How to Read and Present a SmartPLS 4 Results Summary Table

What is a SmartPLS results summary table used for

When you run SmartPLS, you receive separate tables: Outer Loadings, Construct Reliability and Validity, Discriminant Validity, VIF, R Square, f Square, Path Coefficients, and Bootstrapping results. A SmartPLS results summary table is a table you create from those outputs so the reader can see the full process of assessing the measurement model and structural model.

This table is not a single default table generated by the software. You need to export or read each group of results, select the appropriate indicators, and then arrange them according to the logic of your thesis. For a PLS-SEM model, the sensible order is to assess the measurement model first and then draw conclusions about the structural model. This two-step approach is presented in (Anderson and Gerbing, 1988) and is also consistent with the current PLS-SEM procedure in (Hair et al., 2022).

If you have just opened SmartPLS, you can also review the SmartPLS topic to identify the type of analysis you are running. This article focuses on compiling results. It does not replace the process of building the model or processing the original data.

Preparing the model and data

Before creating the table, check that you have the two parts of the model in place. The measurement model connects latent variables to indicators, such as PU1, PU2, and PU3 measuring the Perceived Usefulness construct. The structural model connects latent variables with hypothesized arrows, such as PU affecting BI.

Indicator names should be short, consistent, and unique. An indicator should belong only to the construct specified in the questionnaire. If the data file contains PU1, PU2, and PU3 but SmartPLS uses P1, P2, and P3, the software may not match the correct data columns.

SmartPLS 4 usually accepts data in .csv format. Each row represents one respondent, and each column represents one indicator. The first row contains variable names, while the following rows contain response values. Avoid secondary headings, merged cells, percentage signs, or text in numeric columns. Likert values should be coded as numbers, and reverse-coded variables must be handled before you run the model.

You can also read the guide to reading an HTML report in SmartPLS if your results are in a report file rather than in the main interface. Save a separate project file before changing indicators. Deleting or renaming variables can make it difficult to compare the model with the original questionnaire.

Steps in SmartPLS 4

Step 1: Open the model and check the data

In SmartPLS 4, open the project, select the model, and check the list of constructs and indicators. Double-click the model to confirm that the indicators are connected correctly. If an indicator is missing from a construct, stop here and check the .csv file instead of continuing and correcting the table later.

Select Calculate > PLS-SEM algorithm. In the dialog box, keep the default configuration unless your research design requires a different setup, then select Start calculation. This output is used to read outer loading, reliability, AVE, HTMT, VIF, R², and f².

Step 2: Assess the measurement model

In the results window, open Quality Criteria. The tables you need to record are Outer Loadings, Construct Reliability and Validity, and Discriminant Validity. For HTMT, open the corresponding table under discriminant validity. If you use the Fornell-Larcker criterion, save the correlation table and the square root of AVE as well, if SmartPLS displays them.

Do not draw a conclusion about the scale from a single Cronbach's Alpha value. You need to examine outer loading, Cronbach's Alpha, rho_A or rho_C, Composite Reliability, and AVE together. When an indicator is low, check its content and its correlation with the other indicators before deleting it. Repeatedly deleting variables just to make the table look better can move the scale away from the theoretical model.

Step 3: Assess the structural model

Return to the PLS-SEM results and open Quality Criteria > Collinearity Statistics (VIF), R Square, f Square, and Path Coefficients. VIF helps you check multicollinearity among predictor variables. R² shows how much of the variance in endogenous variables is explained, while f² shows the contribution of one predictor to a dependent variable.

To obtain the p-value and confidence interval, select Calculate > Bootstrapping. In SmartPLS 4, set the number of bootstrap subsamples according to your research design. If you need to follow the convention described in the PLS-SEM literature, 5,000 subsamples are mentioned in (Hair et al., 2022). Select Start calculation, then read Path Coefficients, T Statistics, P Values, and Confidence Intervals.

Step 4: Export the output for checking

On the results page, use the SmartPLS 4 export function to save an HTML or Excel report, depending on your purpose. HTML is convenient when you need to view the complete output structure, while Excel is easier for filtering and arranging the data into Chapter 4 tables. The guide to exporting results to Excel in SmartPLS is useful when you want to preserve the column names from the software.

Create a working file with separate sheets for Measurement Model, Discriminant Validity, Structural Model, and Hypothesis Testing. Do not edit the original export file directly. The original version allows you to trace a number if your supervisor asks why the thesis table contains that result.

Step 5: Create the summary table in thesis order

The summary should contain at least two tables rather than forcing every indicator into one very wide table. Use the first table for the measurement model and the second for the structural model. If your department requires a single table, you can still use column groups or notes to distinguish the two parts.

The table title must state clearly whether the numbers are illustrative or are results from your study. In this guide, the numbers below are illustrative output, not results from a specific project. When you prepare your thesis, replace every number with the output from your own analysis.

Reading the results tables

A measurement model table usually begins with outer loading. Compare each indicator with its corresponding construct, then review reliability and AVE at the construct level. One row should show which construct an indicator belongs to and its outer loading. The construct-level table should then show Alpha, rho_A, Composite Reliability, and AVE.

Illustrative output: measurement model in SmartPLS 4

Read the table horizontally to see how each indicator contributes and vertically to assess whether the construct meets reliability and convergent validity requirements. An outer loading of 0.7 or above is commonly used in PLS-SEM (Chin, 1998). CR from 0.7 and AVE from 0.5 or above are commonly cited benchmarks from (Fornell and Larcker, 1981).

The structural model table should include the hypothesized path, path coefficient, standard deviation or t-value, p-value, and conclusion. If you have a bootstrap confidence interval, you can add the lower and upper bounds for a clearer explanation. Do not put CFI, TLI, or RMSEA in a SmartPLS table. These are fit indices for CB-SEM, presented in (Hu and Bentler, 1999), rather than core indicators for assessing PLS-SEM.

Illustrative output: structural model and hypothesis testing

The numbers in the table only illustrate the shape of the output and the reporting format. You must not present them as findings from your own study. For each hypothesis, compare the p-value or confidence interval with the rule stated in your research design, and check whether the sign of the path coefficient matches the direction of the hypothesis.

Then add R² for endogenous variables, f² for each path, and Q² if your study uses this indicator. In PLS-SEM, R² values of 0.75, 0.50, and 0.25 are commonly interpreted as substantial, moderate, and weak, respectively (Hair et al., 2011). Q² above 0 indicates predictive relevance under the cross-validation logic (Stone, 1974).

Assessment thresholds

Do not copy every threshold you have seen on a website into the same table. State clearly whether the table assesses the measurement model or the structural model, because each indicator answers a different question. The benchmarks below are sourced reference points, not rules for mechanically deleting variables.

Do not use Alpha alone to decide whether to retain or remove a scale. Alpha from 0.7 or above is commonly accepted according to (Nunnally, 1978), while a value from 0.6 may be considered for a new or exploratory scale according to (Hair et al., 2010). Report the reasoning and also compare CR, AVE, outer loading, and the content of the indicators.

How to write this in your thesis

Write the results in the same order as the analysis. First describe the measurement model, followed by discriminant validity, collinearity, R², f², and finally hypothesis testing through bootstrapping. Each table needs a number, a title, and a source note reading “Results processed using SmartPLS 4” if the data are yours, followed by a paragraph interpreting the table.

How to write this in your thesis

You can adapt the following paragraph using your actual numbers: “The results show that the outer loadings of the indicators range from [lowest value] to [highest value]. The Composite Reliability values of the constructs range from [x] to [y], while AVE ranges from [x] to [y], indicating [conclusion about reliability and convergent validity] according to (Fornell and Larcker, 1981). HTMT between the constructs ranges from [x] to [y], below the threshold of [0.85 or 0.90], indicating [conclusion about discriminant validity] according to (Henseler et al., 2015).”

For hypotheses, use a more specific structure: “The path from [independent variable] to [dependent variable] has a path coefficient of [β], a t-value of [t], and a p-value of [p]. Because [p-value or confidence interval], hypothesis [H1] is [accepted or not supported].” Avoid writing “strong effect” if you have only looked at the p-value. Effect size, direction, and statistical significance are three separate pieces of information.

If you want to review the report in another format before creating the table, see how to run and read an HTML SmartPLS 4 report. Before submitting the thesis, check the construct names, decimal places, arrow symbols, and hypothesis codes across the tables, model figure, and interpretation.

Common mistakes

Mixing PLS-SEM and CB-SEM output

CFI, TLI, and RMSEA often appear in SEM materials using AMOS or CB-SEM. If you ran SmartPLS, focus on the PLS-SEM indicators suited to your research objective, such as outer loading, CR, AVE, HTMT, VIF, R², f², Q², and path coefficient.

Using screenshots instead of a summary table

A screenshot can preserve evidence, but it is not convenient for readers comparing constructs. Keep screenshots in your analysis records, while Chapter 4 should contain a properly aligned table with clear column names and a data source note.

Deleting an indicator only because its outer loading is low

A low indicator may reduce AVE or CR, but the decision to delete it must consider the scale content, the number of remaining indicators, and the theoretical model. Record every run, the indicator removed, and the reason. If you remove too many variables after viewing the output, discuss the decision with your supervisor rather than hiding the result.

Reading the p-value while ignoring the path coefficient

A small p-value only indicates statistical evidence under the rule you selected. The path coefficient shows the direction and relative size of the path in the model. These two columns should appear together in the hypothesis testing table.

Entering the wrong data or indicator names

If SmartPLS reports missing data, unusual loadings, or many invalid values, return to the .csv file. Check the header row, missing value coding, reverse-coded variables, and data type before running the model again. Do not edit numbers directly in the results table to make the report look better.

Read more: how to run and read an HTML SmartPLS 4 report, reading an HTML report in SmartPLS, exporting results to Excel in SmartPLS, SmartPLS, SmartPLS topic.

Frequently asked questions

Which tables should a SmartPLS results summary include

At minimum, you should include a measurement model group containing Outer Loadings, Construct Reliability and Validity, and Discriminant Validity. The structural model group usually contains VIF, R², f², and Bootstrapping results with path coefficient, t-value, and p-value.

You do not need to put every output into one table. Dividing the results into two or three tables makes Chapter 4 easier to read and reduces the risk of mixing indicators from the two stages.

Can SmartPLS results be exported directly to Excel

Yes. You can export results from SmartPLS 4 and then use Excel to filter, reorder columns, and combine the indicators you need to report. Keep the original export file separately, because the edited Excel table is no longer the original reference version.

Does SmartPLS 4 use CFI, TLI, and RMSEA

These indicators belong to the model fit assessment commonly used in CB-SEM. For a PLS-SEM summary table, you should not include them as mandatory thresholds. Report the indicators that fit the procedure, objective, and methodological sources selected for your study.

Should you delete a variable immediately when HTMT exceeds 0.90

You should not delete it immediately. A high HTMT value suggests that two constructs may not be sufficiently distinct, but you need to review the conceptual content, the wording of the indicators, the correlation matrix, and the theoretical basis. If you remove an indicator, record the criterion, the number of runs, and the effect on the model.

Can the numbers in the illustrative table be used in a thesis

No. The numbers in the illustrative table only show the column structure and interpretation format. You must replace them with results from your own .sav or .csv file, save the original output, and never report illustrative numbers as research findings.

Open your SmartPLS 4 project, run PLS-SEM and Bootstrapping again, and create separate measurement model and structural model tables before writing the interpretation. If you need to run the analysis on your own .sav or .csv file and compile the results into Chapter 4, see DoThesis M4 data analysis.