Social Indicators Research

https://doi.org/10.1007/s11205-020-02271-5

ORIGINAL RESEARCH

Methodological PLS‑PM Framework for SDGs System

Rosanna Cataldo1   · Corrado Crocetta2 · Maria Gabriella Grassia1 ·

Natale Carlo Lauro3 · Marina Marino1 · Viktoriya Voytsekhovska4

Accepted: 8 January 2020

© Springer Nature B.V. 2020

Abstract
Sustainability is the biggest challenge of our generation, because civilization has reached
a point where natural resources are in rapid decline. It’s a complex multidimensional phe-
nomenon, which was studied for couple decades already. Nowadays different social con-
cepts, such as sustainability, but also quality of life, satisfaction, are difficult and complex
to define. The main problem for researchers is to find appropriate tools to obtain a com-
posite indicator able to synthesize and represent these phenomena. The work focuses on
building a system of composite indicators of Sustainability through to Structural Equation
Modeling, specifically with the use of Partial Least Squares-Path Modeling. In recent years
many advances have been developed, in the context of these models to solve some prob-
lems related to the role that the composite indicators play within that system; in particular
on the aspects linked to the high level of abstraction, when a composite indicator is mani-
fold, lacks its own manifest variables and is described by various underlying blocks. The
aim of this paper is to demonstrate how these recent developments in Partial Least Square-
Path Modeling could help you to build a SDGs system and to provide a better measure of
this complex social phenomenon.

Keywords  Composite indicators · Higher-order construct · PLS-path modeling ·

Sustainable development goals

1 Introduction

The concept of sustainability and particularly of sustainable development figure among the
most ambiguous and controversial in the literature. Sustainability has already been a popu-
lar term in the field of environmental economics. However, its definition is still ambiguous.
The term “sustainability” was first proposed in 1980 by the International Union for Con-
servation of Nature and Natural Resources (International Union for Conservation of Nature
and World Wildlife Fund 1980). One of the important concluding remarks of this report
was that human beings, in their quest for economic development and enjoyment of the
riches of nature, must come to terms with the reality of resource limitation and the carry-
ing capacities of ecosystems, and must take account of the needs of future generations. The

* Rosanna Cataldo
rosanna.cataldo2@unina.it
Extended author information available on the last page of the article

13
Vol.:(0123456789)
 R. Cataldo et al.

Brundtland Report provides the most popular notion of sustainability (sustainable develop-
ment):1 development that meets the needs of the present without compromising the ability
of future generations to meet their own needs (WCED, SPECIAL WORKING SESSION
1987). The definition provided by the Brundtland Report is a characteristic definition of
sustainability (Bithas and Christofakis 2006).
Then there were several other milestones such as Rio Earth Summit in 1992. An impor-
tant achievement of the summit was an agreement on the Climate Change Convention
which in turn led to the Kyoto Protocol (1997) and the Paris Agreement (United Nations
Development Programme 1992).
Measuring sustainable development and setting indicators for assessment and continu-
ous improvement of the development of nations were discussed for the first time at that
Rio Earth Summit, also known as Agenda 21, which called for development of “systems
for monitoring and evaluation of progress towards achieving sustainable development
by adopting indicators that measure changes across economic, social and environmental
dimensions” (United Nations Development Programme 1992). Ever since, a wide range
of indices have been proposed, each aiming to rank countries based on their achievements
across different sets of indicators (Strezov et al. 2017).
Another important step in the history was the adoption of Millennium Development
Goals (MDGs) (United Nations 2009). The MDGs mark a historic and effective method of
global mobilisation to achieve a set of important social priorities worldwide. They express
widespread public concern about poverty, hunger, disease, unmet schooling, gender ine-
quality, and environmental degradation (Sachs 2012).
In 2012 Rio+20 Summit aimed at reconciling the economic and environmental goals of
the global community (United Nations 2015b). It has issued a report recommending that
the world adopt a set of Sustainable Development Goals (SDGs) which will build upon
the MDGs and converge with the post 2015 development agenda. In fact, the conference
on sustainable development, that led to United Nations 2030 Agenda, calls for an exten-
sive set of global indicators, which consists of 17 Goals and 169 Targets as a result of
joint work of all stakeholders: academia, civil society and government’s official delegates
(United Nations 2015a). According to Sachs (2012), the SDGs are an important idea, and
could help finally to move the world to a sustainable trajectory. The detailed content of the
SDGs, if indeed they do emerge in upcoming diplomatic processes, is very much up for
discussion and debate. In his paper, Sachs believes that the SDGs can benefit from both
the successes and the shortfalls of the MDGs. The successes are notable. Unlike many UN
goals, the MDGs are still very much with us almost 12 years after their adoption. Moreo-
ver the success of the SDGs will need societies worldwide to invest adequately in their
success. Sustainable development is the only viable path for humanity, but it will not be
achieved unless a small part of consumption spending is turned into investments for long-
term survival. Sachs (2012). Based on literature review, we can see that it is difficult to
define sustainability precisely and convincingly. Moreover, from this review of definitions,
sustainability consists of two main elements: the triple bottom line as an abstract notion
of environmental, social and economic processes and over time preservation or intergen-
erational equity (Mori and Christodoulou 2012). The concept of development and growth
was studied by Solow (1994) and Daly et al. (1994), who argue that sustainability is linked
not only to the economic component but also to the social and environmental dimensions.

1
 The concept of sustainable development is quite different from that of sustainability in that the word
“development” clearly points to the idea of change, of directional and progressive change (Gallopín 2003).

13
Methodological PLS‑PM Framework for SDGs System

Fig. 1  The graph of Sustain-

ability

So, starting from these considerations, the sustainable development is today viewed from
the perspective of the three pillars model (Hopwood et al. 2005): economic, environmental
and socio-cultural pillars respectively (Fig. 1). The economic dimension should build on
the MDGs, which have helped to advance the world’s agenda in the fight against poverty,
hunger, and disease. The second pillar is environmental sustainability, usefully conceptu-
alised by the global planetary boundaries. The third broad SDG is social inclusion, the
commitment to future economic and technological progress under conditions of fairness
and equitable access to public services, and with the government counter acting social dis-
crimination on the basis of gender, ethnic origin, religion, and race.
Bettencourt and Kaur (2011) explain the importance of international and regional scien-
tific organizations roles that need more of theoretical and methodological framing beyond
the topics of individual research in the very area. Further development of sustainability
science is described by Anand and Sen (2000), Kates (2011) and others. A country’s eco-
nomic, social and environmental performance is now seen through a framework of indica-
tors of Sustainable Development, which were proclaimed in September 2015 during UN
General Assembly. At the UN in New York the Open Working Group created by the UN
General Assembly proposed a set of global SDGs which comprises 17 goals and 169 tar-
gets. Further to that, a preliminary set of 330 indicators was introduced in March 2015.
Some SDGs build on preceding MDGs while others incorporate new ideas (Hák et  al.
2016).2 The successful implementation of Sustainable Development Agenda 2030 and its
17 goals requires scientific concept to be developed in order to monitor sustainable devel-
opment in its three key areas: economic, environmental and social. When it comes to meas-
uring sustainability it’s often linked to well-being indicators, however they are not identical
and shall be tackled differently (Neumayer 2010), Neumayer states, that fully integrated
indicators of well-being and sustainability encounter a fundamental conceptual problem:
those, that affect current well-being would not affect sustainability or not in the same way,

2
  The MDGs were derived from the United Nations Millennium Declaration, adopted by 189 nations in
2000. Most of the goals and targets were set to be achieved by 2015 on the basis of the global situation dur-
ing the 1990s. The baseline for the assessment of progress is therefore 1990 for most of the MDGs targets.

13
 R. Cataldo et al.

and vice versa. Thus he proposes to complement Human Development Index (HDI) with
appropriate sustainability indicators. The Sustainable Development Goals are the blueprint
to achieve a better and more sustainable future for all. They address the global challenges
we face, including those related to poverty, inequality, climate change, environmental deg-
radation, peace and justice. The SDGs are interconnected, the potential synergies or trade-
offs between the goals make the SDGs interactions complex, as they are clearly not inde-
pendent from each other (Pradhan et  al. 2017). However, the correlation does not imply
causality, so often observed synergies between sustainable development indicators could
be independently related to another process driving both indicators and in turn resulting
in correlations. Therefore it becomes increasingly important to study those dependencies.
International interest in sustainability research has significantly increased in recent
years due to boosts of the “Beyond GDP” initiative and the Stiglitz et  al. (2017) report,
and the idea of the SDGs has quickly gained ground because of the growing urgency of
sustainable development for the entire world (Sachs 2012). Researchers stress the need to
operationalise the SDGs’ targets and see an explicit need to structure the SDGs indica-
tors into a coherent framework. They state that there is a need for global, integrated and
scientifically based information on sustainability, and that the achievement of the SDGs
need to be assessed and accompanied by target and indicators in a well structured frame-
work. The path to quantifying and monitoring a multidimensional concept like sustainable
development is fraught with challenges (Swain 2018). It requires a profound understanding
of sustainable development, comprehension about how to operationalize and implement
the SDGs, access to all forms of data and the expertise to analyse and interpret the results.
Moreover, in his critical analysis of the SDGs, Swain writes that the ambitious UN-adopted
SDGs have been criticized for being inconsistent, difficult to quantify, implement and mon-
itor (Swain 2018). Disparaging analysis suggests that there exists a potential inconsistency
in the SDGs, particularly between the socio-economic development and the environmental
sustainability Goals (Spaiser et al. 2017; ICSU, ISSC 2015). Critiques also raise questions
on if and whether the SDGs can be measured and monitored. Despite these criticisms, in
these years, researchers have tried to measure and monitor sustainable development indi-
cators, whose proposals are analyzed in detail in the Sect. 3. Starting from the successful
implementation of Sustainable Development Agenda 2030 and from the existing Elemen-
tary Indicators (EIs) for the estimation of the single areas, we propose a methodological
framework, estimated through to Higher-Order PLS-PM, of SDGs System, in which the
hierarchical relationships between this different LVs are considered and their connections
to a Higher-Order Global Sustainability CI is estimated.
This article does not seek to redefine the sustainability’s concept or propose new SDGs
and/or targets. The aim of this paper is to propose PLS-PM as a method for studying exist-
ing SDGs indicators and demonstrate how it could help you to define the framework for
SDGs indicators in order to provide a better measure of this complex multidimensional
social phenomenon. There have been many and varied approaches in literature, such as
simple arithmetic mean of the original indicators, PCA or FA to measuring sustainable
development, based on simple aggregation techniques and easily calculated, but PLS-PM
represents an important breakthrough with respect to these traditional aggregation meth-
ods. The choice of using the PLS-PM is useful for several reasons, particularly, it should be
used for constructing a System of CIs, help to identify critical indicators, to construct rank-
ing of countries and to provide an interpretation of the results. This approach allows you to
estimate causal relationships, defined according to a theoretical model linking two or more
latent complex concepts, each measured through a number of observable indicators (Lauro
et al. 2018). The paper is organized as follow. Section 2 consists in a brief literature review

13
Methodological PLS‑PM Framework for SDGs System

on different method proposed in the last few years for constructing sustainability indices.
Section 3 propose a PLS-PM as a methodological framework for constructing a SDGs sys-
tem. In this section advantages of this technique are discussed, in particular we explain
how and why we can use PLS-PM and in detail the Higher-Order PLS-PM as a method for
constructing SDGs composite indicators. Section 4 reports theoretical principles of PLS-
PM approach and, in-depth, the Higher-Order PLS-PM and different approaches proposed
in the literature. In Sect.  5 the constructing of the methodological framework for SDGs
using Higher-Order PLS-PM is proposed, while Sect. 6 reports the main result of a case
study for measuring SDgs of European Countries using PLS-PM approach. Finally, Sect. 7
conclusions are drawn and some open issues are explained.

2 Literature Review

The interest towards understanding and measuring the phenomena manifests in numerous
researches and publications (Neumayer 2007; Stiglitz et  al. 2017; Bettencourt and Kaur
2011; Sachs 2012; Gan et  al. 2017; De Smedt et  al. 2018). There have been many and
varied approaches to measuring sustainable development. Many involve the identification
of indicators of sustainable development and methods for collating these values, which
are often indexed to allow for comparison (Evans et al. 2015). There is a significant body
of international studies (Böhringer and Jochem 2007; Moran et  al. 2008; Bilbao-Ubillos
2013; Estoque and Murayama 2014; Evans et al. 2015) aiming to review and compare the
indices developed to measure sustainable development. However, there is still no consen-
sus on a single index that is most acceptable by the scientific and political communities
(Strezov et al. 2017). It is recognized, that measurement of sustainable development goals
is somehow problematic due to its complicated nature (Spaiser et al. 2017; Fabbrizzi et al.
2017). When it comes to measuring sustainability, it’s often linked to well-being indicators
or indicators that go beyond GDP measurement, however they are not identical and shall
be tackled differently (Neumayer 2007) it states, that fully integrated indicators of well-
being and sustainability encounter a fundamental conceptual problem: those, that affect
current well-being would not affect sustainability or not in the same way, and vice versa.
Thus he proposes to complement HDI with appropriate sustainability indicators. In general
there are nine main indices, that measure sustainability to some extent, but only two of
them Sustainable Society Index and Genuine Savings Index include all three areas: eco-
nomic, social and environmental (Strezov et al. 2017). Those indices are easily calculated,
based on simple aggregation techniques therefore can serve as excellent base for monitor-
ing sustainable development on the cross country’s level. However they do not demon-
strate important interlinkages between goals and do show only observed variables. Current
research shows big advances in reducing the dimensionality of the initial data by means
of linear transformation techniques (Dunteman 1989; Smith 2002; Tenenhaus et al. 2005;
Spaiser et  al. 2017; Mazziotta and Pareto 2019) such as Principle Component Analysis
(PCA). Those approaches help to understand social phenomena with large number of data,
since they reduce dimensionality of the original indicator set. Nevertheless they are devel-
oped to examine the relationships between variables but not to determine weights. One of
the difficulties here is that contradictory outcomes may be obtained, since the weights are
correlation based and do not necessarily represent real tendencies among indicators (Her-
mans et  al. 2008). Therefore Factor Analysis (FA) can attribute lower weights to impor-
tant indicator due to their weak correlation with other dimensions. Spaiser et  al. (2017)

13
 R. Cataldo et al.

use Confirmatory FA (CFA) to prove incompatibility of sustainable development goals on

global scale. While it’s important to show factors, that contribute to successful sustainable
development but can not be directly observed or measured (health programmes, govern-
ment investments, renewable energy and others) it’s not clear how first and second order
latent constructs are related in given study. There are other data-driven approaches, such
as Bali Swain and Wallentin (2017) who, employing Structural Equation Models (SEM) to
the Sachs et al. (2016) data set, compare their SDG construct with SDG indices from Sachs
et  al. (2016) and the HDI to investigate if these different measures suggest conflicting
policy inferences to the developing and developed countries in terms of achieving SDGs
(Swain 2018). In their analysis, the three pillars of SDGs are estimated in two steps and in
the first step, the PCA is employed, while in the second step, Exploratory FA (EFA) is con-
ducted to determine the SDGs that measure the latent factors or sustainable development
pillars. According Swain (2018), the result obtained bay (Bali Swain and Wallentin2017)
are in line with the literature that visualizes SDGs as an interlinked set of policies with
trade-offs and synergies. As was stated before sustainability is often linked to well-being
therefore we can as well study the use of PCA in order to select main factors (Mazziotta
and Pareto 2019). The main problem with data-driven approaches is that, they lack theo-
retical foundation and consequently can send misleading signals to policy-makers, such an
approach can be especially crucial for SDGs where multicollinearity is common therefore
can be misleading.

3 An Alternative Proposal: Use of PLS‑PM for Studying Sustainability

The goal of many research in social, economic and political fields is to obtain a whole
description of the various facets of a complex phenomenon, through a suitable synthesis
of the associated EIs. This combination can be obtained by applying methodologies known
as Composite Indicator (CI) (Jarvis et al. 2003; Salzman 2003). The construction of a CI
implies the search for a suitable synthesis of a number of Manifest Variables (MVs) in
order to achieve a simple representation of a multidimensional phenomenon. Accordingly
a CI can be considered as a latent concept, not directly measurable, the estimation of which
can be obtained through the values of the MVs (i.e. the EIs). CIs are useful as a means
of summarising the information provided by base indicators in an overall judgement or
assessment of countries, and developing an assessment that permits countries to be ranked
from best to worst, taking into account a whole set of features. Generally, they are very
useful in order to deal with those phenomena that can not be observed directly. Different
social concepts, such as the quality of life, satisfaction and sustainability, are multidimen-
sional concepts that are difficult and complex to define. The main problem for researchers
is to find appropriate tools to obtain a CIs able to synthesize and represent these phenom-
ena. The existing literature offers different alternative approaches in order to obtain a CI
(Trinchera et al. 2008). According Lauro et al. (2018) and Cataldo et al. (2017), PLS-PM
should be used for constructing a System of CIs, help to identify critical indicators, to con-
struct ranking of countries and to provide an interpretation of the results. PLS-PM allows
you to estimate causal relationships, defined according to a theoretical model linking two
or more latent complex concepts, each measured through a number of observable indica-
tors. The basic idea is that the complexity inside a system can be studied by taking into
account the entirety of the causal relationships among the LVs, each measured by several
MVs (Lauro et  al. 2018). PLS-PM represents an important breakthrough with respect to

13
Methodological PLS‑PM Framework for SDGs System

traditional aggregation methods, such as a PCA or a simple arithmetic mean of the original
indicators. Instead of taking the unweighted sum of the indicators (or unit-weights for all
the indicators), PLS-PM assigns weights to the original variables taking into account the
network of relationships between the constructs and the variance and covariance structure
within and between the blocks of variables. Moreover, PLS-PM provides components with
specific proprieties in order to enhance interpretation of the composites and the relation-
ships among them. In particular, depending on the chosen estimation options, PLS-PM
provides components that are as much correlated as possible to each other while explaining
the variances of their own set of variables. PLS-PM is a procedure based on simple and
multiple Ordinary Least Squares (OLS) regressions; it has minimal demands on measure-
ment scales, sample size, and data distributions and it is particularly applicable for predic-
tive applications and theory building (Chin 1998; Davino et al. 2018).
The choice of using the PLS-PM is particularly useful for several reasons. This approach
has as its main advantages its applicability to small sample, the ability to estimate quite
complex models (with many latent and observable variables) and less restrictive require-
ments concerning normality and variable and error distributions (Henseler et  al. 2009).
Furthermore, PLS-PM approach provides the possibility of working with missing data and
in the presence of multi-collinearity. Another advantage of this approach, as compared to
other multivariate techniques, is that it examines simultaneous a series of dependence rela-
tionship, using a single statistical approach to test the full scope of projected relations (Hair
et  al. 2009). Furthermore, this approach provides researchers with much more flexibility
as it enables using both formative and reflective measurement models, providing a more
nuanced testing of theoretical concepts (Hair et al. 2011). The PLS-PM approach should
be used in order to not only reduce dimension but find relations between Composite Indi-
cators and their blocks (Tenenhaus et al. 2005). However, a fundamental distinction must
be made between reducing dimensionality and constructing composite indicators, because
they are repeatedly confused. Both the procedures aims to summarize a set of variables or
individual indicators, but reducing dimensionality focuses on extracting the most important
information from the data, whereas constructing composite indicators focuses on the use of
a measurement model that can be reflective or formative (Mazziotta and Pareto 2019). For
a detailed explanation about composite indicators, formative and reflective measurement,
see Mazziotta and Pareto (2019), Michalos (2014) and Nardo and Saisana (2008). For this
purpose, it is essential to define the model of measurement in order to describe relation-
ships between the phenomenon to be measured (LVs) and its elementary indicators (MVs).
According Mazziotta and Pareto (2019), measuring well-being requires a formative
approach, where the index to be constructed does not exist as an independent entity, but
it is a composite measure directly determined by a set of non-interchangeable individual
indicators or pillars.3
Starting form these suggestions and observations and from the successful implementa-
tion of Sustainable Development Agenda 2030 and its 17 goals , we propose a System of
SDGs Indicators estimated through the Higher-Order PLS-PM, in which the hierarchical
relationships between this different LVs are considered.

3
  One of the oldest and most famous formative composite indices is the HDI by United Nations Devel-
opment Programme (United Nations Development Programme 2010). It s a composite measure of human
development that include three theoretical dimensions (Health, Education and Income) (Mazziotta and
Pareto 2019).

13
 R. Cataldo et al.

There are three main reasons for the inclusion of an Higher-Order Construct Model in
PLS-PM. Firstly, by establishing Higher-Order Construct Models, researchers can reduce
the number of relationships in the structural model, making the PLS path model more par-
simonious and easier to grasp. Secondly, Higher-Order Construct Models prove valuable if
the constructs are highly correlated; the estimations of the structural model relationships
may be biased as a result of collinearity issues, and a discriminant validity may not be
established. In situations characterized by collinearity among constructs, a Second-Order
Construct can reduce such collinearity issues and may solve discriminant validity prob-
lems. Thirdly, establishing Higher-Order Construct Models can also prove valuable if
formative indicators exhibit high levels of collinearity. Provided that the theory supports
this step, researchers can split up the set of indicators and establish separate constructs in a
higher order structure (Hair et al. 2014).
In the next section, the principles of PLS-PM approach is presented and the Higher-
Order PLS-PM approach is described in detail.

4 Theory of PLS‑PM

Two very important review papers on the PLS approach to SEM are Chin (1998) and
Tenenhaus et al. (2005). Recently, Lauro et al. (2018) have presented some current devel-
opments in PLS-PM for the treatment of non-metric data, hierarchical data, longitudinal
data and multi-block data.

4.1 The PLS Path Model

A PLS-PM is made up of two elements, the measurement model (also called the outer
model), which describes the relationships between the MVs and their respective LVs,
and the structural model (also called the inner model), which describes the relationships
between the LVs. In the measurement model, an LV 𝜉 is an unobservable variable (or con-
struct) indirectly described by a block of observable variables xk which are called MVs or
indicators. There are three ways to relate the MVs to their LVs:

• The reflective way (or outwards directed way): each MV reflects the corresponding LV.
In this case, the MVs should be highly correlated, due to fact that they are correlated
with the LV of which they are expression. In other words, the block has to be homoge-
neous.
• The formative way (or inwards directed way): the LV is supposed to be generated by its
own MVs.
  Each MV or every set of MVs represents a different level of the underlying latent
concept.
• The MIMIC way (a mixture of the reflective and formative ways): combination of the
reflective and formative ways.

In the PLS-PM framework, the structural model specifies the relationships between the
LVs; an LV, if it is supposed to depend on other LVs, is called exogenous, and, other-
wise, endogenous. The evaluation of the measurement and structural model results in
PLS-PM builds on a set of non-parametric evaluation criteria and uses procedures such as

13
Methodological PLS‑PM Framework for SDGs System

bootstrapping and blindfolding. This process involves a separate assessment of the meas-
urement model and the structural model.
A key characteristic of the PLS-PM method is the extraction of CI scores. One of the
greatest advantages of PLS-PM is these CI scores. In the System of indicators built with
PLS-PM, you can obtain the scores for each indicator, exogenous or endogenous, and for
each indicator you can make a ranking among units. Moreover, PLS-PM provides informa-
tion on the relative importance of constructs in explaining other constructs in the structural
model. Information on the importance of constructs is relevant for drawing conclusions.
For this reason, an Importance-Performance Matrix Analysis is a valuable decision mak-
ing tool. The matrix contrasts the structural model’s total effects (the importance) and the
average values of the indicator (the performance). As a result, conclusions can be drawn on
two dimensions (i.e., both importance and performance), which is particularly important in
order to prioritize actions. The Importance-Performance Matrix Analysis is a very intuitive
scatter plot and we can analyze themes according to the quadrant in which CIs are placed:
upper-right quadrant: the area to be maintained, in which the CIs have a high value for
the mean and the path coefficient; lower-right quadrant: the area to improve because the
CIs have a high mean value and a low path coefficient; lower-left quadrant: the area of the
monitoring, in which the CIs have a low value for the mean and the path coefficient; and
upper-left quadrant: the critical area, because the CIs have a high impact but a low mean
value. A similar scatter plot can be considered also for the MVs. In this kind of analysis,
we have the possibility to analyze the strengths, weaknesses, opportunities, and threats of
constructs, that are considered in the model in order to estimate a latent concept.

4.2 Higher‑Order PLS‑PM

PLS-PM is a suitable tool for the investigation of models with a high level of abstraction
(Lohmöller 2013), in cases where the building of a system of Indicators depends on differ-
ent levels of construction. Higher-order constructs in PLS-PM are considered as explicit
representations of multidimensional constructs that exist at a higher level of abstraction
and are related to other constructs at a similar level of abstraction, completely mediating
their influence from or to their underlying dimensions (Chin 1998).
The Higher-Order Construct Models are also known as Hierarchical Models. As their
name says, what makes these models so special is that they contain LVs of a Higher Order.
A Higher-Order Construct is the fancy term that some authors use to refer to a special class
of LVs: constructs that involve more than one dimension. Each of the Higher-Order Con-
struct Models types is characterized by different relationships between the Higher-Order
Constructs and the LVs: the reflective relationship and the formative relationship. There
are four main types of Higher-Order Construct Models discussed in the extant literature
(Wetzels et al. 2009) and used in applications (Johnson et al. 2012). These types of model
have two elements: the Higher-Order Construct, which captures the more abstract entity,
and the Lower-Order Construct which captures sub-dimensions of the abstract entity.
Different approaches have been developed and proposed in the literature: the Repeated
Indicator Approach (Lohmöller 2013); the Two Step Approach (Sanchez 2013); the Mixed
Two Step Approach (Cataldo et al. 2017); and the PLS Components Regression Approach
(Cataldo et al. 2017).
The Repeated Indicators Approach is the most popular approach when estimating
Higher-Order Constructs in a PLS-PM (Wilson 2009). The procedure consists of taking the

13
 R. Cataldo et al.

indicators of the Lower-Order Constructs and using them as the MVs of the Higher-Order
LV.
The Two Step Approach is divided in two phases: in the first step the LV scores of the
Lower-Order Constructs are computed without the Higher-Order Construct (Rajala and
Westerlund 2010); then, in the second step, the PLS-PM analysis is performed using the
computed scores as indicators of the Higher-Order Constructs. The implementation is not
performed through a single PLS run; this implies that any Second-Order Construct, inves-
tigated in stage two, is not taken into account when estimating LV scores in stage one.
Sanchez (2013) suggests this way of computing scores for the LVs of Lower-Order: we
can obtain a score for a First-Order Construct by taking the first principal component of its
indicators. Next, the PCA scores of the Lower-Order Constructs are subsequently used as
indicators for the Higher-Order Construct in a separate PLS path model.
These two approaches present two important limitations related to components of each
block: the first is that only one component is chosen for each block; the second is that this
component has a strong representative power but a weak predictive power in the analysis of
the Higher-Order Construct. For these reasons, in order to overcome these two drawbacks,
two alternative methods to estimate the Second-Order Constructs are proposed (Cataldo
et al. 2017). In particular, in order to resolve the issue related to the predictive power of
the component for each First-Order Construct, the Mixed Two Step Approach is proposed
and, regarding the choice of the number of components for each block, the Partial Least
Squares Component Regression Approach is proposed.
The Mixed Two Step Approach begins with the implementation of the PLS-PM in the
case of the Repeated Indicators Approach. So, initially, because the Second-Order Con-
struct has no MVs of its own, we consider it as formed of all the MVs of the First-Order
Constructs. Starting from this structure, a PLS-PM algorithm is performed in such a way
as to obtain the scores of each block. Once the scores for the blocks have been obtained,
these will be the MVs of the Second-Order Construct and the PLS-PM algorithm can be
implemented. Compared to the Repeated Indicator Approach, this method uses the compo-
nents that seems to be the best representative of its block and, at the same time, has the best
predictive power on the Second-Order LV.
The PLS Component Regression Approach is proposed in order to overcome the prob-
lem related to the number of components of the First-Order Constructs, giving the pos-
sibility of choosing the number of components to be extracted manually or according to a
criterion. In addition, since the aim of PLS-PM is to estimate the relationships between the
LVs, this approach provides components that are at the same time representative of their
blocks and predictive of the Second-Order Construct. In the first step of this approach, the
PLS-Regression (Tenenhaus 1998) for each block of the First-Order is performed, so as to
obtain h components for each block. Next, these h components will represent MVs of the
Second-Order Construct. Once the PLS-Regression Components are assigned as indicators
of the Second-Order Construct, the PLS-PM algorithm can be implemented. In this way,
PLS-Regression for each block of the First-Order is performed, so as to obtain h compo-
nents for each block. Next, these h components will represent MVs of the Second-Order
Construct For a detailed discussion and step-by-step illustration of these approaches, see
Cataldo et al. (2017) and Hair et al. (2014). In this work the Higher-Order PLS-PM model
has been estimated with the PLS Component Regression Approach (Cataldo et al. 2017).
It has been demonstrated through a simulation study that this approach produce less biased
and more stable parameter estimates than the approaches known in the literature (Cataldo
et  al. 2017). Moreover, the “plspm” package in the R statistical software (Sanchez and
Trinchera 2012; Sanchez 2013) has been used.

13
Methodological PLS‑PM Framework for SDGs System

5 A Methodological PLS‑PM Framework for SDGs System

A sound methodological framework is the starting point in constructing CIs. The frame-
work should clearly define the phenomenon to be measured and its sub-components,
selecting individual indicators and weights that reflect their relative importance and the
dimensions of the overall composite. This process should ideally be based on what is desir-
able to measure and not on which indicators are available (Joint Research Centre-European
Commission 2008). In a framework, multi-dimensional concepts can be divided into sev-
eral sub-groups. These sub-groups need not be (statistically) independent of each other,
and existing linkages should be described theoretically or empirically to the greatest extent
possible. According to PLS-PM approach, it is possible to define the SDGs as a multi-
dimensional LVs not measurable directly and related to its single indicators or MVs by
a formative relationship. The PLS-PM approach has enjoyed increasing popularity as a
key multivariate analysis method in various research disciplines. It allows you to estimate
causal relationships, defined according to a theoretical model linking two or more latent
complex concepts, each measured through a number of observable indicators. The Global
Sustainability model is reported in Fig. 2, that gives an overview of the approach used in
the current research.
Global Sustainability is, here, conceived as a Third-Order Construct affecting Second-
Order dimensions, which in turn shape First-Order LVs underlying specific aspects of
the Second-Order dimensions. The study focuses on a formative-formative measurement
model, a model resulting from the combination of formative lower order and formative

Fig. 2  The SDGs theoretical framework

13
 R. Cataldo et al.

Table 1  SDGs and the three Area Goals

areas
Social Area
Goal 1 No Poverty
Goal 2 Zero Hunger
Goal 3 Good Health and Well-Being
Goal 4 Quality Education
Goal 5 Gender Equality
Goal 6 Clean Water and Sanitation
Economic Area
Goal 7 Affordable and Clean Energy
Goal 8 Decent Work and Economic Growth
Goal 9 Industry, Innovation and Infrastructure
Goal 10 Reduced Inequalities
Goal 11 Sustainable Cities and Communities
Goal 12 Responsible Production and Consumption
Environment Area
Goal 13 Climate Action
Goal 14 Life Below Water
Goal 15 Life On Land
Goal 16 Peace, Justice and Strong Institution
Goal 17 Partnerships for the Goals

higher order constructs. This hierarchical structure allows us to synthesize individual indi-
cators into single indexes in order to construct CIs at a global and a partial level. Moreover,
we analyze the relationships among the different goals and areas as well as the effects of
these on Global Index, in order to search for strongly influential factors. In Table  1, the
Goals for each area are reported.4 The environmental dimension of sustainable develop-
ment is measured with indices ranging between air quality, water quality, waste manage-
ment, greenhouse gas emissions and various ecological indicators. The social dimension
was the next most represented, with the widest variation in the selection of indicators. The
economic dimension is generally measured by the capital income, including income distri-
bution and public debt. There is no consensus or standard between the indices on the indi-
cators used to measure each of the three dimensions of sustainable development (Strezov
et al. 2017).
In the literature there is another SDGs classification made by the Overseas Develop-
ment Institute’s report (Nicolai et al. 2015). It classifies the SDGs into: reform, revolution
and reversal. SDGs classified at reform levels are ending extreme poverty, strengthening
economic growth in least-developed countries and halting deforestation (SDGs 1, 8, 15).
Goals that require progress by multiples of current rates were categorized as revolution.
Nine targets belong to this group: ending hunger, reducing maternal mortality, secondary
school completion, ending child marriage, access to sanitation, access to energy (electric-
ity), industrialization in less developed countries, reducing deaths and domestic resource

4
 For a detailed description of the individual Goals, refer to the site “Sustainable Development Goals”
(www.un.org/susta​inabl​edeve​lopme​nt/devel​opmen​t-agend​a/).

13
Methodological PLS‑PM Framework for SDGs System

mobilization (SDGs 2–7, 9, 16, 17). Targets classified under reversals are moving in the
opposite direction and require a reversal of current trends. They include inequality, slum
populations, climate change, waste management and marine (reef) conservation. (SDGs
10–14). Nicolai and others are optimistic that the least-developed countries are halfway
towards their 2030 targets of ending extreme poverty, economic growth in least-developed
countries and halting deforestation (Swain 2018).
Clearly the ability of a CIs to represent multidimensional concepts largely depends on
the quality and accuracy of its components. Missing data is a major problem in developing
countries. For instance, even for well reported variables like child mortality, 136 of the 161
countries track data on this goal (Rodriguez-Takeuchi 2014). About two-thirds of the 75
countries accounting for more than 95% of all maternal, newborn and child deaths do not
have records of births and deaths, whereas twenty-six countries have no data on child mor-
tality since 2009 (Stuart et al. 2015). To facilitate the implementation of the global indica-
tor framework, all indicators are classified by the IAEG-SDGs into three tiers based on
their level of methodological development and the availability of data at the global level, as
follows:5

• Tier 1: Indicator is conceptually clear, has an internationally established methodology

and standards are available, and data are regularly produced by countries for at least 50
per cent of countries and of the population in every region where the indicator is rel-
evant.
• Tier 2: Indicator is conceptually clear, has an internationally established methodology
and standards are available, but data are not regularly produced by countries.
• Tier 3: No internationally established methodology or standards are yet available for the
indicator, but methodology/standards are being (or will be) developed or tested.

Another important problem is concerning the high correlation of Goal’s indicators to

other Goals. Pradhan et al. (2017) highlight the existence of typically more synergies than
trade-offs within and among the SDGs in most countries. This indicates a strong founda-
tion for successfully implementation of the SDG agenda. The large fraction of synergies
within each SDG is consistent with previously highlighted linkages between some indica-
tors. An indicator may apply to more than one theme, as for example with “proportion of
population with access to safe drinking water”, which has primary links to poverty and
health. This indicator is also useful for measuring the impact of regulating and governing
water utilities. Moreover, as domestic fresh water is the major source of drinking water
for most countries, the indicator provides information on availability and use of water
resources and for the availability of infrastructure in utilities (Pradhan et al. 2017). Accord-
ing some works, all SDGs need to act as a system of interacting cogwheels that together
move the global system into the safe and just operating space. No SDG will do that indi-
vidually, and the whole SDGs should not be seen as an additive structure but as a system of
synergistic re-enforcement (Pradhan et al. 2017).

5
  The latest changes reflect the decisions made during the IAEG-SDG WebEx Meeting in January 2019.
The tier classification of many indicators is expected to change as methodologies are developed and data
availability increases. The review of reclassification requests by the IAEG-SDGs will occur when require-
ments are met at the two physical meetings and via WebEx meetings throughout the year, based on a calen-
dar developed by the Group.

13
 R. Cataldo et al.

6 A Case Study of SDGs of European Countries

To illustrate the importance of the proposal approach, in this section it is reported the
analysis of SGDs framework. This section reports the main results of the Higher-Order
PLS-PM analysis, with consideration on the evaluation of the measurement model, on the
structural model and, finally, on the Importance-Performance Matrix Analysis. Also, the
analysis of principal components has been carried out to compare this approach with the
PLS-PM and to underline the advantages of the last one.

6.1 The Data and the Model

The official data derive from the database of United Nation “Sustainable Development
Goals”.6 The analysis was developed with reference to the European Community countries,
so the number of countries considered is 28. The data are updated to the last three years
(2015-2017), in particular the data are related to the year 2017;7 but for some variables,
because of the lack of data for that year, the previous years of this triennium were taken
as the reference. Considering that each EIs have different units and values, for comparison
purposes all units are first normalized to a value between 0 and 1, where 0 was assigned
to the least sustainable while 1 was the value assigned to the most sustainable country for
each EIs. The Eq. (1) is the simple expression used for normalization of each variable:
X − min(X)
Z= (1)
max(X) − min(X)
where X is the sustainability index.
In the case of some EIs, where the larger value meant lower sustainability, the normali-
zation had reverse order, so that the country with the highest EI received a value of 0,
while the country with the lowest EI had a value of 1. The “Global indicator framework
for the Sustainable Development Goals and targets of the 2030 Agenda for Sustainable
Development” of UN 2017 (United Nations Statistical Commission 2017) lists all indica-
tors used to calculate each of the selected SDGs. The indicators were assigned to one of
the three dimensions of sustainable development: the economic, environmental and social
dimensions.
In the analysis three Goals are not considered because they are not available in the three-
year period considered. For the social area, the Goal not available is Goal 1 “No Poverty”;
for the economic area, the Goals not available are “Reduced Inequalities” and “Responsi-
ble Production and Consumption”. The scheme of theoretical model that express the rela-
tionship among indicators is shown in Fig. 2 in Sect. 5. The study focuses on a formative-
formative measurement model, a model resulting from the combination of formative lower
order and formative higher order constructs. Sanchez “plspm” package in the R program-
ming language (Sanchez and Trinchera 2012) was used in order to perform PLS-PM analy-
sis involving formative indicators and the centroid scheme for the inner estimation.

6
  https​://unsta​ts.un.org/sdgs/indic​ators​/datab​ase/.
7
  Data presented in this study were mined in September 2018.

13
Methodological PLS‑PM Framework for SDGs System

Table 2  Correlation among Goals of Social Area

Zero Hunger Good Health Quality Gender Equality Clean Water Social Area
Educa-
tion

Zero Hunger 1.00

Good Health 0.82 1,00
Quality Education 0.89 0.86 1.00
Gender Equality 0.55 0.52 0.69 1.00
Clean Water 0.87 0.79 0.97 0.67 1.00
Social Area 0.92 0.89 0.98 0.74 0.96 1.00

Table 3  Correlation among Goals of Economic Area

Clean Energy Economic Infrastructure Sustainable Economic Area
Growth Cities

Clean Energy 1.00

Economic Growth 0.90 1.00
Infrastructure 0.54 0.56 1.00
Sustainable Cities 0.74 0.67 0.46 1.00
Economic Area 0.96 0.94 0.46 0.79 1.00

Table 4  Correlation among Goals of Environmental Area

Climate Action Life Below Life On Land Environ-
Water mental
Area

Climate Action 1.00

Life Below Water 0.99 1.00
Life On Land 0.99 0.99 1.00
Environmental Area 0.99 0.99 0.99 1.00

6.2 The Main Results

Tables  2,  3 and  4 show the correlation matrix of Sustainable Goals for each area and
between each of them and their own area.
As we can see from the tables, for each area, the Goals are strongly related to each
other and strongly linked to the area indicator. Only “Gender Equality” for Social area
and “Infrastructure” for Economic area are weakly correlated with the other goals of the
same area. In general, this means that the Goals of different dimensions of sustainability
are, with different intensity, concordant among themselves. Before applying the PLS-PM
approach, an exploratory PCA has been performed in order to analyse the relationship of
elementary indicators for each Goal. In Fig. 3 the scree-plot (a) and the correlation circle
(b) of PCA of Goal 1 “No Poverty” are shown.
From the scree-plot examination, it is evident that the first one component explain
the variability of the MVs but it is not sufficient to synthesize the block of MVs of the

13
 R. Cataldo et al.

Fig. 3  Scree-plot and correlation circle of PCA of Goal 1 “No Poverty”

Table 5  Godness of Fit GoF Relative GoF GoF outer model GoF inner model

0.656 0.762 0.718 0.769

Table 6  Overall model quality

Type R2 Communalities Redundancies

Social Area Endogenous 0.934 0.818

Economic Area Endogenous 0.916 0.600
Environment Area Endogenous 0.895 0.998
Global Sustainability Endogenous 0.991 0.636

Goal. The most of the variability of the MVs of Goal 1 (67.59%) can be explained by the
first two axis. By projecting the original indicators in the plane of the first two principal
components, the circle of correlations is obtained, where each indicator of Goal is rep-
resented by a point with coordinates equal to the two coefficients of correlation with the
first and second factor. Note that the first factor is strongly correlated with one indicator
of 5, whereas the second one represents the composite index of the other three indicators.
The graph shows that an indicator is in the center of the circle, this means that it is not
explained by the first two factors. The same analysis has been carried out for all 17 Goals
and, more or less, the same results came out. In this way it is possible to have several main
components representing the different goals but they can be analyzed separately from each
other. This means that for this type of sustainability data the PCA approach can be used
for single each block, but it is unable to summarize the overall structure of the data. For
this reason, since the goal of this work is not only to synthesize the dataset, but to obtain a
composite indicator capable of representing this phenomenon, the PLS-PM approach has
been applied and the results are presented below. First, overall model quality indices were

13
Methodological PLS‑PM Framework for SDGs System

Table 7  Path coefficients linking the first-order constructs to the three areas at second-order construct and
path coefficients linking the three areas to Global Sustainability CI
LVs Path coefficient SE CI (95%)

Zero Hunger → Social Area 0.417 0.025 [0.394; 0.465]

Good Health → Social Area 0.408 0.033 [0.332; 0.429]
Quality Education → Social Area 0.441 0.067 [0.277; 0.465]
Gender Equality → Social Area 0.375 0.018 [0.358; 0.409]
Clean Water and Sanitation → Social Area 0.458 0.062 [0.424; 0.620]
Clean Energy → Economic Area 0.505 0.071 [0.390; 0.590]
Economic Growth → Economic Area 0.569 0.086 [0.455; 0.704]
Infrastructure → Economic Area 0.523 0.032 [0.368; 0.651]
Sustainable Cities → Economic Area 0.465 0.034 [0.392; 0.499]
Climate Action → Environmental Area 0.537 0.080 [0.532; 0.741]
Life Below Water → Environmental Area 0.529 0.023 [0.492; 0.567]
Life On Land → Environmental Area 0.536 0.013 [0.531; 0.569]
Social Area → Global Sustainability 0.312 0.0939 [0.288; 0.683]
Economic Area → Global Sustainability 0.438 0.0658 [0.356; 0.760]
Environmental Area → Global Sustainability 0.241 0.0120 [0.162; 0.398]

shown. The coefficients presented in Table 5 showed that MVs and LVs are well predicted
by the PLS-PM framework. The prediction performance of the PLS-PM is high for both
the outer, inner and global models. As concerning the goodness of fit, there is no overall
fit index in PLS-PM. Nevertheless, a global criterion of Goodness of Fit (GoF) has been
proposed by Tenenhaus et al. (2005). Both GoF and relative GoF reflected the high quality
of the constructs for both outer and inner model (Table 5).
Table 6 reports the main indices indicating the overall model quality: R2 coefficient, the
Communality and Redundancy indices. The R2 coefficient shows that endogenous LVs are
well predicted by the explanatory LVs, while the values of communality index (another
index that measures the goodness of the models of measurement) are appreciably higher
for all blocks (value of 0.50 indicates a sufficient degree of construct validity).
In order to provide an interpretation of the results, we analyse all the path coefficients
linking the first-order constructs to the three areas at second-order construct and path coef-
ficients linking these areas to Global Sustainability CI through Table 7.
The path coefficients are all positive and significant. All dimensions of a first-order con-
struct are important for their areas, respectively. In particular the dimensions of “Quality
Education” and “Clean Water and Sanitation” prove to be most influential among all the
factors of the social area; the most influential dimension of economic area is “Economic
Growth”, while the “Climate Action” has a slightly higher path coefficient than the other
dimensions of the Environmental Area. The economic area has a significant role in the
Global Sustainability CI estimation, with a path of 0.438, following by the social area, with
a path of 0.312. The environmental area has a lower impact to estimation Global Sustain-
ability CI, with a path of 0.241.
The importance played by Goals on the outcome domain can be jointly analysed with
the composites averages. Figure 4 shows the importance and the performance (means and
path-coefficients) of the 12 Goals in order to estimate the different areas of Global Sustain-
ability CI.

13
 R. Cataldo et al.

Fig. 4  Importance-performance matrix analysis

The Importance-Performance Matrix Analysis shows that the Education, Economic growth
and Climate change are respectively located in a critical area of the three areas respectively,
because the critical area contains domains that have a high impact on the outcome but that
register low average values; they prove to be especially critical for the Global Sustainability
CI estimation. Instead “Zero Hunger”, “Good Health”, “Sustainable Cities” and “Life Below
Water” are located in a area of the monitoring. Note how the “Gender Equality” and “Infra-
structure” are placed in the area to improve.
This application can allow only a provisional and preliminary study. Because of the miss-
ing data, some Goals were not considered in the analysis so it is incomplete and partial. The
results presented only want to be example to show the proposed instrument and highlight its
potential. According the United Nations site, starting 2019 major updates are expected to be
released in July, September and December, so maybe it will be possible have Goals that in this
analysis were not considered due to the missing EIs. Moreover, in this analysis we consider
all indicators used on 28 European countries on the same level, but we should avoid the sus-
tainability indices/indicators tend to over- or under-estimate developed countries as compared
with developing countries.

13
Methodological PLS‑PM Framework for SDGs System

7 Final Remarks

The development of a quality framework for composite indicators is not an easy task. In
fact, the overall quality of the composite indicator depends on several aspects, related both
to the quality of elementary data used to build the indicator and the soundness of the pro-
cedures used in its construction. Quality is usually defined as “fitness for use” in terms of
user needs. The most important quality characteristics depend on user perspectives, needs
and priorities, which vary across user-groups (Joint Research Centre-European Commis-
sion 2008).
Lack of limited availability of the data is a major constraint for quantifying and moni-
toring SDGs. While data on economic indicators is widely available for most countries,
data on environmental and social indicators is incomplete and of poor quality. Further-
more SDGs have often been criticized for a lack of underlying comprehensive theoreti-
cal framework. Finally, SDGs are long-term development agenda and have the potential to
be exposed to unforeseen positive and negative shocks. There have been many and varied
approaches, such as simple arithmetic mean of the original indicators, PCA or FA to meas-
uring sustainable development, based on simple aggregation techniques and easily calcu-
lated. According to many authors, such as Mazziotta and Pareto (2019), the construction of
index for measuring this multidimensional phenomenon cannot solve simply by using PCA
or FA, since they are typically used for a reflective approach and in particular a PCA based
index accounts for a limited part of the total variance. PLS-PM represents an important
breakthrough with respect to these traditional aggregation methods. The choice of using the
PLS-PM is useful for several reasons, particularly this approach should be used in order to
not only reduce dimension but find relations between composite indicators and their blocks
(Tenenhaus et al. 2005). It should be used for constructing a system of composite indica-
tors, help to identify critical indicators, to construct ranking of countries and to provide an
interpretation of the results. PLS-PM allows you to estimate causal relationships, defined
according to a theoretical model linking two or more latent complex concepts, each meas-
ured through a number of observable indicators. This approach, as compared to other mul-
tivariate techniques, examines simultaneous a series of dependence relationships, using a
single statistical approach to test the full scope of projected relations (Hair et al. 2009). The
basic idea is that the complexity inside a system can be studied by taking into account the
entirety of the causal relationships among the LVs, each measured by several MVs (Lauro
et al. 2018). It is absolutely essential to define the theoretical framework with an appropri-
ate measurement model. This paradigm should always be considered when the objective of
the research is to measure a multidimensional phenomenon through composite indices. The
PLS-PM provides researchers with much more flexibility as it enables using both forma-
tive and reflective measurement models, providing a more nuanced testing of theoretical
concepts (Hair et al. 2011). This present work would be a first propose of a methodological
framework that can be useful for creating a Sustainability CI in terms of decision-mak-
ing. As a first approach, we analyse the PLS-PM approach without considering synergies
and trade-offs within an SDGs. We propose Higher-Order PLS-PM approach for studying
existing SDGs indicators and demonstrate how it could help you to define the methodo-
logical framework for SDGs system in order to provide a better measure of this complex
multidimensional social phenomenon. A Higher-Order PLS-PM approach was adopted in
a preliminary attempt to investigate relations between different analytical dimensions of
the phenomenon. The results of the analysis suggest that economic area should be the most
important in order to measure sustainability but it cannot be considered the only factor to

13
 R. Cataldo et al.

be analyzed, as the other two dimensions also have their relative impact. However, our
preliminary research can offer only suggestions on this aspect and further investigation are
definitively needed, especially because some Goals were not considered in the analysis due
to missing data. Starting from this methodological PS-PM framework for SDGs system
and works of Pradhan et al. (2017), future works will be to implement and develop these
models also considering synergies within SDGs and understand if and, if yes, how these
synergies modify the relationships among the Goals and the areas. Moreover, we would
like to insert in the analysis qualitative information concerning European community coun-
tries in order to understand if there are differences among them relating to this issue. The
successful implementation of the good methodological framework for SDGs system is the
only way forward to address the global sustainability challenge of ensuring human well-
being, economic prosperity, and environmental protection. Infact,“what we measure affects
what we do; and if our measurements are flawed, decisions may be distorted” (Stiglitz et al.
2017).

References
Anand, S., & Sen, A. (2000). Human development and economic sustainability. World Development, 28,
2029–2049.
Bali Swain, R., & Wallentin, F.-Y. (2017). The sustainable development quagmire: Quantifying the sustain-
able development goals. Unpublished draft, Södertörn University & Stockholm School of Economics.
Bettencourt, L. M.-A., & Kaur, J. (2011). Evolution and structure of sustainability science. Proceedings of
the National Academy of Sciences, 108(49), 19540–19545.
Bilbao-Ubillos, J. (2013). The limits of Human Development Index: The complementary role of economic
and social cohesion, development strategies and sustainability. Sustainable Development, 21(6),
400–412.
Bithas, K. P., & Christofakis, M. (2006). Environmentally sustainable cities. Critical review and operational
conditions. Sustainable Development, 14(3), 177–189.
Böhringer, C., & Jochem, P. E.-P. (2007). Measuring the immeasurable—A survey of sustainability indices.
Ecological Economics, 63(1), 1–8.
Cataldo, R., Grassia, M. G., Lauro, N. C., & Marino, M. (2017). Developments in higher-order PLS-PM for
the building of a system of composite indicators. Quality & Quantity, 51(2), 657–674.
Chin, W. W. (1998). The partial least squares approach to structural equation modeling. In G. A. Mar-
coulides (Ed.), Modern business research methods (pp. 295–336). Mahwah, NJ: Lawrence Erlbaum
Associates.
Daly, H., Cobb, Jr. J. B., & Cobb, J. B. (1994). For the common good: Redirecting the economy toward com-
munity, the environment, and a sustainable future. Boston: Beacon Press.
Davino, C., Dolce, P., Taralli, S., & Vinzi Esposito, V. (2018). A quantile composite-indicator approach for
the measurement of equitable and sustainable well-being: A case study of the italian provinces. Social
Indicators Research, 136(3), 999–1029.
De Smedt, M., Giovannini, E., & Radermacher, V. (2018). Measuring sustainability, for good measure
advancing research on well-being metrics beyond GDP: Advancing research on well-being metrics
beyond GDP (Vol. 241). Paris: OECD Publishing.
Dunteman, G. H. (1989). Principal components analysis (Vol. 69). Sage.
Estoque, R. C., & Murayama, Y. (2014). Social–ecological status index: A preliminary study of its structural
composition and application. Ecological Indicators, 43, 183–194.
Evans, A., Strezov, V., & Evans, T., et al. (2015). Measuring tools for quantifying sustainable development.
Rome: European Center of Sustainable Development.
Fabbrizzi, S., Maggino, F., Marinelli, N., Menghini, S., Ricci, C., & Sacchelli, S. (2017). Sustainability: A
quantitative discourse analysis. Rivista di studi sulla sostenibilità.
Gallopín, G. C. (2003). A systems approach to sustainability and sustainable development (Vol. 64). United
Nations Publications.
Gan, X., Fernandez, I. C., Guo, J., Wilson, M., Zhao, Y., Zhou, B., et al. (2017). When to use what: Methods
for weighting and aggregating sustainability indicators. Ecological Indicators, 81, 491–502.

13
Methodological PLS‑PM Framework for SDGs System

Hair, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. (2009). Multivariate data analysis. Upper Saddle
River, NJ: Pearson Prentice Hall.
Hair, J. F., Hult, T., Ringle, C. M., & Sarstedt, M. (2014). A primer on partial least squares structural equa-
tion modeling (PLS-SEM). Thousand Oaks, CA: Sage.
Hair, J. F., Ringle, C. M., & Sarstedt, M. (2011). PLS-SEM: Indeed a silver bullet. Journal of Marketing
Theory and Practice, 19, 139–151.
Hák, T., Janoušková, S., & Moldan, B. (2016). Sustainable development goals: A need for relevant indica-
tors. Ecological Indicators, 60, 565–573.
Henseler, J., Ringle, C. M., & Sinkovics, R. R. (2009). The use of partial least squares path modeling in
international marketing. In New challenges to international marketing (pp. 277–319). Emerald Group
Publishing Limited.
Hermans, E., Van den Bossche, F., & Wets, G. (2008). Combining road safety information in a performance
index. Accident Analysis & Prevention, 40(4), 1337–1344.
Hopwood, B., Mellor, M., & O’Brien, G. (2005). Sustainable development: Mapping different approaches.
Sustainable Development, 13(1), 38–52.
ICSU, ISSC. (2015). Review of the sustainable development goals: The science perspective. Paris: Interna-
tional Council for Science (ICSU).
International Union for Conservation of Nature and World Wildlife Fund. (1980). World conservation strat-
egy: Living resource conservation for sustainable development. Gland: IUCN.
Joint Research Centre-European Commission, et  al. (2008). Handbook on constructing composite indica-
tors: Methodology and user guide. Paris: OECD Publishing.
Jarvis, D., MacKenzie, S., & Podsakoff, P. (2003). A critical review of construct indicators and measure-
ment model misspecification in marketing and consumer research. Journal of Consumer Research,
30(3), 199–218.
Johnson, R. E., Rosen, C. C., Chang, C. H., Djurdjevic, E., & Taing, M. U. (2012). Recommendations for
improving the construct clarity of higher-order multidimensional constructs. Human Resource Man-
agement Review, 22(2), 67–72.
Kates, R. W. (2011). What kind of a science is sustainability science? Proceedings of the National Academy
of Sciences, 108(49), 19449–19450.
Kyoto Protocol. (1997). United Nations framework convention on climate change (vol. 19). Kyoto Protocol,
Kyoto.
Lauro, C. N., Grassia, M. G., & Cataldo, R. (2018). Model based composite indicators: New developments
in partial least squares-path modeling for the building of different types of composite indicators. Social
Indicators Research, 135(2), 421–455.
Lohmöller, J. B. (2013). Latent variable path modeling with partial least squares. Berlin: Springer.
Mazziotta, M., & Pareto, A. (2019). Use and misuse of PCA for measuring well-being. Social Indicators
Research, 142(2), 451–476.
Michalos, A. C. (2014). Encyclopedia of quality of life and well-being research. Dordrecht: Springer.
Moran, D. D., Wackernagel, M., Kitzes, J. A., Goldfinger, S. H., & Boutaud, A. (2008). Measuring sustain-
able development—Nation by nation. Ecological Economics, 64(3), 470–474.
Mori, K., & Christodoulou, A. (2012). Review of sustainability indices and indicators: Towards a new City
Sustainability Index (CSI). Environmental Impact Assessment Review, 32(1), 94–106.
Nardo, M., & Saisana, M. (2008). OECD/JRC handbook on constructing composite indicators. Putting the-
ory into practice. Brussels: European Commission—Joint Research Centre; Institute for the Protection
and Security of the Citizen Unit of Econometrics and Applied Statistics.
Neumayer, E. (2010). Weak versus strong sustainability: Exploring the limits of two opposing paradigms
(3rd ed.). Cheltenham: Edward Elgar Publishing.
Neumayer, E. (2007). Sustainability and well-being indicators. In Human well-being (pp. 193–213). Lon-
don: Palgrave Macmillan.
Nicolai, S., Hoy, C., Berliner, T., & Aedy, T. (2015). Projecting progress: Reaching the SDGs by 2030. Lon-
don: Overseas Development Institute.
Pradhan, P., Costa, L., Rybski, D., Lucht, W., & Kropp, J. P. (2017). A systematic study of sustainable
development Earth’s future. Wiley Online Library, 5(11), 1169–1179.
Rajala, R., & Westerlund, M. (2010). Antecedents to consumers’ acceptance of mobile advertisements: A
hierarchical construct PLS structural equation model. In XLIIIth Hawaii International Conference on
Systems Sciences (HICSS).
Rodriguez-Takeuchi, L. (2014). The ‘X-file’ and the need for civil registration and vital statistics systems
post-2015. London: ODI.
Sachs, J. D. (2012). From millennium development goals to sustainable development goals. The Lancet,
379(9832), 2206–2211.

13
 R. Cataldo et al.

Sachs, J., Schmidt-Traub, G., Kroll, C., Durand-Delacre, D., & Teksoz, K. (2016). SDG index & dash-
boards: A global report. Gütersloh: Bertelsmann Stiftung.
Salzman, J. (2003). Methodological choices encountered in the construction of composite indices of eco-
nomic and social well-being. Ottawa, ON: Ottawa Center for the Study of Living Standards.
Sanchez, G. (2013). PLS path modeling with R (Vol. 383). Berkeley: Trowchez Editions.
Sanchez, G., & Trinchera, L. (2012). plspm: Partial Least Squares data analysis methods. R package ver-
sion 0.2-2. http://CRAN.R-proje​ct.org/packa​ge=plspm​.
Smith, L. I. (2002). A tutorial on principal components analysis.
Solow, R. M. (1994). Perspectives on growth theory. Journal of Economic Perspectives, 8(1), 45–54.
Spaiser, V., Ranganathan, S., Swain, R. B., & Sumpter, D. J. T. (2017). The sustainable development oxy-
moron: Quantifying and modelling the incompatibility of sustainable development goals. International
Journal of Sustainable Development & World Ecology, 24(6), 457–470.
Stiglitz, J. E., Sen, A., & Fitoussi, J. P. (2017). Report by the commission on the measurement of economic
performance and social progress.
Strezov, V., Evans, A., & Evans, T. J. (2017). Assessment of the economic, social and environmental dimen-
sions of the indicators for sustainable development. Sustainable Development, 25(3), 242–253.
Stuart, E., Samman, E., Avis, W., & Berliner, T. (2015). The data revolution: Finding the missing millions.
London: Overseas Development Institute.
Swain, R. B. (2018). A critical analysis of the sustainable development goals. In Handbook of sustainability
science and research (pp. 341–355). Cham: Springer.
Tenenhaus, M. (1998). La régression PLS: théorie et pratique. Editions technip.
Tenenhaus, M., Esposito Vinzi, V., Chatelin, Y. M., & Lauro, C. N. (2005). PLS path modeling. Computa-
tional Statistics and Data Analysis, 48(1), 159–205.
Trinchera, L., Russolillo, G., & Lauro, C. N. (2008). Using categorical variables in PLS path modeling to
build system of composite indicators. Statistica Applicata, 20(2), 309–330.
United Nations. (2009). Millennium Development Goals Report 2009 (Includes the 2009 Progress Chart).
United Nations Publications.
United Nations. (2015a). Transforming our world: The 2030 agenda for sustainable development. General
Assembly 70 session.
United Nations. (2015b). United Nations Conference on Sustainable Development, Rio+20.
United Nations Development Programme. (1992). Earth Summit 1992: The United Nations Conference on
Environment and Development, Rio de Janeiro, 1992.
United Nations Development Programme. (2010). Human development report 2010. Palgrave Macmillan.
United Nations Statistical Commission, et  al. (2017). Global indicator framework for the Sustainable
Development Goals and targets of the 2030 Agenda for Sustainable Development. UN Resolution A/
RES/71/313.
WCED, SPECIAL WORKING SESSION. (1987). World commission on environment and development.
Our common future (vol. 17, pp. 1–91). Oxford University Press, London.
Wetzels, M., Odekerken-Schröder, G., & van Oppen, C. (2009). Using PLS path modeling for assessing
hierarchical construct models: Guidelines and empirical illustration. MIS Quarterly, 33(1), 177–195.
Wilson, B. (2009). Using PLS to investigate interaction effects between higher order brand constructs. In
Esposito, Vinzi, V., Chin, W. W., Henseler, J., Wang, H. (Eds) Handbook of partial least squares: Con-
cepts, methods and applications in marketing and related fields. Berlin: Springer.

Publisher’s Note  Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.

Affiliations

Rosanna Cataldo1   · Corrado Crocetta2 · Maria Gabriella Grassia1 ·

Natale Carlo Lauro3 · Marina Marino1 · Viktoriya Voytsekhovska4
Corrado Crocetta
corrado.crocetta@unifg.it
Maria Gabriella Grassia
mgrassia@unina.it

13
Methodological PLS‑PM Framework for SDGs System

Natale Carlo Lauro

lauro@unina.it
Marina Marino
mari@unina.it
Viktoriya Voytsekhovska
viktoriia.v.voitsekhovska@lpnu.ua
1
Department of Social Sciences, University of Naples “Federico II”, Vico Monte della Pietà, 1,
80138 Naples, Italy
2
Department of Economics, University of Foggia, Largo Papa Giovanni Paolo II, 71121 Foggia,
Italy
3
Department of Economic and Statistical Sciences, University of Naples “Federico II”, Monte S.
Angelo, 80126 Naples, Italy
4
Department of Economics of Enterprise, Lviv Polytechnic National University, Lviv, Ukraine

13