See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/363536652

Introduction to SmartPLS and result reporting

Presentation · September 2022

CITATIONS READS

0 4,906

1 author:

Mona Ali
The German University in Cairo
17 PUBLICATIONS 6 CITATIONS

SEE PROFILE

All content following this page was uploaded by Mona Ali on 13 December 2022.

The user has requested enhancement of the downloaded file.

SmartPLS workshop
Mona Ali Ali
Under license from SmartPLSTM
Outline
• Ch.1 An introduction to Structural Equation Modelling
• Ch.2 Specifying the path model and examining data
• Ch.3 Assessing PLS-SEM results (Reflective measurement models)
• Ch.4 Reporting results (Reflective measurement models)
• Ch.5 Assessing PLS-SEM results (Formative measurement models)
• Ch.6 Mediator analysis
• Ch.7 Moderator analysis
 First a quick question!!
• What do you know about statistical analysis?????????????
• Normal distribution
• Parametric Vs. Non parametric data
• Factor analysis
• Exploratory Vs. Confirmatory
• Reliability
• Validity
• Regression analysis
• Multiple regression analysis
•Ch.1, An introduction to Structural Equation Modelling
• Application of first-generation statistical methods, such as factor
analysis and regression analysis, dominated the research landscape
through the 1980s.

• But since the early 1990s, second-generation methods have

expanded rapidly and, in some disciplines, represent almost 50% of
the statistical tools applied in empirical research (Hair et al., 2022)
Ch.1: What is SEM
• Structural equation modeling is a multivariate statistical analysis
technique that is used to analyze structural relationships. This
technique is the combination of factor analysis and multiple
regression analysis, and it is used to analyze the structural
relationship between measured variables and latent constructs.

• This method is preferred by the researcher because it estimates the

multiple and interrelated dependence in a single analysis.
 Ch.1: So, what is SEM???
• PLS-SEM is a “flexible” technique
capable of estimating complex
models (many constructs, many
variables, many causal relationships
between constructs – arrows – and
formative models),
Ch1: Why use SEM?
• The sample size is small
• The data is non-normally distributed.
• Test for normality https://stattrek.com/anova/normality/normality-test
• PLS enhances sampling distribution to approach normality
• For theory development and prediction
• Models can use fewer indicators (1 or 2)
• Model can use more indicators(up to 50+)
Ch1. : Different types of SEM (Chin and Newsted, 1999)
Ch.1: Assumptions
• Structural equation modeling is also called casual modeling because it
tests the proposed casual relationships. The following assumptions
are assumed:
Ch.1: Assumptions
• Multivariate normal distribution: The maximum likelihood method is used and assumed for
multivariate normal distribution. Small changes in multivariate normality can lead to a large
difference in the chi-square test.
• Linearity: A linear relationship is assumed between endogenous and exogenous variables.
• Outlier: Data should be free of outliers. Outliers affect the model significance.
• Sequence: There should be a cause and effect relationship between endogenous and exogenous
variables, and a cause has to occur before the event.
• Non-spurious relationship: Observed covariance must be true.
• Model identification: Equations must be greater than the estimated parameters or models
should be over identified or exact identified. Under identified models are not considered.
• Sample size: Most of the researchers prefer a 200 to 400 sample size with 10 to 15 indicators. As
a rule of thumb, that is 10 to 20 times as many cases as variables.
• Uncorrelated error terms: Error terms are assumed uncorrelated with other variable error terms.
• Data: Interval data is used.
Ch2. specifying the path model and data
• All main variables are called latent variables.
• All items reflecting or forming the latent variables are called
measures or indicators
• Arrows represent the relationships between latent variables and or
indicators
• Independent variables are called exogenous variable (arrows exit
from these variables)
• Dependent variables are called endogenous variables (Arrows are
drawn into these variables)
Ch2. Steps in SmartPLS analysis
• Differentiation between structural and measurement model

• Two main components of models are distinguished in SEM:

• The structural model showing potential causal dependencies between
endogenous and exogenous variables, and

• The measurement model showing the relations between latent variables and
their indicators.
Ch. 2 Structural Vs Measurement models
Go to SmartPLS
Draw model 1
Measures (Indicators)
Go to Dataset 1
• Make sure all data is clean
• https://www.formpl.us/blog/data-cleaning

• Make sure your data is in CSV form

Reflective measurement model
• A reflective measurement theory is based on the idea
that latent constructs cause the measured variables and the
error results in an inability to fully explain these measures.
• Reflective relationship with construct is called Loading
• For example, Customer Commitment is believed to cause
specific measured indicators like willingness to obtain brand X,
telling friends about purchasing brand X, and continuing to buy
brand X at higher price. Here Commitment can be expressed
through different ways. Even if one does not tell friends and
shows willingness to buy brand X at a higher price, this is still
termed Commitment.
Formative measurement model
• formative measurement theory is modeled based on the
assumption that measured variables cause the
construct. The error in the formative measurement models is
an inability to fulling explain the construct. This means that the
indicator list must be comprehensive.
• Formative relationship with construct is called Weight
• For example, the social class index, SCI is a composite of
one’s educational level, occupational prestige, and income. SCI
does not cause these indicators as in reflective case, but these
indicators cause the SCI. In case we remove Income as
indicator, we cannot call it SCI.
Reflective Vs. Formative model (Comparison)
Coltman et al. (2008)
Reflective Vs. Formative model (Example)
(Dr. Emmanuel Awuni)
More Formative Vs. Reflective examples
Reflective Vs. Formative presentation in
SmartPLS
Formative Vs Reflective
• It is important to note that by default, SmartPLS assumes the
indicators are reflective when the model is built, with arrows pointing
away from the blue-color latent variable. One of the common
mistakes that researchers made when using SmartPLS is that they
forget to change the direction of the arrows when the indicators are
“formative” instead of “reflective”.
Formative Vs Reflective
• If the indicators cause the latent variable and are not
interchangeable among themselves, they are formative. In general,
these formative indicators can have positive, negative, or even no
correlations among each other (Haenlein & Kaplan, 2004; Petter et
al., 2007).
• As such, there is no need to report indicator reliability, internal
consistency reliability, and discriminant validity if a formative
measurement scale is used. This is because outer loadings, composite
reliability, and square root of average variance extracted (AVE) are
meaningless for a latent variable made up of uncorrelated measures.
Go to Model 1
• Click on any latent variable
• Right click
• Choose invert measurement model
• Change it back to reflective again!
Ch. 3 Analysis of
SEM (Parameters)
Reflective (Parameters)
Inner Model (Parameters)
Ch.3 Assessing PLS-SEM reflective models.
1- Go to Model 1
2- Choose  Calculate  PLS algorithm
3- Enjoy!
4- Report
But how do I report results????
Ch.5 Reporting results of reflective model
assessment
• 1- First report the sample description (Excel is easier)
Ch. 5 Reporting results of reflective model
assessment
• 2- Report the outer model – reliability and convergent validity
(Numbers may differ because the output is from SmartPLS 3.0)
Ch. 5 Reporting results of reflective model
assessment
• 2- Report the outer model – discriminant validity
• (Numbers may differ because the output is from SmartPLS 3.0)
Ch. 5 Reporting results of reflective model assessment
• 2- Report the outer
model – Loadings,
Threshold: Loading>0.7

• (Numbers may differ

because the output is
from SmartPLS 3.0)
Ch. 5 Reporting results of reflective model assessment
• 2- Report the outer model – HTMT (Discriminant validity), Threshold: HTMT<0.85

• (Numbers may differ because the output is from SmartPLS 3.0)

Ch. 5 Reporting results of reflective model
assessment
• 3- Report the inner model – Run the bootstrapping function first!!
• (Numbers may differ because the output is from SmartPLS 3.0)
T- Value (Same as Z statistic)
• Use the following values
Two Tail
90% significance level--t value>1.645
95% significance level--t value>1.96
99% significance level--t value>2.57

One tail
90% significance level--t value>1.28
95% significance level--t value>1.645
99% significance level--t value>21.96
Final Model
Interpretation:
• The tool is both reliable and valid
• Hypothesis testing: Significant relationships could be observed
between………………………
• R squared:
• Firm Performance: 37.3 percent of the variance in Firm Performance is due to the
Variance in both Hospitableness and Hospitality
• Hospitality: 51.5 percent of the variance in hospitality is attributed to the variance in
hospitableness.
• Why doesn’t Hospitableness have an R2 ???
• Beta (Path coefficient)
• Hospitality has a slightly higher effect on Performance than Hospitableness.
• Both have low to medium positive effect. (1 highest and 0 lowest)
• F Squared (Difference in R2 if dropped
• Hospitality contributes more to the R squared than does Hospitableness
• Following Cohen (1988), .02 represents a “small” f2 effect size, .15 represents a
“medium” effect, and .35 represents a “high” effect size. We can say that the effect
of dropping both latent V from the model is Low.
Ch. 6 Mediator analysis

• Mediation considers the presence of an intermediate variable or

mechanism that transmits the effect of an independent variable to an
outcome (Aguinis et al. 2017).
• For instance, mediation usually appears when the effect of reputation
on customer loyalty is transmitted by customer satisfaction, such that
reputation impact on customer satisfaction in turn influences
customer loyalty
Mediation Effect
• The following figure shows the example of a simple mediator model,
whereby p3​ is the direct effect, p1​⋅p2​ is the indirect effect, and the
direct effect (p3​) + the indirect effect (p1​⋅p2​) = the total effect:
To analyze a mediator, Zhao et al. (2010) suggest a model, as shown in the following
figure, which Hair et al. (2017) also propose to use for PLS-SEM:
Steps for Mediation analysis in SmartPLS:
1- Draw the Model
Go to Model 2
Steps for Mediation analysis in SmartPLS:

2- Upload Data Set 2 (Driver Behavior)

The purpose is to see how intention mediates the relationship between

the predictors (attitude, subjective norms and perceived behavior
control) and driver behavior.

H  Intentions mediate the relationship between At, SN, PBC and

driver behavior.
 TPB Driver
Behavior
final model
Steps for Mediation analysis in SmartPLS:
• First we will run the measurement model using the PLS algorithm.
Click on Calculate and PLS Algorithm.
• Click on Start Calculation.
• We will assess the loadings > 0.708, before we move to testing the
AVE and CR.
• Delete all indicators with low loadings.
• We will assess the CR > 0.7 and AVE > 0.5 to confirm convergent
validity before we move to testing the mediating effect.
• We will then run the bootstrapping to get the t-values to assess if the
direct relationships are significant before testing the mediating
effects.
• Click at Calculate and then click on Bootstrapping.
Steps for Mediation analysis in SmartPLS:
• Make sure the researchers click on Two-Tailed, and continue clicking
on Start Calculation.
Steps for Mediation analysis in SmartPLS:
• Then, Click on the Total Indirect Effect Result.
Mediator results:
• Based on the statistics we can conclude that
• The only significant mediating effect is between attitudes and
behavior.
• Significance @ t-value >1.96 and p-value < 0.05

Original sample (O) Sample mean (M) Standard deviation (STDEV) T statistics (|O/STDEV|) P values

Attitudes -> Behavior -0.312 -0.316 0.042 7.508 0

Perceived behavioral
control -> Behavior 0.037 0.037 0.043 0.863 0.388

Subjective norms ->

Behavior -0.061 -0.067 0.045 1.358 0.174
Ch. 7 Moderator analysis
• A moderator variable can be visualized as a third variable that
changes the relationship between the independent variable and the
dependent variable.
Ch.7 Moderation analysis
• There are three approached for moderation testing in SmartPLS 3. You should choose
according to your research settings and objectives:
• If moderator (or predictor) is formatively measured, choose Two Stage approach.
• If moderator (and predictor) are both reflectively measured, you may choose any of the three
approaches depending on your research objective:
• Two-stage has the highest power, i.e. it is the most likely approach to detect a significant interaction.
• Othogonalizing has the smallest bias, i.e., the interaction effect size is most correct and it
maximizes the explained variance in your dependent variable.
• Product-Indicator approach has no particular advantages.
• If your sample size is small, you probably want to focus on statistical power and hence should
choose the Two-Stage approach.

• In general, the difference for reflective constructs is relatively small between the
approaches. However, if you have a formative construct and choose the
orthogonalization or product-indicator approach, you could get very misleading results.
SmartPLS 4: testing Moderation
1- Go Back to Model 1
2- Form Hypothesis
H1  Gender moderates the positive effect of Hospitality on
Firm Performance
Or Even Better!!!
H1  The positive effect of Hospitality on Firm Performance is
stronger for Females than for Males
Remember even for moderators, Hypothesis are based on strong
theoretical background!!!!!!
SmartPLS 4.0 testing Moderation
• Drag the Moderator indicators near the model
• Connect moderator to endogenous variable
• Connect moderator to the line representing the relationship it
moderates, (New to SmartPLS 4.0)
• Run the bootstrapping for significance
• Run the PLS algorithm
• Check the Simple slope analysis for significant relationships
• Below is -1 standard deviation
• Above is +1 standard deviation of moderator.
Moderator analysis- SmartPLS 4
Step 1- Add the moderators LV to the endogenous variable.
Moderator analysis- SmartPLS 4
Step 2- Connect the moderators to the line representing the
relationships it moderates
Moderator analysis- SmartPLS 4.0
• Run the Bootstrapping and check the significance
Bootsrapping output
Hypothesis is not approved (not significant)

Standard deviation T statistics

Original sample (O) Sample mean (M) (STDEV) (|O/STDEV|) P values
AlphaHospitality -> Firm Performance -0.034 -0.019 0.397 0.087 0.931
EDUCATION -> Firm Performance 0.027 0.025 0.084 0.319 0.75
GENDER -> Firm Performance -0.594 -0.616 0.299 1.988 0.047
Hospitableness -> AlphaHospitality 0.73 0.731 0.038 19.245 0
Hospitableness -> Firm Performance 0.225 0.219 0.366 0.616 0.538
GENDER x Hospitableness -> Firm
Performance 0.047 0.051 0.227 0.206 0.837
GENDER x AlphaHospitality -> Firm
Performance 0.256 0.253 0.23 1.11 0.267
EDUCATION x Hospitableness -> Firm
Performance 0.008 0.007 0.062 0.135 0.893
EDUCATION x AlphaHospitality -> Firm
Performance -0.043 -0.041 0.056 0.77 0.441
Bootsrapping output
Hypothesis is not approved (not significant)
Run the PLS algorithm and check the simple slope
analysis for interpretations (If significance is found)
- No moderation is found (Hospitableness and
Performance with Gender as a moderator)
Very interesting!!!!
The higher the education
the more negative the
relationship between
Hospitality and
Performance. But with
lower education levels
the relationships shifts to
a positive relationship !
Gender as a moderator of the relationship
between Hospitality and performance
Education as a
moderator of the
relationship between
hospitableness and
performance. The
more the education
the more positive the
relationship is. Less
education leads to
less positive
relationship
Triangulation with other methodologies:
(Richter et al. 2022)
 References
• Joseph F. Hair, Tomas M. Hult, Christian M. Ringle, Marko Sarstedt. A Primer On Partial Least Squares Structural Equation Modeling (PLS-SEM), Sage January 2022.

• Nicole F. Richter, Sven Hauff, Christian M. Ringle, Siegfried P. Gudergan. The Use Of Partial Least Squares Structural Equation Modeling And Complementary Methods In International Management
Research, Management International Review, Springer 2022.

• Coltman, T., Devinney, T. M., Midgley, D. F., & Venaik, S. (2008). Formative Versus Reflective Measurement Models: Two Applications Of Formative Measurement. Journal Of Business
Research, 61(12), 1250-1262.

• Aguinis, H., Edwards, J. R., & Bradley, K. J. (2017). Improving Our Understanding Of Moderation And Mediation In Strategic Management Research. Organizational Research Methods

• Nitzl, C., Roldan, J. L., & Cepeda-Carrion, G. (2016). Mediation analysis in partial least squares path modeling: Helping researchers discuss more sophisticated models. Industrial Management &
Data Systems, 116(9), 1849–1864.

• Chin, W. W., & Newsted, P. R. (1999). Structural equation modeling analysis with small samples using partial least squares. In R. H. Hoyle (Ed.), Statistical strategies for small sample research (pp.
307-341). Thousand Oaks: CA: Sage Publications.

• Henseler, Jörg and Wynne W. Chin (2010). A Comparison of Approaches for the Analysis of Interaction Effects Between Latent Variables Using Partial Least Squares Path Modeling, Structural
Equation Modeling, 17 (1), 82-109.

View publication stats