15. SEMinR Series. Reporting Measurement Model Results

Once a PLS-SEM model is estimated in R with the measurement and structural models set up, results can be reported in a thesis or paper by pulling key measurement-model outputs from the SEMinR “summary” object and exporting them into clean tables. The core workflow is: estimate the model, store outputs in the summary object, then extract factor loadings, reliability, convergent validity, and discriminant validity results—typically by writing them to CSV files—so they can be pasted into Word with consistent formatting (e.g., rounding to three decimals).

Reporting starts with factor loadings for each indicator. After loading data, checking it, specifying the measurement model with multiple latent variables and the structural model with structural relationships, the PLS algorithm is run and the results are saved in a summary object. Calling the relevant subobject (via the object’s loading-related component) displays the factor loadings directly in R. Instead of manually copying from the console, the workflow recommends exporting the loadings to a CSV file in the same directory as the R script, then opening that file to copy into a paper table. For presentation, the transcript suggests formatting numeric cells to three decimal places and optionally removing leading zeros. In a thesis-style table, loadings can be organized under headings for each construct (e.g., “Vision,” “Development”), with reliability and validity columns placed alongside the loadings.

Next comes reliability analysis. The same general approach applies: extract reliability results from the SEMinR outputs, export them to a CSV file, and format them to three decimals before pasting into the document. The reliability table includes the values needed for reliability reporting (the transcript specifically references AV values as part of the reliability table layout). After reliability, convergent validity is reported using a convergent validity function, which produces results indicating whether convergent validity was established. Those outputs are then copied into a dedicated table.

Finally, discriminant validity is handled through a discriminant validity function that generates multiple assessment methods. The transcript lists the specific discriminant validity checks to report: the Fornell–Larcker Criterion, cross loadings, and HTMT. Each of these results can be exported to CSV from the same directory as the R file, then pasted into the thesis as part of a discriminant validity table. Taken together, the process turns SEMinR’s measurement-model outputs into publication-ready tables: factor loadings first, then reliability and convergent validity, and concluding with discriminant validity using multiple criteria.