QuickPLS Does an environmental profile connect alignment with performance?
A business can report alignment with an environmental classification without immediately showing stronger financial performance. A mediation model can distinguish the direct association from an indirect route through its environmental, social and governance profile. An uncertain indirect result remains informative if the measures and model are clearly defined.
Ramírez-Orellana and colleagues examined 217 pharmaceutical firms using reported company information and PLS modelling. Their study connected EU Taxonomy alignment, an ESG profile and financial performance. This lesson uses original assessment indicators to teach measurement direction, covariate adjustment and an imprecise indirect effect. Its data are not company accounts, taxonomy disclosures or a reproduction of the paper's indicators.
Keep the teaching ratings distinct from financial records
Use 250 fictional pharmaceutical businesses. ALIGN, PROFILE and PERF each have three invented positively directed indicators. SIZE and DEBT are single continuous teaching scores for organisational scale and financing burden. Each row is a business.
Specify ALIGN → PROFILE, PROFILE → PERF and ALIGN → PERF. Add SIZE and DEBT as predictors of both PROFILE and PERF. The two covariates contribute four paths, giving seven paths overall. They are included because the practice question asks for associations at comparable scale and financing burden.
All scores use a continuous 1–7 teaching scale. PERF is a perceived performance assessment, not return on assets, return on equity or a market valuation. DEBT is an invented burden score, not a leverage ratio. The distinction matters because units and allowable values would be different in real accounts.
The reflective blocks are a teaching simplification. A collection of financial ratios or taxonomy percentages should not automatically be declared reflective merely because the software can estimate a Mode A model.
On a small screen, swipe the table sideways to see every column.
| Columns | Meaning | Role |
|---|---|---|
| ALIGN1–ALIGN3 | Environmental alignment assessment | Reflective predictor |
| PROFILE1–PROFILE3 | Responsible-business profile | Reflective mediator |
| PERF1–PERF3 | Performance assessment | Reflective outcome |
| SIZE | Organisational scale score | Observed covariate |
| DEBT | Financing burden score | Observed covariate |
| case_id | Fictional business identifier | Not an indicator |
Step 1. Resolve units and directions before estimating paths
In a writable project choose File > Import Data… and Choose data file…. In Review & import confirm raw data, Header row 1, comma separator, Decimal convention Dot (1.25) and Continuous analytical columns. Choose Import dataset and confirm Dataset imported and verified.
Check the 250 complete rows. All supplied indicators have positive directions, and no reversal is required. In a real ESG dataset, larger emissions may mean worse environmental performance while larger renewable-energy use may mean better performance. Resolve such meanings explicitly before forming a score.
Do not reverse a financial ratio by subtracting it from 8 just because the teaching indicators use 1–7. Negative profitability, percentages above an expected range and missing disclosures each require a measure-specific interpretation. A missing disclosure is not automatically zero activity.
Align reporting periods in real company data. Combining a current alignment measure with a different year's performance can be a deliberate lagged design, but it must be stated. The synthetic single-period exercise cannot test a lagged causal effect.
Step 2. Fit the mediation model with the stated covariates
Check that the result belongs to the seven-path model with ALIGN, PROFILE, PERF, SIZE and DEBT. SIZE and DEBT are single-indicator covariates; they do not have multi-item reliability estimates to interpret.
Create the model
Choose File > New Model…, select the imported Dataset version and Cases, then Create model. Select each three-column block in Indicators and choose Create construct.
Set construct names and roles
Name ALIGN, PROFILE and PERF in Properties > Model > Name. Set Parameter > Measurement model to Reflective. Create separate single-indicator constructs for SIZE and DEBT; do not include them in the reflective blocks.
Draw seven paths
Use Connect for ALIGN → PROFILE, PROFILE → PERF and ALIGN → PERF. Add SIZE → PROFILE, SIZE → PERF, DEBT → PROFILE and DEBT → PERF. Save the model after checking arrow directions.
Calculate the point estimates
Choose Calculate… and PLS-SEM Algorithm. Keep Starting weights Standard (+1), use Missing data > Treatment Casewise deletion for the complete file, review Readiness and Start calculation with Open Results when finished.
Read measurement before mediation
Inspect convergence, Estimates > Outer loadings, Reliability and validity > Construct reliability and validity, Discriminant validity > HTMT+ and Collinearity and effects > Inner VIF values.
Step 3. Check the fitted answers for these 250 cases
Use all 250 cases in pharmaceutical-taxonomy-mediation.csv and the exact blocks and arrows in this lesson. The check uses ordinary PLS, path weighting, standardised indicators, starting weights +1, a 3,000-iteration limit and stopping criterion 0.0000001. Environmental alignment assessment, Responsible-business profile and Performance assessment are reflective blocks and use Mode A. SIZE and DEBT enter as single observed covariates; the saved calculation represents each as a one-indicator Mode A block.
ALIGN → PROFILE is 0.226, while PROFILE → PERF is 0.100 after the stated controls. Their product is about 0.023. The adjusted direct ALIGN → PERF estimate is 0.009. The small direct estimate does not tell you whether the indirect interval excludes zero.
About these answers: Calculated with the QuickPLS engine and independently checked against cSEM.
On a small screen, swipe the table sideways to see every column.
| Specified path | Point estimate | Reference 95% percentile interval |
|---|---|---|
| Environmental alignment assessment → Responsible-business profile | 0.226 | 0.111 to 0.345 |
| Responsible-business profile → Performance assessment | 0.100 | -0.034 to 0.229 |
| Environmental alignment assessment → Performance assessment | 0.009 | -0.118 to 0.133 |
| Organisational scale score → Responsible-business profile | 0.114 | -0.019 to 0.235 |
| Financing burden score → Responsible-business profile | -0.106 | -0.226 to 0.027 |
| Organisational scale score → Performance assessment | 0.235 | 0.119 to 0.354 |
| Financing burden score → Performance assessment | -0.162 | -0.283 to -0.041 |
How these answers were checked
All 31 point-value comparisons and 7,000 direct-path comparisons from 1,000 common-row fits passed an absolute tolerance of 0.00000001. QuickPLS v1.03.11 calculation code and cSEM 0.7.1 fitted the same file and model directly. These checks were not new runs in the installed app. All 1,000 fits succeeded in both calculations.
The reference drew rows with R seed 71011. Intervals use percentile endpoints with R quantile type 7. Both programs fitted the same sampled rows. This separate calculation did not use the QuickPLS bootstrap command. Entering the same seed in another program need not draw the same rows. The downloads include sampled row numbers, unrounded estimates and settings.
Step 4. Check the measurement values and explained variation
Step 4. Check the measurement values and explained variation
For the reflective blocks in this exercise, Loading relates each indicator to its construct score; Weight shows how the indicator contributes to forming that score. A weight in this table does not make the block formative. SIZE and DEBT each represent a single observed score, so their single-indicator loading and weight do not establish measurement reliability. Assess the multi-item blocks using reliability, discriminant-validity and content checks as well.
About these answers: Point estimates calculated with QuickPLS analysis code and cSEM using the same model and data file.
On a small screen, swipe the table sideways to see every column.
| Block / indicator | Loading | Weight |
|---|---|---|
| Environmental alignment assessment / ALIGN1 | 0.956 | 0.385 |
| Environmental alignment assessment / ALIGN2 | 0.949 | 0.352 |
| Environmental alignment assessment / ALIGN3 | 0.942 | 0.316 |
| Organisational scale score / SIZE | 1.000 | 1.000 |
| Financing burden score / DEBT | 1.000 | 1.000 |
| Responsible-business profile / PROFILE1 | 0.952 | 0.345 |
| Responsible-business profile / PROFILE2 | 0.949 | 0.357 |
| Responsible-business profile / PROFILE3 | 0.952 | 0.349 |
| Performance assessment / PERF1 | 0.955 | 0.358 |
| Performance assessment / PERF2 | 0.945 | 0.333 |
| Performance assessment / PERF3 | 0.949 | 0.362 |
Explained variation in each outcome
R² describes this fitted sample. It is not a test of prediction on new observations.
| Outcome | R² |
|---|---|
| Responsible-business profile | 0.078 |
| Performance assessment | 0.091 |
Step 5. Read each indirect route separately
Each row follows one specific route. Multiply its component coefficients within each repeated fit before taking the interval; multiplying the endpoints of component intervals is incorrect. These are statistical associations in synthetic data, not evidence of causal mediation.
About these answers: Indirect effects recalculated for each shared resampled dataset. These reference intervals are not QuickPLS bootstrap output.
On a small screen, swipe the table sideways to see every column.
| Specific route | Point product | Reference 95% percentile interval |
|---|---|---|
| Environmental alignment assessment → Responsible-business profile → Performance assessment | 0.023 | -0.007 to 0.060 |
| Organisational scale score → Responsible-business profile → Performance assessment | 0.011 | -0.004 to 0.040 |
| Financing burden score → Responsible-business profile → Performance assessment | -0.011 | -0.035 to 0.005 |
Step 6. Understand what adjustment changes
The PROFILE → PERF coefficient is conditional on the other predictors of PERF. It is not the simple association between the two scores. If SIZE or DEBT overlaps strongly with PROFILE, the conditional estimate can differ substantially from an unadjusted relationship.
Controls should follow the study design and causal reasoning. Adding a variable because it happens to change significance is not a defensible control-selection rule. In real research, controlling for a consequence of alignment or for a common effect can introduce bias rather than remove it.
Inspect the reflective blocks for coherent item content and direction. Do not treat a high reliability value as permission to combine incomparable financial and environmental measures. Their measurement relationship must be justified before interpreting an indirect route.
Read R-squared as in-sample explained variation under the specified equations. It does not show that alignment predicts future profitability, and it does not establish that the covariates removed all confounding.
Step 7. Read the indirect effect even if the total association is weak
A significant total effect is not a required first step before examining a prespecified indirect association. Opposing direct and indirect relationships can offset each other. Conversely, clear individual path estimates do not replace uncertainty for their product.
Open the matching bootstrap
Choose Calculate… and PLS-SEM Bootstrapping for the saved seven-path model. Check the model and missing-data treatment.
Set the teaching plan
Enter Bootstrap samples 5000, Confidence level (%) 95 and Seed 71011. Choose Start calculation and open the completed result.
Inspect counts and interval labels
Read requested, attempted, usable and failed resamples and the exact interval method. Do not infer that every interval option described in a textbook is available in this basic panel.
Read matching effect rows
Find the specific indirect row ALIGN → PROFILE → PERF, then the direct ALIGN → PERF and total-effect rows. Check that SIZE and DEBT remain in both equations. Copy each estimate with its own interval from this completed run; the reference intervals below come from a separate repeated-fit calculation.
Describe an interval that allows several plausible conclusions
Do not report a one-sided interpretation after seeing the interval merely to rescue the hypothesised positive direction. Keep the prespecified inferential question and explain the uncertainty.
Hypothetical example
Illustrative reading example
Suppose illustrative ALIGN → PROFILE is 0.30 and PROFILE → PERF is 0.10. The indirect estimate is 0.03. Suppose its 95% interval is [−0.02, 0.09]. This range includes small negative, zero and positive associations.
A useful account is that the indirect relationship was estimated imprecisely under the model. It does not show that the environmental profile has no possible role, and it does not support claiming a positive mechanism. Discuss whether values across the interval would matter in the study.
If the illustrative direct effect is −0.02, the total is 0.01. The near-zero total reflects offsetting point estimates in this example. It is not evidence that the analysis failed or that a mediator should be removed.
Indirect = 0.30 × 0.10 = 0.03; total = 0.03 − 0.02 = 0.01. The interval [−0.02, 0.09] and all coefficients are invented for reading practice, not fitted results or findings copied from the paper.
Use a meaningful comparison, not a search for support
Keep the planned adjusted model as the primary analysis. A separately labelled unadjusted model can illustrate how SIZE and DEBT change the conditional associations. Explain that the comparison changes the estimand, meaning the relationship being estimated, rather than proving one result correct by its significance.
For real company data, investigate defensible alternatives in measurement period, ratio definition and treatment of exceptional accounting values. Do not silently trim valid losses or extreme but genuine firms because their inclusion weakens a path.
If a coding error is found, correct it and rerun measurement, paths and bootstrap together. If the indirect interval remains wide, report that result. More bootstrap draws reduce simulation noise but cannot add independent businesses or repair a poorly measured concept.
Find whether the problem is a unit, a measure or a relationship
On a small screen, swipe the table sideways to see every column.
| Symptom | Check | Action |
|---|---|---|
| Emissions and positive ESG indicators point in opposite directions | Their substantive definitions | Resolve direction with documented coding before score formation. |
| The total effect is near zero | Direct and specific indirect signs | Explain possible cancellation rather than abandoning the mediation question. |
| The indirect interval includes zero | Its range and practical meaning | Report uncertainty without declaring proof of no mechanism. |
| Controls reverse a path sign | Predictor overlap and the control rationale | Explain the conditional estimate and inspect collinearity. |
| A single-indicator covariate lacks reliability statistics | Its measurement structure | Do not invent multi-item reliability for one observed score. |
| A real ratio is treated like a 1–7 response | Units and valid range | Use measure-specific preparation rather than the teaching recode. |
Step 8. Make a qualified finding useful
For Results, insert the indirect estimate and interval first, then the direct and total effects. Describe the plausible range rather than using supported/unsupported labels alone. State how covariate adjustment and any planned measurement alternatives affected the conclusion.
| Report item | Information |
|---|---|
| Measurement | Score definitions, units, direction and indicator roles |
| Adjustment | Covariate rationale and equations entered |
| Indirect route | Product estimate and matching interval |
| Direct and total | Estimates, intervals and any cancellation |
| Sensitivity | Declared alternative and changed interpretation |
| Limits | Synthetic exercise or real observational design boundaries |
Methods writing outline
We fitted an original reflective teaching model linking ALIGN to PERF directly and through PROFILE, with SIZE and DEBT predicting both endogenous constructs. We used [case count], [measurement checks] and a case bootstrap with [settings, interval method and usable counts]. The data are fictional assessment scores rather than company accounts.
Step 9. Put this result into a report
We fitted the specified ordinary PLS model to 250 original synthetic teaching cases. The measurement roles and complete structural model were fixed before interpreting the results. Point estimates were checked against an independent cSEM calculation.
Report each path estimate with its matching interval. If you use the reference intervals here, explain that both programs fitted the same resampled datasets in a separate checking procedure. Do not describe them as results from a QuickPLS command you have not run. For a real study, add the sampling design, measurement evidence and reasons for the model choices.
Keep a weak result scientifically interpretable
- Verify seven paths and two separate covariates.
- Explain why PERF is not a profitability ratio.
- Trace a direction decision to measurement meaning.
- Read the indirect interval without requiring a significant total effect.
- Explain what the covariates change.
- Report uncertainty without changing the hypothesis after seeing it.