships. Identify the data as synthetic. Distinguish the checked point estimates from measurement assessment or bootstrap inference that you have actually completed.
