#SmartPLS4 Series 25 - How to Report the Structural Model/Hypotheses Results?

Q: What information is needed to report a direct structural path hypothesis in SmartPLS?

For each hypothesized direct relationship, the reporting template uses the path coefficient (beta/weight of impact), the t statistic, and the p value. The conclusion line then states whether the hypothesis is supported based on whether the p value meets the significance threshold (e.g., p < 0.01 is used when p is effectively 0).

Q: How does the transcript show reporting when a path is significant?

For H1 (internal marketing → organizational performance), the results show a significant positive effect. The beta is reported alongside a t statistic of 7.87, and the p value is effectively 0, which is written as p < 0.01. The final line states: hence H1 was supported.

Q: How does the transcript show reporting when a path is insignificant?

For H6 (perceived organizational support → organizational performance), the effect is insignificant. The beta is −0.018, the t statistic is 0.298, and the p value is 0.383. The write-up concludes: hence H6 was not supported. The same beta–t–p structure is used for each insignificant hypothesis.

Q: What practical steps are recommended to turn SmartPLS structural results into a publishable table?

Export the results to CSV, paste into Excel, delete the sample column, format cells to three decimal points, then copy into Word. Clean up borders (remove unnecessary top/bottom borders), rename columns (beta coefficient as weight of impact, standard deviation, t statistics), and align hypothesis/path labels (e.g., H1 corresponds to IM2 → OP). Add notes to define any abbreviations.

Q: How is the structural model figure incorporated into the report?

The graphical output diagram is rerun to display path estimates, then pasted into the report. This figure complements the table of direct relationships by visually showing the hypothesized paths and their estimated values.

TL;DR

After validating the measurement model, structural model assessment tests whether each hypothesized path is significant using beta, t statistics, and p values.

Briefing Cornell Notes

Briefing

Structural model reporting in SmartPLS hinges on one decision: whether each hypothesized path between constructs is statistically significant. After the measurement model is validated, the analysis moves to the structural model to test direct relationships—such as internal marketing (IM) predicting organizational performance (OP)—and to determine which hypotheses are supported or rejected based on path coefficients (betas), t statistics, and p values.

In the example template used for reporting, multiple direct paths are specified from internal marketing dimensions to organizational performance. The core hypothesis H1 tests whether internal marketing has a significantly positive effect on organizational performance. The results show a significant and positive relationship, with the reported beta value corresponding to a t statistic of 7.87 and a p value effectively at 0, which is then written in the conventional format as p < 0.01. With that evidence, the reporting line concludes that H1 is supported.

The reporting workflow also covers the opposite case—insignificant effects—using the same structure. For a hypothesis such as perceived organizational support (POS) predicting organizational performance, the results indicate an insignificant impact. The beta is negative (−0.018), the t statistic is 0.298 (described as almost negligible), and the p value is 0.383. Because the p value does not meet the significance threshold, the write-up concludes that the corresponding hypothesis (labeled H6 in the example) is not supported. This pattern is repeated for each direct relationship from H1 through H9: report the beta direction and magnitude, the t statistic, the p value, and then explicitly state whether the hypothesis is supported.

Once all direct effects are summarized, the results are presented in a table. The transcript gives practical formatting guidance for exporting SmartPLS output to a publication-ready table: export to CSV, paste into Excel, remove the sample column, format numeric cells to three decimal places, and then transfer the table into a Word document. It also recommends cleaning up table borders (e.g., removing extra borders and controlling top/bottom lines), renaming columns for clarity (beta coefficient as the weight of impact, standard deviation, and t statistics), and aligning the hypothesis labels with the corresponding paths (for instance, mapping H1 to IM2 → OP). If abbreviations are used, they should be defined in notes.

Finally, the structural model figure is generated from SmartPLS graphical output. The diagram is rerun to obtain the version with path estimates and then pasted into the report. The overall takeaway is that direct relationship reporting is systematic: significance testing determines hypothesis support, and consistent tabular and figure formatting turns SmartPLS estimates into readable results. Subsequent sessions are positioned to extend this same reporting discipline to mediation and moderation analyses.

Cornell Notes

The structural model stage in SmartPLS is where hypothesized paths are tested for significance after the measurement model is validated. Each direct relationship is reported using the path coefficient (beta), t statistic, and p value, followed by a clear conclusion that the hypothesis is supported or not supported. In the example, internal marketing → organizational performance is significant (beta positive, t = 7.87, p < 0.01), so H1 is supported. For perceived organizational support → organizational performance, the effect is insignificant (beta = −0.018, t = 0.298, p = 0.383), so H6 is not supported. Results are then organized into a cleaned table (export CSV, format to three decimals, align hypotheses to paths) and accompanied by a structural model figure with estimates.

What information is needed to report a direct structural path hypothesis in SmartPLS?

For each hypothesized direct relationship, the reporting template uses the path coefficient (beta/weight of impact), the t statistic, and the p value. The conclusion line then states whether the hypothesis is supported based on whether the p value meets the significance threshold (e.g., p < 0.01 is used when p is effectively 0).

How does the transcript show reporting when a path is significant?

For H1 (internal marketing → organizational performance), the results show a significant positive effect. The beta is reported alongside a t statistic of 7.87, and the p value is effectively 0, which is written as p < 0.01. The final line states: hence H1 was supported.

How does the transcript show reporting when a path is insignificant?

For H6 (perceived organizational support → organizational performance), the effect is insignificant. The beta is −0.018, the t statistic is 0.298, and the p value is 0.383. The write-up concludes: hence H6 was not supported. The same beta–t–p structure is used for each insignificant hypothesis.

What practical steps are recommended to turn SmartPLS structural results into a publishable table?

Export the results to CSV, paste into Excel, delete the sample column, format cells to three decimal points, then copy into Word. Clean up borders (remove unnecessary top/bottom borders), rename columns (beta coefficient as weight of impact, standard deviation, t statistics), and align hypothesis/path labels (e.g., H1 corresponds to IM2 → OP). Add notes to define any abbreviations.

How is the structural model figure incorporated into the report?

The graphical output diagram is rerun to display path estimates, then pasted into the report. This figure complements the table of direct relationships by visually showing the hypothesized paths and their estimated values.

Review Questions

When writing results for a direct path, what three statistics must be reported before stating whether the hypothesis is supported?
How would you format the p value if SmartPLS reports it as 0—what threshold notation is used in the transcript?
What formatting and alignment steps are suggested to ensure the exported beta/t/p table matches the correct hypotheses (e.g., H1 mapping to IM2 → OP)?

Key Points

1
After validating the measurement model, structural model assessment tests whether each hypothesized path is significant using beta, t statistics, and p values.
2
A significant positive path (e.g., internal marketing → organizational performance) is reported with a positive beta, a large t statistic (7.87 in the example), and p < 0.01, followed by a supported conclusion.
3
An insignificant path is reported with its beta (including sign), a small t statistic (0.298 in the example), and a p value above the threshold (0.383), followed by a not supported conclusion.
4
Each hypothesis (H1 through H9 in the template) should follow the same reporting pattern: beta, t, p, then an explicit support/not-support statement.
5
Results should be presented in a cleaned table: export to CSV, remove the sample column, format to three decimals, and standardize column labels (beta/standard deviation/t).
6
Hypothesis labels must be rearranged to match the corresponding regression weights/paths (e.g., H1 mapped to IM2 → OP).
7
A structural model figure with estimates should be generated from graphical output and pasted alongside the table for clarity.

Highlights

Internal marketing → organizational performance is supported with t = 7.87 and p < 0.01, using a beta that indicates a significant positive effect.

Perceived organizational support → organizational performance is not supported: beta = −0.018, t = 0.298, p = 0.383.

Reporting consistency matters: every hypothesis gets the same beta–t–p structure plus a clear supported/not supported conclusion.

A publication-ready table requires cleanup after export—removing the sample column, formatting to three decimals, and aligning hypothesis/path labels.

The structural model diagram should be rerun to include estimates and then pasted into the report.

Topics

Structural Model Reporting
Direct Relationships
SmartPLS Hypotheses
P Values and Significance
Results Tables