XLSTAT: Consumer satisfaction analysis in Excel with PLSPM in Excel

Background

Partial Least Squares Path Modeling (PLS-PM) is used to model complex relationships among observed and latent variables in structural equation models. It focuses on optimizing predictive accuracy rather than testing causal hypotheses.

PLS-PM emphasizes prediction over explanation, making it well-suited for exploratory analysis. Key references include Chin (1998) for applications and Tenenhaus et al. (2005) for theory. The latent variable scores from PLS-PM are useful for constructing global indices like consumer satisfaction.

 

PLS Path Modeling using XLSTAT-PLSPM

This article guides you step-by-step to create a project, define a model, estimate parameters, and analyze results for consumer satisfaction analysis.

Using real data from 250 mobile phone customers, the PLS-PM model is based on the European Customer Satisfaction Index (ECSI), where interrelated latent variables represent concepts that cannot be directly measured.

Each latent variable is linked to one or more measured (manifest) variables, scored from 0 to 100. For example, the latent variable Image is measured by five questions on trust, stability, social contribution, customer concern, and innovation.

The example uses an XLSTAT-PLSPM project file, which includes data and results. These projects are Excel templates with a *.ppmx extension. Save files using the “Save” or “Save As” options from the XLSTAT-PLSPM menu to ensure proper formatting.

On project creation, select Marketing mode for simplified dialogs and outputs. If the graphical display appears distorted, click “Optimize the display.”

Each project includes two essential sheets:

• D1: For input data.

• PLSPMGraph: For model design. Selecting this sheet activates the "Path modeling" menu.

You can switch display modes anytime via the PLSPM menu options.

In this example, we use the Marketing display for consumer satisfaction analysis. Save the project using the Save the project as function—always use PLSPM functions, not standard Excel ones, to save or open *.ppmx files.

Next, paste the data into the D1 sheet. Then switch to the PLSPMGraph sheet to start building the model. The toolbar appears in the upper left corner of that sheet.

To create several latent variables in a row, double click on the circle button so that it stays pressed while you add variables:

To add arrows between latent variables, click the starting variable, hold Ctrl, click the ending variable, then click the arrow button or press Ctrl+L.

You can also load predefined models using the seventh button, which accesses a library of well-known models. Selected models (.ppmxmod files) are automatically imported into the project.

After adding arrows, define the manifest variables for each latent variable (this can also be done earlier). To do this, select the latent variable and click the MV button in the toolbar.

This activates the D1 sheet and displays a dialog box where you give a proper name to the latent variable, select the manifest variables on D1 and define a few settings.

Next, define the mode of measurement. In Mode A (reflective), the latent variable influences the manifest variables; in Mode B (formative), the manifest variables define the latent variable. In this example, we use Mode A for all variables.

Below is the dialog box setup for the Expectation latent variable.

The obtained model has the following form:

Once the manifest variables have been defined for each latent variable and latent variables are linked, you can start computing the model. To run the model, click the play button.

This displays the Run dialog box, where many options are available. For this tutorial the following options have been used:

In our example, we use the automatic mode for the scale of the manifest variables and the 0-100 scale for the latent variables . Keep the default Options tab configuration unchanged. Configure the outputs tab as follows:

Activate the Simulation table option and set Satisfaction as the target variable to analyze the impact of changes in model variables. Changes range from -10% to 10% in 1% steps. Also enable the IPMA table (Importance-Performance Matrix Analysis).

Interpreting the Results

The first part of the results gathers information on the data and on the created model (descriptive statistics related to the manifest variables and model specification). It is first important to check blocs unidimensionality. In our case (reflective variables), blocs must be unidimensional. The Composite reliability table provides this information: The Cronbach alpha is below the 0.7 threshold for the Expectation and Loyalty variables. However, Dillon and Goldstein’s rho is above the 0.7 threshold for all variables. Finally, the first eigenvalue is higher than the second in many cases. Those results let us consider that blocs are unidimensional, although it could be interesting to investigate supplementary dimensions in expectation and loyalty. The Complaint latent variable does not appear in the table as it is associated to one single manifest variable.

The next table contains information on the goodness of fit index . The absolute GoF = 0.484. It is very close to its bootstrap estimation. This value is not easy to interpret. It is mostly used to compare different groups of individuals or different models. The relative GoF and GoFs based on internal and external models are very high and reflect a likely good fit of the model to the data.

The two following tables display external weights and the correlations associated to the measurement model. Once the measurement model is studied, the structural model can be analyzed. Each latent variable is associated to several outputs. We will comment the outputs of the satisfaction variable: We may consider that the latent variable is well explained (R²=0.681). We see that the perceived quality has the most important effect on satisfaction followed by the perceived value . The expectation impact is insignificant. The last table summarizes the preceding results. We see that the perceived quality has a 59% contribution to the R². The following chart illustrates the tables we just commented:

Next, the marketing-related outputs are displayed, starting with the IPMA (Importance-Performance Matrix Analysis). This chart shows the impact and performance of latent variables on the target variable, Satisfaction.

 

Quality is the most important and high-performing latent variable, leaving little room for improvement. Expectation and Image have similar importance, but Image stands out due to its lower performance, making it a good target for improvement.

Next, simulation charts show the effect of changes in manifest variables. The first chart summarizes how a X% increase in each manifest variable affects satisfaction. For a 10% increase, the results are:

  The QUALITY5 variable has the strongest effect. The following chart is the same as the current one but uses the mean value as a base. Then the changes to the latent variables are investigated:  

A 10% increase in quality leads to a 17% rise in satisfaction. A similar chart based on means follows.

The remaining outputs include latent variable scores and their descriptive statistics, which can be used for further analyses in XLSTAT.

Conclusion

This article showed how to investigate a dataset using a satisfaction model. We illustrated the use of the PLS Path Modeling approach lets you easily understand the interactions between concepts. Once the model is validated, it coefficient interpretation and result analysis are very simple.

Related Article

Partial Least Squares Path Modeling