QuickPLS Paths and explained variation
Path coefficients: all 13 relationships
A path coefficient describes the fitted relationship with an outcome after accounting for its other predictors. These coefficients are standardised, so their scale is comparable within this model.
For Customer loyalty, the coefficient from Customer satisfaction is 0.505 in both columns. The coefficient from Likeability is 0.344. These are estimates of different relationships in the same outcome equation. Their size alone does not establish a causal effect or statistical significance; a bootstrap analysis would address uncertainty.
On a small screen, swipe the table sideways to see every column.
| Result | QuickPLS | SmartPLS |
|---|---|---|
| Attractiveness → Competence | 0.086 | 0.086 |
| Attractiveness → Likeability | 0.167 | 0.167 |
| Corporate social responsibility → Competence | 0.059 | 0.059 |
| Corporate social responsibility → Likeability | 0.178 | 0.178 |
| Performance → Competence | 0.295 | 0.295 |
| Performance → Likeability | 0.117 | 0.117 |
| Quality → Competence | 0.430 | 0.430 |
| Quality → Likeability | 0.380 | 0.380 |
| Competence → Customer satisfaction | 0.146 | 0.146 |
| Competence → Customer loyalty | 0.006 | 0.006 |
| Likeability → Customer satisfaction | 0.436 | 0.436 |
| Likeability → Customer loyalty | 0.344 | 0.344 |
| Customer satisfaction → Customer loyalty | 0.505 | 0.505 |
R-squared: explained variation in four outcomes
R-squared describes how much variation in an outcome score the model accounts for in these cases. Customer loyalty has R-squared 0.562 in both programs, or about 56.2% of its variation in this sample.
Customer satisfaction has R-squared 0.292. That figure answers a different question from whether its measurement is reliable. It also says nothing by itself about prediction for customers outside this dataset.
On a small screen, swipe the table sideways to see every column.
| Result | QuickPLS | SmartPLS |
|---|---|---|
| Competence | 0.631 | 0.631 |
| Likeability | 0.558 | 0.558 |
| Customer satisfaction | 0.292 | 0.292 |
| Customer loyalty | 0.562 | 0.562 |
Loadings and weights
Reflective loadings: the ten measurement relationships
A loading describes how closely an indicator follows its construct score. For example, comp_2 has a loading of 0.821 in both columns. Read it as a measurement result, separate from the path between Competence and an outcome.
The single cusa indicator has a loading of 1.000 because it supplies the Customer satisfaction score on its own. That value does not demonstrate perfect measurement or provide an internal-consistency test.
On a small screen, swipe the table sideways to see every column.
| Result | QuickPLS | SmartPLS |
|---|---|---|
| Competence / comp_1 | 0.824 | 0.824 |
| Competence / comp_2 | 0.821 | 0.821 |
| Competence / comp_3 | 0.844 | 0.844 |
| Likeability / like_1 | 0.880 | 0.880 |
| Likeability / like_2 | 0.869 | 0.869 |
| Likeability / like_3 | 0.844 | 0.844 |
| Customer satisfaction / cusa | 1.000 | 1.000 |
| Customer loyalty / cusl_1 | 0.833 | 0.833 |
| Customer loyalty / cusl_2 | 0.917 | 0.917 |
| Customer loyalty / cusl_3 | 0.843 | 0.843 |
Formative weights: the 21 contributions to the composites
A weight determines an indicator’s contribution to forming the composite score. The Performance indicator perf_1 has a weight of 0.468 in both columns. It is a contribution within the Performance measurement block, not the effect of Performance on Competence.
The small negative weight for qual_4 is -0.005 in both programs. Keep that sign. A small or negative formative weight calls for checking the indicator’s meaning, overlap with other indicators and uncertainty. It is not a reason to delete the indicator automatically.
On a small screen, swipe the table sideways to see every column.
| Result | QuickPLS | SmartPLS |
|---|---|---|
| Attractiveness / attr_1 | 0.414 | 0.414 |
| Attractiveness / attr_2 | 0.201 | 0.201 |
| Attractiveness / attr_3 | 0.658 | 0.658 |
| Corporate social responsibility / csor_1 | 0.306 | 0.306 |
| Corporate social responsibility / csor_2 | 0.037 | 0.037 |
| Corporate social responsibility / csor_3 | 0.406 | 0.406 |
| Corporate social responsibility / csor_4 | 0.080 | 0.080 |
| Corporate social responsibility / csor_5 | 0.416 | 0.416 |
| Performance / perf_1 | 0.468 | 0.468 |
| Performance / perf_2 | 0.177 | 0.177 |
| Performance / perf_3 | 0.194 | 0.194 |
| Performance / perf_4 | 0.340 | 0.340 |
| Performance / perf_5 | 0.199 | 0.199 |
| Quality / qual_1 | 0.202 | 0.202 |
| Quality / qual_2 | 0.041 | 0.041 |
| Quality / qual_3 | 0.106 | 0.106 |
| Quality / qual_4 | -0.005 | -0.005 |
| Quality / qual_5 | 0.160 | 0.160 |
| Quality / qual_6 | 0.398 | 0.398 |
| Quality / qual_7 | 0.229 | 0.229 |
| Quality / qual_8 | 0.190 | 0.190 |
Formative loadings: the other 21 indicator relationships
QuickPLS also reports loadings for formative indicators. These describe each indicator’s relationship with the resulting composite score. They answer a different question from the weights above. For instance, perf_1 has loading 0.846 and weight 0.468; neither number should replace the other.
The SmartPLS column uses the algorithm loading matrix in the published chapter. All 21 entries agree at three decimals. The later summary gives a different value for attr_1, which is shown in the next section.
On a small screen, swipe the table sideways to see every column.
| Result | QuickPLS | SmartPLS |
|---|---|---|
| Attractiveness / attr_1 | 0.754 | 0.754 |
| Attractiveness / attr_2 | 0.506 | 0.506 |
| Attractiveness / attr_3 | 0.891 | 0.891 |
| Corporate social responsibility / csor_1 | 0.771 | 0.771 |
| Corporate social responsibility / csor_2 | 0.571 | 0.571 |
| Corporate social responsibility / csor_3 | 0.838 | 0.838 |
| Corporate social responsibility / csor_4 | 0.617 | 0.617 |
| Corporate social responsibility / csor_5 | 0.848 | 0.848 |
| Performance / perf_1 | 0.846 | 0.846 |
| Performance / perf_2 | 0.690 | 0.690 |
| Performance / perf_3 | 0.573 | 0.573 |
| Performance / perf_4 | 0.717 | 0.717 |
| Performance / perf_5 | 0.638 | 0.638 |
| Quality / qual_1 | 0.741 | 0.741 |
| Quality / qual_2 | 0.570 | 0.570 |
| Quality / qual_3 | 0.749 | 0.749 |
| Quality / qual_4 | 0.664 | 0.664 |
| Quality / qual_5 | 0.787 | 0.787 |
| Quality / qual_6 | 0.856 | 0.856 |
| Quality / qual_7 | 0.722 | 0.722 |
| Quality / qual_8 | 0.627 | 0.627 |
Additional paired diagnostics
f-square: the contribution of each structural predictor
f-square describes how much a predictor contributes to explaining its outcome, relative to the unexplained variation. Customer satisfaction → Customer loyalty has f-square 0.412 in both columns. Its path coefficient was 0.505; these two statistics describe different aspects of the same relationship.
Competence → Customer loyalty displays 0.000. The saved QuickPLS value is approximately 0.00004269, so the displayed zero means very small at this precision. It is not an exact zero or a significance test.
On a small screen, swipe the table sideways to see every column.
| Result | QuickPLS | SmartPLS |
|---|---|---|
| Attractiveness → Competence | 0.009 | 0.009 |
| Attractiveness → Likeability | 0.030 | 0.030 |
| Competence → Customer satisfaction | 0.018 | 0.018 |
| Competence → Customer loyalty | 0.000 | 0.000 |
| Corporate social responsibility → Competence | 0.005 | 0.005 |
| Corporate social responsibility → Likeability | 0.035 | 0.035 |
| Customer satisfaction → Customer loyalty | 0.412 | 0.412 |
| Likeability → Customer satisfaction | 0.159 | 0.159 |
| Likeability → Customer loyalty | 0.138 | 0.138 |
| Performance → Competence | 0.082 | 0.082 |
| Performance → Likeability | 0.011 | 0.011 |
| Quality → Competence | 0.143 | 0.143 |
| Quality → Likeability | 0.094 | 0.094 |
HTMT: six comparisons between the reflective constructs
HTMT helps assess whether two reflective constructs are sufficiently distinct. It compares relationships between indicators from different constructs with relationships among indicators of the same construct. Competence and Likeability have HTMT 0.780 in both columns. To judge whether the constructs are distinct in a study, also examine their item wording, the chosen assessment criteria and confidence intervals.
The published HTMT calculation uses absolute indicator correlations. QuickPLS calls this output HTMT+. The source explains that adding the formative constructs leaves these six pairs unchanged, so the published matrix applies to the extended example. Formative constructs are not included in this table.
These are point values. No HTMT confidence interval has been compared here. Customer satisfaction is a single-item construct, so interpret its pairs with that measurement choice in mind.
On a small screen, swipe the table sideways to see every column.
| Result | QuickPLS | SmartPLS |
|---|---|---|
| Customer satisfaction / Competence | 0.465 | 0.465 |
| Customer loyalty / Competence | 0.532 | 0.532 |
| Customer loyalty / Customer satisfaction | 0.755 | 0.755 |
| Likeability / Competence | 0.780 | 0.780 |
| Likeability / Customer satisfaction | 0.577 | 0.577 |
| Likeability / Customer loyalty | 0.737 | 0.737 |
AVE: the published Competence value
Average variance extracted, or AVE, summarises how much indicator variation a reflective construct represents on average. The extended-model discussion reports Competence AVE as 0.688, about 68.8%. The saved QuickPLS result also rounds to 0.688.
This is one matched AVE value. The complete QuickPLS reliability and AVE output follows separately; it is not presented as a complete paired SmartPLS table.
On a small screen, swipe the table sideways to see every column.
| Result | QuickPLS | SmartPLS |
|---|---|---|
| Competence | 0.688 | 0.688 |
QuickPLS outputs without a complete paired reference
Full reliability and AVE table: QuickPLS results
This table is the saved QuickPLS output. It shows where the familiar reliability measures appear and keeps the inapplicable entries visible. Only the Competence AVE value was paired with the published extended-model reference above.
For Competence, Cronbach’s alpha is 0.776, rho_A is 0.786, composite reliability is 0.869 and AVE is 0.688. Reliability describes consistency among the indicators; AVE concerns the variation represented by the reflective construct. Neither is a replacement for inspecting the item content.
The four formative constructs have no reflective reliability or AVE values. Customer satisfaction has one indicator: alpha and rho_A are undefined here. Its displayed 1.000 values are a consequence of the single-item setup, not evidence that the measurement has no error.
The published chapter 4 reliability table uses the smaller model. Adding the formative predictors changes some loadings and related measures, so that table has not been used as a matching reference for the extended-model results below.
On a small screen, swipe the table sideways to see every column.
| Construct | Cronbach’s alpha | rho_A | Composite reliability | AVE |
|---|---|---|---|---|
| Attractiveness | Not applicable | Not applicable | Not applicable | Not applicable |
| Competence | 0.776 | 0.786 | 0.869 | 0.688 |
| Corporate social responsibility | Not applicable | Not applicable | Not applicable | Not applicable |
| Customer satisfaction | Not defined for one item | Not defined for one item | 1.000 | 1.000 |
| Customer loyalty | 0.831 | 0.839 | 0.899 | 0.748 |
| Likeability | 0.831 | 0.836 | 0.899 | 0.747 |
| Performance | Not applicable | Not applicable | Not applicable | Not applicable |
| Quality | Not applicable | Not applicable | Not applicable | Not applicable |
Adjusted R-squared: QuickPLS results
Adjusted R-squared accounts for the number of predictors as well as the explained variation. For Customer loyalty, the saved result is R-squared 0.562 and adjusted R-squared 0.558. The adjustment is useful when you describe the fitted equation; it does not measure prediction on new data.
The source pages checked for this comparison do not supply a corresponding adjusted-R-squared table for the extended model. The values below are therefore shown as QuickPLS results only.
On a small screen, swipe the table sideways to see every column.
| Outcome | QuickPLS R-squared | QuickPLS adjusted R-squared |
|---|---|---|
| Competence | 0.631 | 0.627 |
| Customer satisfaction | 0.292 | 0.288 |
| Customer loyalty | 0.562 | 0.558 |
| Likeability | 0.558 | 0.552 |
Source discrepancy and comparison scope
The dataset and model
Corporate Reputation is a sample project about the relationships between a company’s reputation, customer satisfaction and loyalty. The extended model uses 344 cases, eight constructs and 31 indicators. It includes both reflective and formative measurement.
The QuickPLS calculation uses the same prepared data and model specification as the SmartPLS teaching example. The SmartPLS column is taken from the published teaching tables, linked at the end of this page.
The main estimates were calculated on 6 November 2026 by running the QuickPLS calculation code outside the desktop app. The additional diagnostic tables use a saved application calculation from 3 November. The software details are recorded below.
What “matching” means here
The published SmartPLS tables report three decimal places. We round the full-precision QuickPLS results directly to three decimals before comparing them.
The app screenshots show four decimals. Use the paired tables for the comparison: rounding a screenshot a second time can give a different last digit.
The source gives two values for attr_1
The same chapter prints attr_1 as 0.754 in its algorithm loading matrix and 0.755 in a later summary. The QuickPLS result rounds directly to 0.754.
The comparison above uses the algorithm matrix, where the two results agree. Against the later summary, the displayed difference is -0.001. Both comparisons are shown below. The original full-precision SmartPLS result is unavailable here, so we have not assigned a cause to the difference.
On a small screen, swipe the table sideways to see every column.
| Published source for attr_1 | QuickPLS | SmartPLS | QuickPLS minus SmartPLS |
|---|---|---|---|
| Algorithm loading matrix, Exhibit A5.8, page 11 | 0.754 | 0.754 | 0.000 |
| Later summary, Exhibit A5.16, page 22 | 0.754 | 0.755 | -0.001 |
Unrounded QuickPLS loading
The fresh QuickPLS value is 0.7544992556030526. It rounds to 0.754 at three decimals. This unrounded value is available only for QuickPLS; the published SmartPLS source supplies the rounded values shown above.
Check effect sizes and measurement diagnostics
The next three tables compare 20 additional results from a QuickPLS application calculation saved on 3 November 2026. It uses the same prepared Corporate Reputation data and model. All 69 estimates shared with the fresh calculation also agree at the displayed precision.
The saved calculation supplies the effect sizes, HTMT values and AVE comparison below. They were checked against the published tables and were not rerun for this page. You can inspect the dates and software versions in the calculation records.
How the saved calculation was connected to this example
The saved project contains the exact 344-case input, the 31 indicator assignments and the 13 structural paths. These were checked against the fresh calculation. The largest absolute difference across the 69 overlapping point estimates is 3.3306690738754696e-16.
These values were read from the saved application result. The new calculation performed outside the desktop app supplies the 69 point estimates above; the separately identified application result supplies these additional diagnostics.
Repeat the model and check a difference
Use the extended Corporate Reputation model
The example studies how perceived company qualities relate to competence and likeability, and how those two reputation dimensions relate to customer satisfaction and loyalty. The table gives the exact measurement assignments. Keep the indicator names when you prepare the data.
The four formative constructs use their indicators to form the composite. The reflective constructs use indicators as measures of the construct. Customer satisfaction has one indicator, cusa. Changing a measurement direction changes the model being compared.
- Draw paths from Attractiveness, Corporate social responsibility, Performance and Quality to both Competence and Likeability. These are eight paths.
- Draw Competence → Customer satisfaction, Competence → Customer loyalty, Likeability → Customer satisfaction and Likeability → Customer loyalty.
- Add Customer satisfaction → Customer loyalty. The completed model has 13 paths.
On a small screen, swipe the table sideways to see every column.
| Construct | Indicators | Measurement |
|---|---|---|
| Competence | comp_1, comp_2, comp_3 | Reflective |
| Likeability | like_1, like_2, like_3 | Reflective |
| Customer loyalty | cusl_1, cusl_2, cusl_3 | Reflective |
| Customer satisfaction | cusa | Reflective, one indicator |
| Corporate social responsibility | csor_1, csor_2, csor_3, csor_4, csor_5 | Formative |
| Attractiveness | attr_1, attr_2, attr_3 | Formative |
| Performance | perf_1, perf_2, perf_3, perf_4, perf_5 | Formative |
| Quality | qual_1, qual_2, qual_3, qual_4, qual_5, qual_6, qual_7, qual_8 | Formative |
Prepare the same 344 cases
Download the project from the official SmartPLS page. In the archive, open Example - Corporate reputation (primer) and extract Complete data.txt. This is a semicolon-separated data table. Open it with the separator set to semicolon, then keep the 31 model columns listed above. Preserve the original row order and all 344 cases.
The current official file has 11 empty measurement cells. The published example replaces each missing answer with the mean of the observed answers for that indicator. Use the values below only in the missing cells of the named columns. Keep the observed answers unchanged.
An older workbook used -99 instead of an empty cell. If your source file contains that code, treat it as missing before calculating a mean. Do not include -99 in the average. These are preparation steps for this published example, not a recommendation to use mean replacement for every study.
Save a separate prepared CSV
Use comma-separated CSV with a header row. Keep the raw source file as well. The prepared file must contain 344 data rows, the 31 named numeric columns and no missing model values.
Check the import before drawing the model
Import the prepared CSV into QuickPLS. Confirm 344 cases and inspect the four columns that contained missing answers. Do not analyse the unprepared file by deleting its incomplete cases.
On a small screen, swipe the table sideways to see every column.
| Column | Missing cells | Value used in the prepared CSV |
|---|---|---|
| cusl_1 | 3 | 5.12903225806452 |
| cusl_2 | 4 | 5.27647058823529 |
| cusl_3 | 3 | 5.65102639296188 |
| cusa | 1 | 5.4402332361516 |
Exact preparation and file check
The benchmark keeps this column order: comp_1, comp_2, comp_3, like_1, like_2, like_3, cusl_1, cusl_2, cusl_3, cusa, csor_1, csor_2, csor_3, csor_4, csor_5, attr_1, attr_2, attr_3, perf_1, perf_2, perf_3, perf_4, perf_5, qual_1, qual_2, qual_3, qual_4, qual_5, qual_6, qual_7, qual_8.
Follow the preparation above and compare the recorded settings in the download. The comparison uses the same data values and model; spreadsheet programs may save the CSV text differently.
Repeating the preparation on the official project downloaded on 6 November 2026 recreated the same benchmark input. Download the dataset and project from the official source linked below.
Calculate the model and check the case count
Create the eight constructs with the assignments and paths above. Click Calculate… and select PLS-SEM Algorithm. Leave Starting weights at Standard (+1). Under Missing data, set Treatment to Casewise deletion for the prepared file.
The file has already had its missing answers replaced, so casewise deletion removes no rows. This is how the QuickPLS comparison retains the same 344 cases as the published mean-replacement analysis.
Click Start calculation and inspect the completed result. The recorded calculation converged in eight iterations, used 344 cases and omitted none. The table below lists the calculation settings used for the comparison.
| Calculation choice | Recorded value |
|---|---|
| Method | Ordinary PLS-SEM |
| Weighting scheme | Path |
| Indicator scale | Standardised |
| Reflective blocks | Mode A |
| Formative blocks | Mode B |
| Starting weights | +1 for every indicator |
| Maximum iterations | 3,000 |
| Stopping criterion | 0.0000001 |
| Missing-data preparation | Indicator mean replacement before QuickPLS import |
| Missing-data treatment during the fit | Casewise deletion; no incomplete rows remain |
| Bootstrap samples in this comparison | 0 |
Convergence and missing-data details
The ordinary raw-data dialog provides Starting weights and Missing data controls. Path weighting, the maximum iteration count and the stopping criterion are listed here to describe the calculation; they are not separate fields to edit in that dialog.
Convergence means the repeated estimation steps reached the stated stopping rule. It does not establish that the research model is correct. The fresh calculation reached a maximum outer-weight change of 2.851066022158122e-8 after eight iterations, below the 1e-7 stopping criterion.
The saved 3 November application calculation uses the same 344 prepared cases, indicator assignments and paths. All 10,664 numeric input cells were checked against the fresh example. No case weights or group filter were used.
Find the familiar result tables
Open Overview → Graphical Output for the model picture, then use the folders below to find each result table. Keep the Corporate Reputation result selected while moving between them. A different saved model can have similar construct names and different numbers.
The screenshots below show where to find these tables in the bundled sample project. The paired tables on this page contain the recorded benchmark values; a screenshot is an aid to navigation.
| If you want to inspect | Open in QuickPLS |
|---|---|
| Relationships between constructs | Effects → Direct effects |
| Explained variation | Estimates → R-squared |
| Indicator relationships | Estimates → Outer loadings |
| Indicator contributions to scores | Estimates → Outer weights |
| Reliability and average variance extracted | Reliability and validity → Construct reliability and validity |
| Separation between constructs | Discriminant validity → HTMT+ |
| Contribution of each structural predictor | Collinearity and effects → f-square effect sizes |
| Explained variation adjusted for the number of predictors | Model fit → R-square and adjusted R-square |
Repeat the comparison in your own project
Use the official data and the model assignments above. A different Corporate Reputation dataset or the Simple model is a different analysis. In particular, the separate 308-case dataset linked from some publications is not this 344-case example.
Prepare the 344-case file
Follow the missing-value replacement table and keep only the 31 named indicators. Confirm there are no missing model values before using Casewise deletion in QuickPLS.
Check the measurement and structural model
Confirm four formative constructs, three reflective multi-item constructs and single-item Customer satisfaction. Trace all 13 paths against the model list.
Calculate ordinary PLS-SEM
Select Standard (+1) starting weights and use the prepared data. Keep this first comparison separate from consistent PLS, bootstrapping and prediction.
Check the completed result
Confirm convergence, 344 used observations and zero omitted observations. The recorded comparison converged after eight iterations.
Compare each named result
Use the exact indicator or source-to-target path label, rather than a row number. Compare values rounded directly to three decimals. Keep the attr_1 source note with the loading table.
Save what you compared
Save the result and export the tables you inspected. Keep the prepared file, model, settings and software version with your notes so another researcher can identify the same calculation.
If your results differ, check the setup first
A difference can identify a changed data treatment, measurement assignment or result definition. Work through the relevant row before changing observations or removing an indicator. Keep the original result so you can explain what changed.
On a small screen, swipe the table sideways to see every column.
| What you see | Check | What to do next |
|---|---|---|
| Fewer than 344 used cases | The raw incomplete file may have been analysed with casewise deletion. | Prepare the missing entries as documented, import a separate prepared file and recalculate. |
| A different Competence loading or AVE | Check whether you opened the Simple model rather than the extended eight-construct model. | Use all eight constructs and the 13 listed paths. |
| A weight compared with a different-looking number | Check whether one output is a loading and the other an outer weight. | Read the named output in each program; keep the two quantities separate. |
| HTMT appears to differ by 0.001 | Check whether you rounded a four-decimal screenshot a second time. | Round the unrounded result directly to three decimals. |
| attr_1 differs from the later source summary | Check the page and exhibit: A5.8 prints 0.754; A5.16 prints 0.755. | Record which source you used and retain the 0.001 displayed difference. |
| Bootstrap or prediction output differs | Those procedures add sampling and prediction settings beyond this point-estimate example. | Set up a separate comparison; these tables do not provide bootstrap or prediction answers. |
Reporting and calculation records
Describe this result comparison
A useful report names the example, settings and results checked. It also distinguishes agreement with a rounded published table from a comparison of two full-precision exports. If you repeat this exercise, replace the software and calculation dates with the ones you actually used.
- This is one dataset and one model, with several output values checked.
- The comparison does not cover bootstrap standard errors, p-values, confidence intervals, prediction results or other algorithms.
- The published reference names SmartPLS 4 but does not identify an exact minor version. The QuickPLS build details are recorded below.
Wording for this recorded check
The QuickPLS calculation engine reproduced 69 published SmartPLS algorithm values to three decimal places for the 344-case extended Corporate Reputation model. These comprised 13 paths, four R-squared values, 21 formative weights and 31 loadings. The same model and indicator-mean replacement were used. One later published summary gives attr_1 loading as 0.755 rather than the algorithm table’s 0.754; QuickPLS rounds to 0.754. A separately saved QuickPLS application result also matched 13 f-square values, six HTMT point values and the published Competence AVE at three decimals.
How the comparison was made
Download the values behind these tables, including the unrounded QuickPLS results and links to the published SmartPLS references. The file also records which software and settings were used.
Calculation dates and sources
The main estimates were calculated on 6 November 2026 using QuickPLS calculation code, outside the desktop app. The download records the software and calculation settings.
The additional diagnostics came from a saved QuickPLS application result completed on 3 November 2026. They were checked against the published reference; that calculation was not rerun for this page.
The images illustrate the current sample project and its result navigation. They are separate from the recorded benchmark calculations.
The SmartPLS values come from its official teaching case. The comparison uses the precision printed in those tables. The download links each set of values to its source.