You are on page 1of 29

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

net/publication/341250748

PLS-SEM using SmartPLS 3.0: Chapter 12: Assessment of Mediation Analysis

Chapter · July 2017

CITATION READS
1 15,972

5 authors, including:

T. Ramayah Cheah Jun Hwa


Universiti Sains Malaysia University of East Anglia
742 PUBLICATIONS   27,160 CITATIONS    121 PUBLICATIONS   7,290 CITATIONS   

SEE PROFILE SEE PROFILE

Francis Chuah Hiram Ting


Universiti Utara Malaysia UCSI University
58 PUBLICATIONS   2,148 CITATIONS    175 PUBLICATIONS   5,860 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Parenting Style and Academic Performance View project

Comparison between Celebrity Endorsement and Selfie Promotion View project

All content following this page was uploaded by Hiram Ting on 08 May 2020.

The user has requested enhancement of the downloaded file.


How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

Partial Least Squares


Structural Equation Modeling
(PLS-SEM) using SmartPLS 3.0
An Updated and Practical Guide to Statistical Analysis

• BASIC LEVEL •SECOND EDITION •

Only Chapter Twelve Available:


Assessment of Mediation Analysis

Edited by

T. Ramayah
Jacky Cheah, Francis Chuah
Hiram Ting & Mumtaz Ali Memon

July, 2017

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

Partial Least Squares Structural Equation Modeling (PLS-SEM) using SmartPLS


3.0: An Updated and Practical Guide to Statistical Analysis
(Basic Level; Second Edition)

Table of Contents

Chapter 1 Development of Structural Equation Modeling


Chapter 2 Partial Least Squares Structural Equation Modeling (PLS-SEM)
Chapter 3 Reflection on SEM Application
Chapter 4 When should I use PLS-SEM instead of CB-SEM?
Chapter 5 Using the SmartPLS Program
Chapter 6 Understand the Measurement Model Assessment between EFA and CFA
Chapter 7 Assessment of Reflective Measurement Models
Chapter 8 Assessment of Formative Measurement Models
Chapter 9 Assessment of Goodness-of-Fit (GOF)
Chapter 10 Assessment of Structural Model
Chapter 11 Assessment of Importance and Performance Matrix Analysis (IPMA)
Chapter 12 Assessment of Mediation Analysis
Chapter 13 Assessment of Moderation Analysis

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

Chapter Twelve
Assessment of Mediation Analysis

Having learned how to conduct basic analysis using SmartPLS 3.0, we now focus on the
assessment of mediation analysis. This chapter begins by providing a basic understanding of
mediation analysis. This includes the different approaches of mediation analysis and their
limitations. The current developments in mediation analysis are then explained. Next, a step-by-
step assessment of mediation effect is explained using SmartPLS 3.0. The chapter concludes by
addressing several frequently asked questions, and provides guidelines for research students,
academics and reviewers, in order to enhance their understanding of mediation analysis.

12.1 Mediation

Mediation models have witnessed considerable adoption among social science researchers. A
mediating variable is, “The generative mechanism through which the focal independent variable is
able to influence the dependent variable of interest . . . (and) Mediation . . . is best done in the case
of a strong relation between the predictor and criterion variable” (Baron and Kenny, 1986, p. 1173,
1178). According to Venkatraman (1989), mediation is the, “… existence of a significant
intervening mechanism between antecedent and the consequent variables.” (p. 428). In other
words, a mediator specifies how, or the mechanism by which, a given effect occurs. Mediation is
also known as a special case of “indirect effect”, where it depends on a strong priori theoretical/
conceptual support, which is essential to explore meaningful mediation effects (Hair et al., 2017;
Preacher and Hayes, 2012). When the support is present, mediation can be a useful statistical
analysis, if carried out properly.

12.2 Basic understanding on mediation effects

Let a, b, c, and c’ be quantifications of causal effects, such as regression coefficients in an


Ordinary Lease Squares (OLS) model, or the maximum likelihood path estimates in a structural
equation model. For this discussion, we assume M and Y are continuous, or modelled as in Figure
12.1a and Figure 12.1b. Consider a variable that is assumed to cause another variable Y (see Figure
12.1a). The variable X is called the causal variable and the Y is called the outcome (the unmediated
model). Path c in the above model is called the total effect.

Figure 12.1a: Total Effect

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

Figure 12.1b: Simple Mediation Model

The effect of X on Y may be mediated by a process, or mediating variable M, and the variable X
may still affect Y (see Figure 12.1b). Path c' is called the direct effect of X on Y when M is held
constant, and Path a*b is called the “indirect effect,” of X on Y. The total effect of Figure 12.4 is
the direct effect + indirect effect [c = c’ + (a x b)]. Note also that the indirect effect equals the
reduction of the effect of the causal variable on the outcome, or ab = c - c' (indirect effect = total
effect - direct effect). When estimating the effects in a mediation model, a*b is the measure of
the amount of mediation.

12.3 Testing for mediation effects: The Baron and Kenny’s (1986) Method

The Baron and Kenny’s (1986) causal procedure method has been the highly employed
procedure for the testing of mediation effect in social science research. This is based on the
original guidelines provided by Judd and Kenny (1981). This causal procedure includes several
steps in order to confirm a mediation effect, as explained in Figure 12.2 (see Figure 12.2a, 12.2b,
12.2c and 12.2d):

i. Figure 12.2a: The IV significantly affects the DV in the absence of the mediator

ii. Figure 12.2b: The IV significantly affects the mediator

iii. Figure 12.2c: The mediator has a significant unique effect on the DV, and

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

iv. Figure 12.2d: The effect of the IV on the DV shrinks upon the addition of the mediator to
the model.

Figure 12.2 Baron and Kenny’s approach Method of Mediation Analysis

This procedure remains popular in spite of its problems. This popularity is no doubt due to the
fact that it is quite simple to understand, easy to describe and teach to anyone. Moreover, it can
be summarized in a few sentences in a scientific report, no specialized software required, and it
can be implemented without taking a strong background in statistics or data analysis. Therefore,
many researchers today are still being taught, and recommended to use the Baron and Kenny
method for mediation analysis. However, despite its popularity, recent developments in
quantitative research methods suggest several limitations in Baron and Kenny’s causal
procedure approach.

12.4 Limitations of Baron and Kenny’s Method of Mediation Testing

There are several shortcomings in using the Baron and Kenny’s method. For example, under this
method, all four steps must be significant before mediation can be accepted. It means one has
to stop analysis if any of the above steps becomes insignificant, and, “no mediation” will be
concluded. To illustrate this, if, the relationship is not significant in step one, then the mediation
test has to be abandoned. If researchers pass the first step but not the second step, then they
will have to stop. Likewise, the relationship is significant in the first two steps but not in the third
step, it shall be concluded that there is not mediation. Once all three steps are significant, then
only the fourth step can be done to assess the reduction in the beta of the independent variable.
This method has been criticized as having very low power, and the multiple steps increase the
occurrence of Type I error, i.e.; a false conclusion that there is mediation when, in fact, there is
no mediation effect (Rungtusanatham et al. 2014). Additionally, the Baron and Kenny’s method
does not measure the magnitude of the mediation effect, nor does it accommodate models with
inconsistent mediation (MacKinnon et al., 2000; Preacher and Kelly, 2011; Rungtusanatham et al.
2014).

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

12.5 Sobel Test and its Limitations

The other method that has been used widely in the “20th-century” is the Sobel test approach,
which is available at: http://quantpsy.org/sobel/sobel.htm (online calculator) (Refer to Figure
12.3a to 12.3c).

Figure 12.3a: Aroian and Goodman test

Figure 12.3b: Alternative test

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

Figure 12.3c: Sobel Test Formula

For the Sobel test, the p-value is derived by assuming normality of the sampling distribution of
the indirect effect and using the standard normal distribution. Although this assumption is fairly
sensible in large samples, it is not in smaller ones. What is a sufficiently large sample is
situationally-specific, and typically you would not know whether to trust a large sample theory
when going into the analysis. In other words, Sobel test is not appropriate to be used in study
because the distributional assumptions do not hold for the indirect effect especially when
looking into a*b (Indirect Effect achieving Leptokurtic Distribution) (Refer to Figure 12.4).

Figure 12.4: Distribution of Mediation Effect

This assumption, which typically will not hold, yields a test that is lower in power than
alternatives especially if researchers use on small sample sizes or when they require
unstandardized path coefficients to calculate the result of Sobel Test. Overall, experts like Hair
et al., (2017) and Preacher and Hayes (2012) in mediation analysis do not recommend the use of
this test, though it remains popular.

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

12.6 Current Directions in Mediation Analysis

Preacher and Hayes (2004; 2008), and Hayes (2009), have criticized the “causal procedure” of
Baron and Kenny. Notwithstanding a single inferential test of the indirect effect is all that is
needed. Several scholars appear to recommend that direct effect does not have to be significant
while analysing for mediation (Shrout & Bolger, 2002; Zhao, Lynch & Chen, 2010). This is because
a significant direct relationship may not be identified because of a small sample size or other
extraneous factors (e.g., moderation), or there may not be enough power to predict the effect
that actually exists. Therefore, what matters in mediation analysis is indirect effect (Hayes &
Rockwood, 2016). According to Preacher and Hayes (2004; 2008), the mediation method is called
“bootstrapping the indirect effect.” (Refer to Figure 12.5).

Figure 12.5: The Path Coefficient in Mediation

Preacher and Hayes (2004; 2008), have argued that the path coefficient, a, and path coefficient,
b, can be normally distributed but the product, i.e.; a*b will not be normally distributed (see
Figure 12.4). As such, the Sobel test online, which is based on normal distribution, cannot be
utilized as it will lead to a wrong conclusion. Since the indirect effect is not normally distributed,
this will affect the “standard errors.” Thus, to correct this situation, a “bootstrapping,”
procedure needs to be put into place.

Bootstrapping, a nonparametric resampling procedure, has been recognized as one of the more
rigorous and powerful methods for testing the mediating effect (Hayes, 2009; Shrout and
Bolger, 2002; Zhao et al., 2010). There are several types of bootstrapping researchers can use to
perform the mediation test, namely; Percentile Bootstrap, Standardized Bootstrap, Bias-
Corrected and Accelerated (BCa) Bootstrap, Davidson and Hinkley’s Double Bootstrap, and Shi’s
Double Bootstrap. Informed advice is to use the default button of Bias-Corrected and
Accelerated (BCa) Bootstrap. Hayes and Scharkow (2013) show that the bias-corrected
bootstrap confidence interval is the best approach for detecting mediating effects when a
mediation is present. Alternatively, the percentile bootstrap confidence interval, that is not bias-
corrected, is a good compromise for researchers if Type-I errors are the concern (Hayes &
Scharkow, 2013).

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

The application of bootstrapping for mediation analysis has recently been advocated by Hair et
al., (2013), who have noted that, “… when testing mediating effects, researchers should rather
follow Preacher and Hayes (2004, 2008), and bootstrap the sampling distribution of the indirect
effect, which works for simple and multiple mediator models.” (P. 223). Furthermore, this
method is said to be perfectly suited for PLS-SEM because it makes no assumption about the
shape of the variables’ distribution, or the sampling distribution of the statistic, and therefore
can be applied to small sample sizes (Hair et al., 2017; Preacher & Hayes, 2008).

12.7 Steps on How to Assess Basic Mediation Model in SmartPLS 3.0

To illustrate how to perform on mediation analysis, this chapter will be using a Theory of Planned
Behaviour (TPB) PLS (n=375) data file - available on the website with this book. The purpose is to
see how intention mediates the relationship between the predictors (attitude, subjective norms
and perceived behaviour control) and share. Based on Figure 12.6, there are three exogenous
constructs: Attitude (Att1 to Att5), Subjective Norm (Sn1 to Sn4), and Perceived Behavioral
Control (Pbc1 to Pbc4). As for the mediation construct, Intention has five indicators (Intent1 to
Intent5) and the endogenous construct has three indicators (Share 1 to Share 3).

Figure 12.6: TPB model

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

Step 1: First we will run the measurement model using the PLS algorithm. Click on Calculate and
PLS Algorithm.

Step 2: Click on Start Calculation.

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

Step 3: We will assess the loadings > 0.708, before we move to testing the AVE and CR.

All the indicators show that the loadings are more than 0.708.

Step 4: We will assess the CR > 0.7 and AVE > 0.5 to confirm convergent validity before we move
to testing the mediating effect.

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

The result exhibits that the CR’s are all more than 0.7, and the AVE’s are all more than 0.5. As
such convergent validity can be confirmed. Once this has been done we will move on to test the
mediating effect through the use of the bootstrapping technique (refer to step four). But first,
let us write down the hypotheses for the three mediating effects. The three mediation
hypotheses can be written as follows:

H1: The relationship between attitude and sharing will be mediated by intention.
H2: The relationship between subjective norm and sharing will be mediated by intention.
H3: The relationship between perceived behavioural control and sharing will be mediated by
intention.

Step 5: 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.

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

Step 6: Make sure the researchers click on Two-Tailed, and continue clicking on Start
Calculation.

Most important! The


researcher needs to click on
Bias-Corrected and
Accelerated (BCa)
Bootstrap, and Two-Tailed

Note: For mediation analysis, the result must be in Two-Tailed because of the way we highlight
our hypothesis (i.e., intention mediates the relationship between attitude and share).

Step 7: The researchers will have the t-value result of the TPB’s structural model.

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

Step 8: Click on the Bootstrapping Result.

Click on the
Bootstrapping

Step 9: Then, Click on the Indirect Effect Result.

Click on the Indirect Effect


Result to get the result of a*b
(the mediation result)

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
Original Sample Standard
T Statistics
Sample Mean Deviation P Values
(|O/STDEV|)
(O) (M) (STDEV)
Attitude -> Intention
Attitude -> Share 0.489 0.497 0.068 7.153** 0.000
Intention -> Share
Perceived Behavioral Control -> Intention
Perceived Behavioral Control -> Share 0.136 0.134 0.031 4.379** 0.000
Subjective Norm -> Intention
Subjective Norm -> Share 0.201 0.197 0.068 2.944** 0.003
Note: t-value >1.96

Based on the result, we can conclude that all three mediations are significant at t-values >1.96
and p-value <0.05. Next, we also need to calculate the 95% bootstrapped confidence interval bias
using the following formula:

Lower Limit (LL) a*b – z (SE) (z value, for 0.05 level is 1.96)
Upper Limit (UL) a*b + z (SE)

Step 10: To obtain further results on the mediation analysis, the researchers can click on the
Confidence Intervals Bias.

Click on the Confidence


Intervals Bias to have the Lower
Limit (LL) and Upper Limit
(UL) of the mediation result

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
Original Sample
Path Bias 2.5% 97.5%
Sample (O) Mean (M)
Attitude -> Intention
Attitude -> Share 0.489 0.497 0.008 0.335 0.611
Intention -> Share
Perceived Behavioral Control -> Intention
Perceived Behavioral Control -> Share 0.136 0.134 -0.002 0.076 0.193
Subjective Norm -> Intention
Subjective Norm -> Share 0.201 0.197 -0.004 0.075 0.337

12.8 Reporting Mediation Analysis

The bootstrapping analysis has shown that all three indirect effects, β = 0.489, β = 0.201 and β =
0.136, are significant with t-values of 7.153, 2.944 and 4.379. The indirect effects 95% Boot CI Bias
Corrected: [LL = 0.335, UL = 0.611], [LL = 0.075, UL = 0.337], and [LL = 0.076, UL = 0.193], do not
straddle a 0 in between indicating there is mediation (Preacher and Hayes, 2004, 2008). Thus,
we can conclude that the mediation effects are statistically significant. The results of mediation
analysis are presented in Table 12.1.

Table 12.1: Hypothesis Testing on Mediation


No Relationship Std. Std. t-value Confidence
Beta Error Interval (BC) Decision
LL UL
H1 Attitude->Intention->Share 0.489 0.068 7.153** 0.611 0.335 Supported
H2 Subjective Norm->Intention->Share 0.201 0.068 2.944** 0.075 0.337 Supported
H3 Perceived Control->Intention->Share 0.136 0.031 4.379** 0.076 0.193 Supported
Note: *p< 0.05, **p< 0.01, BC = Bias Corrected, UL = Upper Level, LL = Lower Level

12.9 A Common Misconception

A common misconception among early career researchers is that when they see a third variable
in between the independent and dependent variable (as shown in Figure 12.7a), they consider it
a mediating model. Notwithstanding, drawing an additional box between the independent and
dependent variables does not automatically make it a mediating model. For instance, if
researchers intend to examine the effect of X on M, and the effect of M on Y, this is simply a step-
wise analysis of a model, and does not conceptualize mediating effect. Mediation depends on a
strong a priori theoretical/ conceptual support to explore meaningful mediation effects (Hair et
al., 2017). According to McKinnon et al. (2012), “the conceptualization of a mediation analysis
requires forethought about the relationships between the variables of interest and the
theoretical meaning behind those relationships” (p. 5). As such, the conceptualization, and a
hypothesis related to mediation, must be included explicitly, as shown in Figure 12.7b. Note,
theory is always the foundation of empirical analyses.

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

Figure 12.7: Mediation Model

12.10 Frequently Asked Questions about Mediation


1. How many hypotheses should I develop and test for mediation analysis?
We receive many questions from postgraduate research students inquiring about the number of
hypotheses for a mediation effect. Rungtusanatham et al. (2014), recommends two approaches
for theorizing mediation effect: (i) Segmentation, and (ii) Transmittal. Researchers following a
segmentation approach, should consider at least three hypotheses; 1) X on M, 2) M on Y, and 3)
M mediates the effect of X and Y, as shown in Figure 12.8.

Figure 12.8: Segmentation Approach for Theorizing Mediation

Moreover, a transmittal approach focuses mainly on indirect effect. As explained by


Rungtusanatham et al. (2014), in a transmittal approach, researchers should “… develop the
hypothesis that M mediates the effect of X on Y, or that X has an indirect effect on Y through M,
without needing to articulate hypotheses relating X to M and M to Y.” (p. 106). Figure 12.9
explains a transmittal approach for mediation analysis.

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

Figure 12.9: Transmittal approach for Theorizing Mediation


Table 12.2: Hypotheses for Theorizing Mediation

Segmentation Approach for Theorizing Mediation

H1 The effect of X on M
Example: There is a positive effect of X on M.
H2 The effect of M on Y
Example: There is a positive effect of M on Y.
H3 The effect of M on Y
Example 1: M mediates the relationship between X and Y, OR
Example 2: X has an indirect effect on Y through M.

Transmittal Approach for Theorizing Mediation

H1 The effect of M on Y
Example 1: M mediates the relationship between X and Y, OR
Example 2: X has an indirect effect on Y through M.

2: Should I include (test) the direct effect between X and Y?


Many researchers inquire whether it is a must to include the direct effect of X on Y when testing for
mediation effect. According to Rungtusanatham et al. (2014), “With either approach
(segmentation and transmittal), hypothesize the direct effect of X on Y when other mediators,
besides M1 (when analysing multiple mediators), may theoretically or empirically explain the
total effect of X on Y, or when the direct effect of X on Y may have a sign that is opposite of the
sign for a specific indirect effect.” (p. 107).
Apart from the aforementioned recommendations, researchers must consider several other
aspects when deciding whether they test the direct effect between X and Y while testing
mediation effect. First, the theoretical consideration - whether underpinning theory suggests
testing the relationship between X and Y? Second, it depends on the analytical approach. For
example, if a researcher is following the Baron and Kenny’s (1986) method, it is a must to test
the direct relationship between X and Y. For those following Preacher and Hayes (2004, 2008),
the focus should be mainly on an indirect effect. Lastly, the problem statement related to
mediation plays an important role. If a researcher argues that X affects Y because of M, then
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

testing direct relationship is of less importance. On the other hand, one might conceptualize that
there is a well-established relationship between X and Y, and the researcher wants to explore
whether M can mediate the relationship. In such a case, testing the direct effect of X and Y is
important to confirm whether there are any additional conclusions while controlling mediator.

3: Do I need to explain the type of mediation (full mediation or partial mediation)?


Researchers, especially postgraduate students, are often unclear whether there is a need to
further determine the types of mediation (i.e., Full Mediation and Partial Mediation). The partial
or full mediation concept is rooted in the Baron and Kenny’s method. Full mediation (also called
complete mediation) is assumed when the effect of the variable X to Y is completely transmitted
with the help of another variable M. It also means the condition Y completely absorbs the
positive or negative effect of X. On the other hand, the situations under the condition that both
the direct effect c’ and the indirect effect a*b are significant, represent partial mediation.

For instance, if the path coefficient in Step one (Figure 12.2a) is 0.485 significant, and then in
Step four (figure 12.2d) becomes 0.025 and not significant, we will conclude, “full mediation.” It
implies that the mediator has absorbed all the direct effect of the IV. However, if the path
coefficient in Step one is 0.485 significant, and then in Step four, the same beta now becomes
0.255, but is still significant, we will conclude, “partial mediation,” indicating that the mediator
has absorbed only part of the direct effect of the IV. Although these terms are used in abundance
in the scientific literature, and are frequently the subject of hypotheses being tested, current
developments in mediation analysis suggest, “… avoid the nomenclature of ‘complete
mediation’ or ‘partial mediation’ when hypothesizing mediation processes” (Hayes, 2012;
Rungtusanatham et al. 2014, p. 106).

However, recent PLS-SEM literature (see Hair et al., 2017; Nitzl, Roldan, and Cepeda, 2016)
suggests the extraction of more information of mediation effect via looking into the types of
mediation would be useful for any studies. Similarly, MacKinnon et al. (2007) highlighted that the
types of mediation effects (Full mediation and Partial Mediation of Complementary and
Competitive) have the potential to deliver theoretically interesting findings (see Figure 12.10).
According to Nitzl, Roldan, and Cepeda (2016), a mediating effect must first exist when the
indirect effect, a × b, is significant. Next, the researcher can advance the discussion of the two
different types of mediation, full and partial mediation. In addition, partial mediation can be
divided again into complementary and competitive partial mediation.

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

Figure 12.10: Mediator analysis procedure suggested by Zhao et al. (2010) and Nitzl et al. (2016)

Complementary Partial Mediation


In a complementary partial mediation, the direct effect c’ and indirect effect a*b point in the
same (positive or negative) direction (Baron and Kenny, 1986; see Figure 12.10 and Figure 12.11).
It is an often-observed result that a*b and c’ are significant, and a*b × c’ is positive, which
indicates that a portion of the effect of X on Y is mediated through M, whereas X still explains a
portion of Y that is independent of M. This complementary mediation hypothesis suggests that
the intermediate variable explains, possibly confounds or falsifies, the relationships between the
independent and dependent variables. Complementary partial mediation is often called a
“positive confounding” or a “consistent” model (Zhao et al., 2010). For example, Nitzl and Hirsch
(2016) showed, in addition to the abovementioned full mediating effect, that 30 percent of the
trust disposition (X) of a superior is mediated through the organizational (M) setting. Thus, the
superior with a higher trust disposition (X) perceives the organizational context to be more
positive. This, in turn, positively influences whether a subordinate will be perceived as
trustworthy (Y).

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

Figure 12.11: Possible Patterns of Complementary Mediation

Competitive Partial Mediation


In a competitive partial mediation, the direct effect c’ and indirect effect a*b point in a different
direction. A negative a*b*c’ value indicates the presence of competitive mediation (see Figure
12.10 and Figure 12.12). This indicates that a portion of the effect of X on Y is mediated through
M, whereas X still explains a portion of Y that is independent of M. In the past, researchers are
often focused only on complementary mediation (Zhao et al., 2010). In the competitive partial
mediation hypothesis, it is assumed that the intermediate variable will reduce the magnitude of
the relationship between the independent and dependent variables. However, it is possible that
the intermediate variable could increase the magnitude of the relationship between the
independent and dependent variables. Competitive partial mediation has often been called a
“negative confounding” or an, “inconsistent” model. For example, McFatter (1979) suggested
that intelligence (X) has a positive influence on individual performance (Y); however, this effect
could be suppressed by the task boredom variable (M) because intelligence (X) leads to greater
task boredom (M), and this variable has a negative effect on individual performance (Y).

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

Figure 12.12: Possible Patterns of Competitive Mediation

VAF Approach

Many researchers often inquire whether to use the VAF approach to conclude the mediation
type. VAF is the calculation of the ratio of the indirect-to-total effect. VAF determines the extent
to which the mediation process explains the dependent variable’s variance. For a simple
mediation, the proportion of mediation is defined as:

Using VAF as classification for mediation portion is not uncritical. If the indirect effect is
significant, but does not mediate much of the total effect c, the VAF would be low. As shown in
Figure 12.10, a significant indirect effect a*b and insignificant direct effect c’ would indicate a full
mediation. Such differences between significance testing and VAF interpretation especially
occur when samples sizes are small in terms of the power, or a high multicollinearity between
the constructs exists (Rucker et al., 2011). Researchers should be aware that detecting a
significant indirect effect, a*b, is always higher than detecting a direct effect c’ (Cohen, 1988).
The rule of thumb is, if the VAF is less than 20 percent, one should conclude that nearly zero
mediation occurs. A situation in which the VAF is larger than 20 percent and less than 80 percent
could be characterized as a typical partial mediation (Hair et al., 2017); and a VAF above 80
percent indicates a full mediation. However, in this situation, the VAF may amount to, for
example, only 60 percent, in such case, researchers should not assume full mediation (see Figure
12.13).

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

Additionally, the interpretation of VAF is clear only for consistent or complementary mediating
effects (i.e. c and a*b having the same effects, positive or negative. See Figure 12.11). In one case,
VAF can be greater than one when the total effect, c, is smaller than the indirect effect, a*b. This
is the case for a suppressor effect. In situations where the VAF is greater than one and the direct
effect c’ is not significant; there is no strong indication that suppression is present. In this
situation, Shrout and Bolger (2002), suggest considering a VAF equals to one representing a full
mediation. In another case, one could consider inconsistent mediation (i.e. c and a*b having
different effects) as yielding a negative VAF, or a VAF tending to infinity as c approaches zero
(Hayes, 2009). Therefore, some researchers are advised of the calculation of VAF only when the
absolute value of the standardized total effect c = a*b + c’ is at least 0.20 (Hair et al., 2017).

Figure: 12.13: Variance Accounted For (VAF)


Source: Primer on Partial Least Squares
Structural Equation Modeling (PLS-SEM) Book (Hair et al., 2017)

For example, using the result produced in Figure 12.14, the calculation of VAF is 0.184/ (0.184 +
0.172), which gives the result 0.517. Thereby, we can determine that the mediation result from
Figure 12.14 could be characterized as a typical partial mediation because the VAF value of 0.517,
or 51.7%, is in the range of VAF being larger than 20% and less than 80%.

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

Note: β = 0.184 is derived from multiplication of a*b = 0.456*0.404


Figure 12.14: Example of using VAF to assess the mediation result of Technology Adoption

In general, we would suggest to researchers that the VAF concept may provide some deeper
insights into mediation analysis, but it should be interpreted very cautiously without mixing the
use of full, partial and no mediation. The logic is that VAF is only useful when the researcher is
interested in determining the extent to which the variance of the dependent variable is directly
explained by the independent variable, and how much of the target construct's variance is
explained by the indirect relationship via the mediator variable. VAF also works well if a
researcher would like to compare the strength of multiple mediators in a model on each indirect
relationship.
Overall, we would like to warn researchers to be cautious about the procedure when
determining the types of mediation effects, and also the use of VAF to determine the strength
of indirect effect, because it might lead to deficient practice of research. Yet, we are also not
bold enough to say that such logic, suggested by Zhao et al., (2010), and Nitzl et al., (2016), is
completely incorrect. We would want researchers to continue to keep abreast of the latest
literature and developments of testing mediation because this will help to guide researchers
through the process of publication in higher impact journals, and have a higher success rate in
the examination process of postgraduate viva.

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

12.11 Limitations of Full and Partial Mediation Approach

The problem with this reasoning is that establishing variable M completely mediates the effect
of X on Y, says nothing whatsoever about the existence, or absence, of other possible mediators
of X’s effect (Rucker, Preacher, Tormala, & Petty, 2011). Even if one has completely accounted
for the effect of X on Y with the favoured mediator, this does not prohibit another researcher
from being able to make the same claim as you, but using an entirely different mediator.
However, if there are multiple mediators that completely mediate X’s effect when considered in
isolation, then what value is there in claiming that your favoured mediator does? Therefore, “…
it is an empty claim, with no real value or meaning and nothing especially worthy of celebration
much less even hypothesizing in the first place.” (Hayes, 2013, p. 171).

Moreover, the claim of partial mediation is a celebration of a mis-specified model (Hayes, 2013).
Philosophically, all relationships are mediated by something. When researchers claim that M
partially mediates the relationship between X and Y, they are admitting that part of X’s effect on
Y has not been accounted for by M. So what is accounting for X’s remaining effect as evidenced
by a statistically significant direct effect? In other words, something is missing in that model.
Although all models are mis-specified to some extent, why hypothesize it? And why celebrate
when you support a hypothesis of partial mediation? When you are hypothesizing a partial
mediation, you are celebrating a mis-specified model (Hayes, 2013).

Table 12.3 provides a brief description of several issues related to mediation assessment, its
consequences, and proposed solutions. This information can be used as a resource for
researchers as well as a checklist for reviewers who evaluate manuscripts, including mediation
tests.

Table 12.3: Problems Regarding Mediation Assessment, Detrimental Consequences for


Substantive Conclusions, and Proposed Solutions.
No Problems Detrimental Consequences for Proposed Solution
Substantive Conclusions
1 A significant • This assumed • Conduct the mediation
relationship between prerequisite can lead test without the
the antecedent and the researchers to overlook precondition that the
outcome is a mediation when the relation between the
prerequisite for testing direct and indirect antecedent and the
mediation. effects in a model have outcome should be
opposite signs. significant.

2 The causal-steps • The causal-steps • Compute the size of the


procedure yields procedure does not indirect effect by
information about the involve the multiplying the paths to
magnitude of the computation or test of and from the mediator
indirect effect the indirect effect, and (i.e., paths a & b in Figure
transmitted through the therefore does not 12.13) and test this product
mediator. indicate the magnitude using nonparametric
of the mediating effect, procedures such as the
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
No Problems Detrimental Consequences for Proposed Solution
Substantive Conclusions
or allow its comparison bootstrap (rather than the
with other effects (e.g., Sobel test that assumes
the indirect vs. the normality in the
direct effect). distribution of product
coefficients).
3 The evaluation of • Because the direct • Conclude that mediation
mediation should effect has no bearing exists when the indirect
include a test of the on the presence of the effect is supported,
direct effect that indirect effect, this test regardless of the presence
bypasses the mediator. can lead researchers to or absence of a direct
inappropriately dismiss effect.
mediating effects.
4 The causal-steps • Routinely including • If the theory under
procedure routinely direct effects violates consideration predicts
includes a direct path the principle of complete mediation, then
from the antecedent to parsimony and prompts use the full mediation
the outcome, researchers to test model as a baseline (i.e.,
regardless of whether models that are not a*b ≠ 0 and c’ = 0 in Figure
this path is conceptually aligned with theory. 12.4) and formally test the
justified. consequences of omitting
the direct effect on the fit
of the model.
5 Mediation can be tested • Mediated models • When possible, assess
satisfactorily with cross contain causal paths mediation using
sectional data. that imply the passage longitudinal data,
of time, and testing preferably with panel
these paths with cross- models that allow the
sectional data can comparison of alternative
produce biased causal flows.
estimates.

6 Lack of attention to • Measurement error can • Create and use reliable


measurement error in bias path estimates measures. As a second
tests of mediation. upward or downward, best option, use multiple-
leading to conclusions item measures for all
that are unwarranted. constructs and analyze the
data using structural
equation modeling with
latent variables.

Source: Extracted from Improving Our Understanding of Moderation and Mediation in Strategic
Management Research by Aguinis, Edwards & Bradley (2016)

References:
Aguinis, H., Edwards, J. R., & Bradley, K. J. (2016). Improving our understanding of moderation and mediation in
strategic management research. Organizational Research Methods, 1-21.
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research:
conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51(6), 1173–
1182.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
Cheung, G.W. & Lau, R.S. (2008). Testing mediation and suppression effects of latent variables: bootstrapping with
structural equation models. Organizational Research Methods, 11(2), pp. 296-325.
Cohen, J. (1988), Statistical Power Analysis for the Behavioral Sciences, Lawrence Erlbaum Associates, Hillsdale, NJ.
Green, J. P., Tonidandel, S., & Cortina, J. M. (2016). Getting Through the Gate: Statistical and Methodological Issues
Raised in the Reviewing Process. Organizational Research Methods, 19(3), 402–432.
Hair, J. F., Hult, T. M., Ringle, C. M., & Sarstedt, M. (2017). A primer on partial least square structural equation
modeling (PLS-SEM). Sage Publications.
Hair, J. F., Ringle, C. M., & Sarstedt, M. (2011). PLS-SEM: Indeed A Silver Bullet. Journal of Marketing Theory and
Practice, 19(2), 139-151.
Hayes, A. F. (2009). Beyond Baron and Kenny: Statistical mediation analysis in the new millennium. Communication
Monographs, 76(4), 408-420.
Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based
approach. New York: Guilford Press.
Hayes, A. F., & Scharkow, M. (2013). The relative trustworthiness of inferential tests of the indirect effect in statistical
mediation analysis does method really matter?. Psychological Science, 0956797613480187.
James, L. R., & Brett, J. M. (1984). Mediators, moderators, and tests for mediation. Journal of Applied
Psychology, 69(2), 307.
James, L. R., Mulaik, S. A., & Brett, J. M. (2006). A tale of two methods. Organizational research methods, 9(2), 233-
244.
Judd, C. M., & Kenny, D. A. (1981). Process analysis: Estimating mediation in treatment evaluations. Evaluation
review, 5(5), 602-619.
Kenny, D. A. (2003), “Mediation,” http://www.users.rcn.com/ dakenny/mediate.htm/
Little, T. D., Card, N. A., Bovaird, J. A., Preacher, K. J., & Crandall, C. S. (2007). Structural equation modeling of
mediation and moderation with contextual factors. Modeling contextual effects in longitudinal studies, 1,
207-230.
Machamer, P., Darden, L., & Craver, C. F. (2000). Thinking about mechanisms. Philosophy of science, 67(1), 1-25.
MacKinnon, D. P., Coxe, S., & Baraldi, A. N. (2012). Guidelines for the Investigation of Mediating Variables in Business
Research. Journal of Business and Psychology, 27(1), 1-14.
MacKinnon, D. P., Fairchild, A. J., & Fritz, M. S. (2007). Mediation analysis. Annual. Revision Psychology, 58(1), 593-614.
Malhotra, M. K., Singhal, C., Shang, G., & Ployhart, R. E. (2014). A critical evaluation of alternative methods and
paradigms for conducting mediation analysis in operations management research. Journal of Operations
Management, 32(4), 127-137.
McFatter, R. M. (1979). The use of structural equation models in interpreting regression equations including
suppressor and enhancer variables. Applied Psychological Measurement, 3(1), 123-135.
Mehrabian, A., & Russell, J. A. (1974). An approach to environmental psychology. The MIT Press.
Morgan, S.L., Winship, C., 2007. Counterfactuals and Causal Inference: Methods and Principles for Social Research.
Cambridge University Press, New York, NY.
Nitzl, C. (2016). The use of partial least squares structural equation modelling (PLS-SEM) in management accounting
research: Directions for future theory development. Journal of Accounting Literature, 37, 19-35.
Nitzl, C., & Hirsch, B. (2016). The drivers of a superior’s trust formation in his subordinate: The manager–
management accountant example. Journal of Accounting & Organizational Change, 12(4), 472-503.
Nitzl, C., Roldán, J. L. & Cepeda, G. (2016). Mediation analyses in partial least squares structural equation modeling:
Helping researchers discuss more sophisticated models. Industrial Management & Data Systems. 116(9), 1849-
1864.
Preacher, K. J., & Hayes, A. F. (2004). SPSS and SAS procedures for estimating indirect effects in simple mediation
models. Behavior Research Methods, Instruments, & Computers, 36(4), 717-731.
Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect
effects in multiple mediator models. Behavior Research Methods, 40(3), 879-891.
Roldán, J. L., & Sánchez-Franco, M. J. (2012). Variance-based structural equation modeling: guidelines for using
partial least squares. Research methodologies, Innovations and Philosophies in Software Systems Engineering
and Information Systems, 193.
Rucker, D. D., Preacher, K. J., Tormala, Z. L., & Petty, R. E. (2011). Mediation analysis in social psychology: Current
practices and new recommendations. Social and Personality Psychology Compass, 5(6), 359-371.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.
Rungtusanatham, M., Miller, J. W., & Boyer, K. K. (2014). Theorizing, testing, and concluding for mediation in SCM
research: Tutorial and procedural recommendations. Journal of Operations Management, 32(3), 99-113.
Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and nonexperimental studies: new procedures and
recommendations. Psychological methods, 7(4), 422.
Wood, R. E., Goodman, J. S., Beckmann, N., & Cook, A. (2008). Mediation testing in management research a review
and proposals. Organizational research methods, 11(2), 270-295.
Zhao, X., Lynch, J. G., & Chen, Q. (2010). Reconsidering Baron and Kenny: Myths and truths about mediation analysis.
Journal of Consumer Research, 37(2), 197-206.

How to Cite:
Ramayah, T., Cheah, J., Chuah, F., Ting, H., & Memon, M. A. (2018). Partial least squares structural equation modeling (PLS-SEM) using
SmartPLS 3.0: An updated guide and practical guide to statistical analysis (2nd ed.). Kuala Lumpur, Malaysia: Pearson.

View publication stats

You might also like