Modeling in SmartPLS

One of the techniques to analyze multivariate data is by using Structural Equation Modeling (SEM) Approach.

One of the popular SEM software is SmartPLS.

It is a graphical user interface for variance-based SEM using the partial least squares path modeling method. By using this software, the relationship between variables can be examined.

Basic Composition of PLS-SEM Model

LATENT AND MANIFEST VARIABLES

Latent Variable

The Variable that is not directly observed but are rather inferred (through a mathematical model) from other variables that are observed (directly measured).

This variable needs a manifest variable assigned to it as an indicator.

An outer model is also called measurement model.

It displays the relationships between the constructs and the indicator variables (rectangles).

Manifest Variables

Variable that can be directly measured or observed.

Example:

Latent Variablel/Construct: JIT

Manifest Variables/Indicators:

JIT1 Our company implements pull production system

JIT2 We use kanban to authorize production.

JIT3 We use kanban to authorize material movement.

JIT4 We produce a product only when it is requested by its user.

Latent Variablel/Construct: Performance

Manifest Variables/Indicators:

PERF1 Our average unit manufacturing costs have reduced

PERF2 Our overall quality of product has been outstanding

PERF3 Our overall flexibility performance has increased

PERF4 Our overall iventory performance has increased

OUTER AND INNER MODELS

Outer Model

An outer model is also called measurement model.

It displays the relationships between the constructs and the indicator variables (rectangles).

Inner Model

An inner model is also called structural model that represents the constructs (circles or ovals).

It also displays the relationships (paths) between the constructs.

ENDOGENEOUS & EXOGENEOUS MODELS

In SEM, the terms of independent and dependent variables are no longer used as suggested by Gefen et al., (2000) (Download here). As a replacement to those terms, SEM is using endogenous and exogeneous variables.

In some condition, a variable can be an endogenous in one situation, it can be also exogeneous in another situation.

Endogenous Variable

Constructs to be explained in the model.

Variable that is influenced/affected by other variables.

Variable that has arrow(s) pointing into it.

Exogenous Variable

Construct that explains other constructs in the model.

Variable that is not affected by other variables, rather it affects others

Variable that has arrow(s) pointing out from it.

The 2-Step Approach in SEM

This approach was suggested by Anderson and Gerbing (1988) (Download here).

This approach is also the same approach done in SPSS. However, if we use SPSS, we don not label them as measurement model and structural model assessment.

Step 1: Measurement Model assessment

To assess construct validity of the model (i.e., the relationships between latent variable & its manifest variables).

Step 2: Structural Model assessment

To test hypotheses (i.e., the relationships between latent variables).

The measurement model serves to create a structural model including paths representing the hypothesized associations among the research constructs.

A Systematic Procedure for Applying PLS-SEM

SmartPLS 2.0

SmartPLS 2.0.M3 has run out of support. But since its still very popular, this version is still provided for free. Please see here for details.

At the beginning of the year 2022, SmartPLS has now finally discontinued support for SmartPLS 2.0.M3, which we first released in 2005. SmartPLS will no longer offer SmartPLS 2.0.M3 for download and ship its activation keys.

You can easily transfer your old projects and continue using them in the much more advanced SmartPLS 3 and SmartPLS 4.

To produce identical results you have to check and apply the same settings in SmartPLS 2.0.M3, SmartPLS 3 or SmartPLS 4.

To give you enough time to migrate to SmartPLS 3 or SmartPLS 4, you can use the following final activation key that will allow you to use SmartPLS 2.0.M3 until the end of 2022.

D48AF1C13A2A6E28124089E838DA1A5EA260E4403F0AA71353A5E7C95416DBE4F236F4CC7B703DB019B6BFBCC510D95D6E420429DFE1385A343E409BC5D7DE6941E77828ACD071B2

SmartPLS 3.0

Brief Instruction of Using SmartPLS 3.0

Step 1 : Load SmartPLS 3 Software.

Step 2 : Create new project – assign name of project.

Step 3 : Double click to import data.

Step 4 : Draw model

Step 5 : Save model

Step 6 : Click on Calculate icon, select PLS-Algorithm on the drop-down list. Now, accept the default options, then click Finish.

Step 7 : Click on calculate icon, select Bootstrapping on the drop-down list. Now, accept the default options, then click Finish.

SmartPLS 4.0

Brief Instruction of Using SmartPLS 4

Step 1 : Load SmartPLS 4 Software.

Step 2 : Choose the workspace. You may create one folder allocated to save all SmartPLS Project (Only for the first time usage of SmartPLS 4).

Step 3 : Create new project – assign name of project.

Step 4 : Import data data file (i.e., either in *.csv, *.txt, *.sav, *.xls, or *.xlsx formats).

Step 4 : Create model

Step 5 : Save model

Step 6 : Click on Calculate icon, select PLS-Algorithm on the drop-down list. Now, accept the default options, then click Finish.

Step 7 : Click on calculate icon, select Bootstrapping on the drop-down list. Now, accept the default options, then click Finish.

Some highlights available in SmartPLS 4

a fundamentally renewed and optimized GUI
significantly improved performance
data import from Excel or SPSS in addition to CSV files
multiple moderation (e.g., three-way interactions)
testing endogenity with Gaussian copulas
Regression models, including many useful diagnostics and reports to fully perform the analyses
PROCESS type analysis, including bootstrapping of conditional direct and indirect effects
Necessary condition analysis (NCA) including significance testing
Accounting for scale type of variables in most algorithms
Standardized, unstandardized and mean-centered PLS-SEM analysis
New datafiles can be created from calulation results directly (useful for higher-order models)
Many new sample models are now included

interface features

Selecting a Workspace and Importing Project

Importing data and building a model

Let's give a try!

Instruction:

Download a data set HERE.
Draw the following model.

Google Sites

Report abuse