Escolar Documentos
Profissional Documentos
Cultura Documentos
PRISME Research Group, Universite de Sherbrooke, Faculte dadministration, 2500 Boulevard de lUniversite, Quebec, Canada J1K 2R1
Universite de Sherbrooke, Faculte dadministration, 2500 Boulevard de lUniversite, Quebec Canada J1K 2R1
A R T I C L E I N F O
A B S T R A C T
Article history:
Received 5 March 2014
Received in revised form 20 May 2014
Accepted 23 May 2014
Available online 3 June 2014
Many studies have demonstrated the impact of information technology (IT) on decision making but few
have used decision making process performance (DMPP) as a dependent variable. Our study proposes a
rich formative conceptualization of DMPP, a valid and reliable measure for this construct, and studies its
inuence on the quality of decision making. The results show that DMPP is a formative second-order
aggregate construct composed of procedural rationality, exhaustivity of the information analyzed,
openness of spirit, and effort. This study illustrates the importance of building proper denitions of
constructs and contributes to the development of shared meaning in IS.
2014 Elsevier B.V. All rights reserved.
Keywords:
Decision making process performance
Formative construct
Procedural rationality
Exhaustivity of information
Openness of spirit
Decision-making effort
1. Introduction
Information technologies (IT) have long been presented as
useful tools that support structured, semi-structured and unstructured decision making. For example, expert systems have been
associated with semi-structured decisions, enterprise resource
planning systems have been associated with structured decisions,
and business intelligence systems have been associated with
unstructured decisions. It is now widely acknowledged that
information systems contribute to the overall quality of decision
making in organizations.
Many studies have demonstrated the impact of IT on decision
making. For example, group decision support systems (GDSS) have
been shown to provide better access to electronic databases and to
facilitate the sharing of information in group decisions [30].
Executive information systems have been shown to play an
important role in gathering data for strategic decision making by
top executives [34]. In this respect, the use of a data warehouse
may result in superior performance compared to the use of either a
* Corresponding author. Tel.: +1 819 821 8000x62983; fax: +1 819 821 7934.
E-mail addresses: Manon.Guillemette@USherbrooke.ca (M.G. Guillemette),
Maxime.Laroche2@usherbrooke.ca (M. Laroche), Jean.Cadieux@USherbrooke.ca
(J. Cadieux).
1
Tel.: +1 819 821 8000x61925; fax: +1 819 821 7934.
http://dx.doi.org/10.1016/j.im.2014.05.012
0378-7206/ 2014 Elsevier B.V. All rights reserved.
the decision. In the business world as much as in research, decisionmaking process performance is an important prerequisite for a
quality decision, and a relationship appears to exist between these
two distinct constructs [33,3,21].
We found relatively few studies in our literature review that use
the decision making process performance construct, due in part to
the complexity of the analysis required but also because the
construct is not always properly conceptualized. As a result, few
reliable and valid measures have been proposed to accurately
measure this construct and those that have been proposed have
not been sufciently tested.
Consequently, this paper proposes a rich formative conceptualization of decision making process performance (DMPP) and
studies its inuence on the quality of decision making. First, we
conceptualize DMPP as a multidimensional construct in the form
of a formative second-order aggregate construct [26] composed, at
the rst-order level, of four reective dimensions: procedural
rationality, exhaustivity of the information consulted, openness of
spirit, and effort. We then develop a measure of this construct that
we validate with a sample of 198 respondents. Our results
demonstrate that the DMPP construct is, from a conceptual
perspective, a true formative construct with the above-mentioned
four dimensions and that it is also an important and signicant
antecedent of decision quality. Our study is a response to the call
made by other researchers to focus on evaluating decision
processes rather than the consequences of those decisions [21].
It proposes a reusable tool for researchers that is a valid measure of
this important dependent variable.
2. Theory development
It has been argued that managers spend most of their working
time communicating and making decisions. However, all decisions
are not the same, and different decisions may require that
managers adopt different decision making processes. Our study
focuses on important decisions for organizations that we call
high-impact decisions. These decisions are mostly ill-structured,
relatively complex, and important for the organization (tactical or
strategic decisions). Examples of high-impact decisions include the
decision to implement a new information technology, the decision
to invest in the development of a new product or to abandon
production of an existing one, and decisions regarding the
termination of a marketing campaign.
Such decisions are not usually made by a single individual in
complete isolation. Rather, decision makers are inuenced by other
people (top-managers, colleagues and employees), structures,
politics and culture. High-impact decisions are likely to involve the
use of information and knowledge in the decision making process
to cope with uncertainty that is intrinsically related to the
organizational context [11,20].
Past research has found that this type of decision is approached
more carefully and rationally by managers compared to less
important decisions. Recent research has shown that the degree of
structure of the decision making process is somewhat related to
the degree of adaptation required by the context. For example, in
planning a new strategic initiative, or in reaction to a new
uncertainty, decision makers tend to carefully reevaluate their
implementation plans and use a structured decision making
process to do so. However, in a crisis-like context, managers tend to
adopt a more straightforward approach [13,20]. Indeed, the
positive impact of high-quality analysis on the quality of decision
making is well documented [24,3]. The results show that highimpact decisions, such as tactical and strategic decisions [34] or
ethical decisions [15], in high-performing rms are taken in a more
rational and less intuitive and political way than some have
suggested [23]. Even if managers are under the inuence of
619
620
decision makers often choose solutions with which they are most
familiar, and do not always explore new avenues as part of their
decision making processes. Openness of spirit allows decision
makers to make decisions by going beyond their usual limitations.
Extant research associates openness of spirit with creativity,
dened as using ones imagination and intellectual abilities to
develop a new line of thinking. Generally, decision makers who
demonstrate openness of spirit in their decision-making processes
are more inclined to make decisions that take all of the issues into
account, and therefore engage in more effective decision making
processes [31].
Procedural rationality, exhaustivity of the information analyzed, effort and openness of spirit may be conceptualized as the
underlying dimensions of DMPP, the dimensions that are
conceptually distinct.
2.2. Formative conceptualization of DMPP (steps 35)
The third step in the conceptualization of a multidimensional
construct is deciding whether the construct is reective or
formative in nature. Because the DMPP construct combines the
four dimensions presented earlier in a way that creates meaning,
our construct can be qualied as formative. Taking this approach,
the DMPP construct is conceptualized as a Type II formative
construct [17]. More specically, the proposed model is reective
at the rst-order level and formative at the second-order level (see
Fig. 1).
We expressed the DMPP construct at the second-order level as a
formative model with four dimensions: procedural rationality,
exhaustivity of the information analyzed, effort, and openness of
spirit. As in any formative conceptualization, a change in the value
of one of the four dimensions will not necessarily affect one of the
other dimensions. For example, a decision maker who does not
engage in a rational process may still review large quantities of
information as part of his or her decision making process.
Consequently, there is causality, from each dimension to the
construct, underscoring, once again, the formative nature of the
construct under study [17].
At the rst-order level, we conceptualized each of these
dimensions as created from reective measures. The purpose of
the reective measures at the rst-order level is to reveal the
presence of phenomena related to the latent constructs. The main
feature of these reective items is that they are interchangeable.
For example, if a decision maker is not following a rational process,
all of the measures used for this dimension will be weak because
they act as reective measures. The rst-order, reective nature of
the measurement items is particularly appropriate because it
targets manifestations of their main indicators. Because each of
these rst-order dimensions has been used in the past as reective
and observable measures, reective items have already been
developed, and they are used and tested in this study.
Finally, the fourth and fth steps consist of determining the
nature of the construct (aggregate or prole), and then, if it is
aggregate, determining mathematical relationships between the
dimensions. It has been suggested that once a construct has several
dimensions of the same order, the dimensions can be aggregated
into an index [35]. One would expect that performance can be
interpreted in the same way because it generally consists of an
aggregation of measures rather than a single observable measure.
3. Methodology
A Q-Sort activity was carried out to establish the content
validity of the reective rst-order dimensions of our second-order
formative construct [25]. This methodology entailed reproducing
all of the questions of the rst-order dimension and asking a few
621
0;
0;
e1
e2
Q21
DMPP2
Q23
0;
e3
0;
DMPP
Formave
0;
0;
0;
0;
0;
0;
0;
0;
0;
e10
Q4.7
e9
Q4.6
e8
Q4.5
e7
Q4.4
e6
Q4.2
e5
Q4.1
e4
Q4.3
EF
EI
PR
0;
0;
OS
Q2.1
Q3.5
Q3.1
Q3.2
1
0;
1
0;
1
0;
e15
e14
e13
Q3.4 RQ3.3
1
0;
e12
1
0;
e11
Q1.4
Q1.3
Q1.1
Q1.2
Q1.5
1
0;
1
0;
1
0;
1
0;
1
0;
e20
e19
e18
e17
e16
e21
Q2.3
e22
Q2.5
e23
Q2.4
e24
e25
Q2.2
0;
0;
0;
0;
Reecve
0;
Indicator
References
# of items
PR
EI
EF
OS
Procedural rationality
Exhaustivity of information
Effort
Openness of spirit
[11,9,10]
[27,4]
[21,8,18]
[31]
7
5
5
5
622
Table 2
Analysis of indicators of DMPP.
PR
EI
EF
OS
Average of items
Standard dev.
Cronbachs alpha
VIF
5.4610
5.0071
5.7980
5.1859
0.72535
1.06182
0.75212
0.76960
0.783
0.834
0.811
0.688
1.674
1.135
1.520
1.659
4. Results
4.1. Estimating the measurement model
Table 2 presents various statistics relating the items to the
latent dimensions and relationships between the indices (variance
ination factors). The right part of the table presents correlation
values between the four indices, created by averaging the related
items at the reective rst-order level.
Cronbachs alpha was used to analyze the internal consistency
of the rst-order reective dimensions. As seen in Table 2, the
Cronbachs alpha values are all greater than or equal to 0.688,
which is slightly less than the usually accepted 0.7 threshold. This
e1
0.735
1
0.51
0.35
1
0.47
e3
0.193
0.255
EI
DMPP22
DMPP
DMPP
PR
0.382
1
0.56
0.69
0.53
e2
DMPP 1
0.208
0.817
Correlations
0.200
EF
OS
Beta
Measures of goodness of t
X2 (df; p value)
R2
NFI
CFI
GFI
SRMR
RMSEA
DMPP index
Procedural rationality (RP) component
Exhaustivity of information (EI) component
Effort (EF) component
Openness of spirit (OE) component
N = 198
0.208
0.255
0.382
0.200
0.817
0.735
623
GFI statistics were all equal to or greater than 0.919, and the SRMR
and RMSEA statistics were equal to or greater than 0.102. In
addition, the DMPP index explains a good portion of the variance in
the decision quality construct (R2 = 0.362). In sum, the statistics
support external validity of the DMPP index.
By way of comparison, the formative index is superior to the
reective measure in terms of criterion validity. An analysis of the
explained variances for decision quality, as measured by their R2,
shows that the R2 for the index is higher than those for the
reective measures. The gap between DMPP and decision quality is
greater with the formative index. These results suggest that the
proposed formative measure provides a better t but also that it
predicts decision quality better than the various reective
measures used in the past.
5. Discussion
e1
e4
e5
e6
e7
e8
DQ1
DQ2
DQ3
DQ4
DQ5
DMPP 1
0.817
DQ
1
1
DMPP
PR
EI
EF
0.362
1
e3
0.601
OS
e6
624
e7
e8
e9
e10
e11
DQ1
DQ2
1
DQ3
DQ4
DQ5
DQ
0.291
1
0.539
DMPP
e2
PR
EI
1
1
e3
e4
EF
OS
e5
e6
Table 4
Goodness of t of the formative and reective models of external validity.
Formative model (Fig. 3)
Concurrent reective
model (Fig. 4)
0.601
0.362
Measures of goodness of t
X2 (df; p value)
85.805 (28; 0.00)
NFI
0.931
CFI
0.951
GFI
0.919
SRMR
0.053
RMSEA
0.102
0.539
0.291
625
6. Conclusions
Openness of spirit
This study illustrates the importance of building proper
denitions of constructs. We argue that the conceptualization of a
construct must reect its nature and the context of study. We have
shown that DMPP should be conceptualized as a formative
construct, consisting of four dimensions that capture its different
characteristics (i.e. procedural rationality, the exhaustivity of the
information analyzed, effort, and openness of spirit) that ensure
better coverage of the construct. This improved denition of DMPP
should lead to more accurate results in future studies by eliminating
some biases that may have affected past results and conclusions.
It has been empirically demonstrated that DMPP is a signicant
antecedent of decision quality. Using our formative measure, we
made a similar observation, indicating that our conceptualization of
DMPP benets from a certain nomological logic. Overall, our
formative index is conceptually superior to previous approaches
that used reective measures and includes four components that
make signicant contributions to DMPP and appear to be generalizable to many contexts. Future studies, particularly in the DSS area,
will benet from a validated and well conceptualized construct,
encouraging knowledge creation through the reuse of common tools.
Moreover, the construct may prove particularly useful to
managers interested in studying the performance of their
decision making process. The development of a simple instrument, in the form of a questionnaire, represents an attractive
self-assessment tool for managers who want to make sure that
they build on each DMPP component to make the best possible
decision. Our measure will also help them identify weaknesses in
the decision-making process, which could lead to specic
training in these areas. Additionally, being aware of personal
weaknesses may help a manager exercise more care in decision
making. This will ultimately help the manager avoid continuously making the same mistakes, thereby improving personal
performance in this area and, ultimately, the organizations
overall performance.
626