of real options reasoning in electronic commerce investment based on
the works of Tiwana et al. (2007). All
19 items were loaded onto one factor, so
we used real options aggregately in our
Perceived Environmental Uncertainty
We assessed perceived state (a 5 0.89),
effect (a 5 0.79), and response (a 5 0.82)
uncertainties using nine items (Ashill &
Jobber, 2009). Examples of items include:
“You have the information to understand how your business environment
will change in the future” (state uncertainty); “You fully understand the effect
of the environment factor on your decision making” (effect uncertainty); and
“You can accurately anticipate the consequences/outcomes of decisions before
implementing them” (response uncertainty) (Ashill & Jobber, 2009). In the
current study, multi-item Likert scales
were used to measure the dependent
and independent constructs (see the
Appendix for items and factor loading).
Some variables not considered in the
hypotheses are presented herein, and
these variables may still influence real
options reasoning and project performance. At the individual level, we
controlled project manager age, education, and gender. At the firm level,
we controlled new technology venture
size (logarithm of number of employees)
and age (logarithm of years of operation). At the project level, we controlled
project net present value (NPV), team
size (logarithm of number of individuals involved in the project), complexity,
duration (logarithm of number of months
required for project completion), and
direct responsibility of project managers
(whether respondent had direct personal
responsibility for initiating the project).
All project level control variables were
adapted from Tiwana et al. (2007).
In Table 1, we summarize the descrip-
tive statistics and correlations of all
variables in the study. To test the hypoth-
esized relationships meaningfully, it
was first necessary to establish that
the types of uncertainty were, in fact,
differentiable. An exploratory factor
analysis (EFA) revealed that the fac-
tors grouped themselves according to
the theory (Kaiser-Meyer-Olkin 5 0.932;
Bartlett’s 5 0.000).
To reduce the threat of multicollinearity, dependent variables were
centered before conducting regression
analyses. In addition, we checked variance inflation factors (VIF) to exclude
multicollinearity. The results of VIF testing of all our variables were significantly
below 5, suggesting that our model of
study did not have a serious multicollinearity problem (Cohen, Cohen, West,
& Aiken, 2013).
To test the study hypotheses, we performed a hierarchical multiple regression analysis. In the first step, we entered
the control variables with real options
reasoning as dependent variables. In the
second step of each regression equation,
we entered the independent variables
(state, effect, and response uncertainty).
Table 2 and Figure 2 list the direct effects
of the three types of uncertainties on
real options reasoning.
We first address Hypothesis 1,
which deals with the direct and positive
relationship between perceived environmental state uncertainty and real
options reasoning. As can be inferred
from Table 2, the relationship between
perceived environmental state uncertainty and real options reasoning is
positive and statistically significant
(B 5 0.318, p , 0.05). Thus, hypothesis
1 was fully supported. This means that
under higher state uncertainty, project
managers use greater numbers of real
Our second hypothesis indicated
that there is a positive and significant
relationship between perceived effect
uncertainty and real options reasoning.
As can be inferred from Table 2, the path
from perceived environmental effect
uncertainty to real options reasoning is
positive but non-significant (B 5 0.031,
p , 0.05). Thus, Hypothesis 2 was not
supported by our results.
In the third hypothesis, we predicted
that perceived response uncertainty
was negatively related to real options
reasoning. As summarized in Table 2,
the perceived environmental response
uncertainty is significantly negatively
related to real options reasoning (B 5
20.503, p , 0.01). Hypothesis 3 was
fully supported by our results. This
means that under higher environmental
response uncertainty, project managers
prefer to consider fewer real options.
The results of testing Hypotheses 4,
5, and 6 by hierarchical multiple regres-
sion analysis are summarized in Table 3
and Figure 2.
The fourth hypothesis predicted that
real options reasoning would be nega-
tively associated with project timeliness.
As predicted, our results demonstrated
that higher usage of real options rea-
soning decreases the project timeliness
(b 5 20.632; p , 0.001). Thus, Hypoth-
esis 4 was supported by our results.
Hypothesis 5 proposed that real
options reasoning would be positively
associated with project effectiveness.
As predicted, our results confirmed that
the higher usage of real options reason-
ing leads to higher project effectiveness
(b 5 0.393; p , 0.001). Hypothesis 5
was fully supported. And finally, con-
sistent with Hypothesis 6, we found that
real options reasoning significantly and
positively relates to project efficiency
(b 5 0.393; p , 0.001).
Among the control variables, surprisingly, education was negatively related
to project timeliness (b 5 20.389; p ,
0.05) and positively related to project
effectiveness (b 5 0.434; p , 0.05) and
efficiency (b 5 0.381; p , 0.05). Consistent with the argument of Tiwana,
Wang, Keil, and Ahluwalia (2007), we
found a negative relationship between
project net present value (NPV) and the
usage of real options (b 5 20.376; p ,
0.001). This means that project managers associate real options reasoning
with perceived project value in ecom-merce projects with low NPV.