Technological Forecasting & Social Change 119 (2017) 18–26

Contents lists available at ScienceDirect

Technological Forecasting & Social Change

The difference in foresight using the scanning method between experts

and non-experts
Hidehito Honda a,⁎, Yuichi Washida b, Akihito Sudo c, Yuichiro Wajima d, Keigo Awata e, Kazuhiro Ueda a,⁎
a
Graduate School of Arts and Sciences, The University of Tokyo, Japan
b
Graduate School of Commerce and Management, Hitotsubashi University, Japan
c
Institute of Industrial Science, The University of Tokyo, Japan
d
Interfaculty Initiative in Information Studies, The University of Tokyo, Japan
e
Future Design Lab, The Japan Research Institute, Ltd., Japan

a r t i c l e i n f o a b s t r a c t

Article history: We examined the factors that produce differences in generating scenarios on the near future using the scanning
Received 25 August 2016 method. Participants were asked to briefly read (scan) 151 articles about new technology, the latest customs,
Received in revised form 20 February 2017 fashion, social change, value system transition, or emerging social problems, and then to generate three scenarios
Accepted 6 March 2017
about the near future based on the articles. We compared the generated scenarios between scanning method ex-
Available online 14 March 2017
perts and non-experts with no prior experience with the scanning method. We found that experts generated
Keywords:
more unique scenarios than non-experts did, and that experts and non-experts differed in the diversity of articles
Foresight referenced when generating scenarios. We discuss the relationship between the present findings and previous
Scanning method findings on divergent thinking.
Divergent thinking © 2017 Elsevier Inc. All rights reserved.
Difference between experts and non-experts
Creativity

1. Introduction
Although prediction of the future is highly important in product
manufacturing or service provision, it is a highly difficult task. How,
then, can people make insightful projections about the future? In the
present study, we discussed one support method for thinking about
the near future, the scanning method, and examined factors that produce
differences in the projection of the future.
The scanning method is a support method for thinking about the
near future. In the scanning method, one tries to generate scenarios
(foresight) about near-future social situations (10–15 years later). Ac-
cording to Loveridge (2008), the scanning method was originally devel-
oped by Stanford Research Institute. Aguilar (1967), Fahey and King
(1977), and Mueller and Smith (1984) discussed an elementary proce-
dure for the scanning method. Since then, various procedures have been
developed (e.g., Ansoff, 1975; Lesca, 2013; Loveridge, 2008; Mueller and
Smith, 1984; Stoffels, 1994). In the present study, we used the following
two-stage procedure based on previous studies (Washida, 2007;
Washida et al., 2009). In the first phase (scanning phase), an individual
briefly read (scanned) 150–200 newspaper, magazine, or webpage
news articles (see Fig. 1 for an example) describing new technology,
⁎ Corresponding authors at: Graduate School of Arts and Sciences, The University of
Tokyo. 3-8-1, Komaba, Meguro-ku, Tokyo, 153-8902, Japan.
E-mail addresses: hitohonda.02@gmail.com (H. Honda), ueda@gregorio.c.u-tokyo.ac.jp
(K. Ueda).
the latest customs, fashion, social change, value system transition, or
emerging social problems. After scanning all of the articles, the individ-
ual completed a scenario generation phase. In this phase, s/he was asked
to generate scenarios about near-future social situations based on the
scanned articles. S/he was required to provide a title for the scenario,
IDs and keywords for articles referenced when generating the scenario,
and a detailed summary of the scenario (the scenario generation sheet
is shown in Fig. 2). Previous studies have discussed the efficacy of the
scanning method and scenario generation for strategy planning, such
as improvement of generated scenarios (Washida, 2007; Washida et
al., 2009), reduction of cognitive bias, and increase of confidence and
flexibility in strategy planning (Meissner and Wulf, 2013; Phadnis et
al., 2015; Wright and Goodwin, 2009). As Phadnis et al. (2015) pointed
out, few empirical studies have been conducted on the scanning
method.1 Although the scanning method can improve generation of
near-future scenarios, there are individual differences.
In the present study, we examined this issue by comparing the
scanning method between experts and non-experts. Experts were
1
Exceptions are Kuhn and Sniezek (1996), Schoemaker (1993), and Phadnis et al.
(2015). Phadnis et al. (2015) pointed out the following three factors with which empirical
evidence on scanning method and scenario generation has not been accumulated; (1) sce-
nario planning practices are highly personalized and hence difficult to compare, (2) strat-
egy practices used by firms are often not publicized, (3) numerous factors outside
researcher's control may exist.

http://dx.doi.org/10.1016/j.techfore.2017.03.005
0040-1625/© 2017 Elsevier Inc. All rights reserved.
 H. Honda et al. / Technological Forecasting & Social Change 119 (2017) 18–26
Fig. 1. Example article. The top box provides a summary of the article. Keywords and the full article are presented below. This example is translated into English from Japanese. In the actual
study, most articles (149/151) were written in Japanese.
individuals who worked as consultants and used the scanning method
daily.2 Non-experts were individuals with no prior scanning method ex-
perience. Differences in scenario generation between these two groups
would reflect important factors related to difference in foresight. In par-
ticular, as described above, the scanning method and scenario genera-
tion can lead to reduction of cognitive bias and increase of confidence
and flexibility in strategy planning (Meissner and Wulf, 2013; Phadnis
et al., 2015; Wright and Goodwin, 2009). Hence, experts in the scanning
method and scenario generation may be more immune to various cog-
nitive biases and more confident and flexible in scenario generation
2
In the present study, we assumed that the experts for the present scanning method
were those who, in their daily work, collected articles for the scanning phase and used
the method many times. Based on this criterion, the number of experts in Japan is around
ten.
19
than non-experts. However, previous studies have not examined de-
tailed factors that produce differences between experts and non-ex-
perts in the scanning method and scenario generation. For example,
although the literature on how cognitive or contextual factors affect cre-
ativity is large (e.g., Amabile, 1979; Förster et al., 2004; Friedman and
Förster, 2001; Galinsky and Moskowitz, 2000; Maddux and Galinsky,
2009; Markman et al., 2007; Sternberg, 1999), few studies have been
conducted in the context of scanning method and scenario generation.
Then, what is the difference in generating scenarios using the scan-
ning method between experts and non-experts? There may be differ-
ences in both or one of the two phases of the scanning method. For
the scanning phase, we examined differences in subjective impressions
for articles between experts and non-experts. Brown et al. (1998), and
Nijstad and Stroebe (2006) claimed that the knowledge accessed
when generating ideas plays an important role in generating creative
20 H. Honda et al. / Technological Forecasting & Social Change 119 (2017) 18–26
Fig. 2. Scenario generation sheet. At the top, participants provide a title for the scenario. In the middle box, participants provide the IDs and keywords for the articles referenced while
generating the scenario. In the bottom box, participants describe the scenario. In the actual study, this sheet was written in Japanese.
ideas. This suggests that information sources will play an important role
in generating ideas. In the scanning method, people access information
by scanning articles and generate scenarios based on that information. If
experts and non-experts scan the same articles, the source of informa-
tion should be the same. However, there may be differences in subjec-
tive impressions for articles between experts and non-experts. For
example, experts, while not non-experts, may realize that a certain arti-
cle is informative for foreseeing the future. Utilizing the information
from such an article may lead to the difference in foresight. Hence, dif-
ferences in foresight between experts and non-experts may derive
from differences in subjective impressions for articles.
In the scenario generation phase, we examined the diversity of refer-
enced articles. Nijstad et al. (2002) indicated that the diversity of infor-
mation participants used to generate ideas played an important role in
generating creative ideas. In research on the scanning method,
Kiyokawa et al. (2010) showed that the diversity of information partic-
ipants access when generating scenarios affects the quality of generated
scenarios. Based on these studies, there may be a difference between ex-
perts and non-experts in the diversity of articles referenced during sce-
nario generation.
In sum, we examined the two possible differences in scenario gener-
ation using the scanning method between experts and non-experts: ar-
ticle impressions in the scanning phase, and diversity of referenced
articles in the scenario generation phase.
2. Hypothesis testing
2.1. Method
Six experts (five males and one female, Mage = 44.2, SDage = 7.91)
and six non-experts (four males and two females, Mage = 42.5,
SDage = 5.50) participated in this study. They were all Japanese compa-
ny employees. All of them were graduates, with a Bachelor of Arts de-
gree. Non-experts were engaged in the following work in companies:
research and development (2), design (2), planning (1), and consulting
(1), while all experts were engaged in consulting.
Each expert or non-expert completed a scenario generation task
using the scanning method described above. In the scanning phase, s/
he briefly read 151 newspaper, magazine, or webpage news articles
about new technology, the latest customs, fashion, social change,
value system transition, or emerging social problems.3 Participants
rated six items about their impressions of each article. Table 1 shows
the six items. We set the six items based on the careful discussion con-
ducted among the authors. In this discussion, we considered some ad-
vices by an expert of the scanning method about the perspective from
which he was scanning articles (this expert did not participate in the ex-
periment). In the scenario generation phase, participants were asked to
generate three scenarios about social situations in Japan around 2025
based on the articles. For each generated scenario, participants provided
the IDs and keywords of the articles referenced when generating the
scenario, the scenario title, and a detailed scenario summary (see Fig. 2).
Participants of this study (i.e., experts or non-experts) completed
the task individually in a face-to-face setting with the interviewer. In
the scanning phase, the interviewer handed each article to the partici-
pant. The participant briefly read the article and rated her/his impres-
sion. Impression ratings were provided verbally. For example, if the
participant felt that the article was highly positive and moderately use-
ful, they would say “A + 2 and C + 1.” The interviewer recorded re-
sponses. We adopted this procedure for the following two reasons.
First, in the usual procedure of the scanning method, one scans the arti-
cles that are printed on papers one by one because scanning method is
usually utilized in physical workshop-style sessions. Thus, as much as
possible, we intended to set the analogous environment like that in a
usual scanning method. Second, if participants rated impressions
using a questionnaire, they could compare ratings between articles.
This comparison is not normally possible during the scanning phase
(i.e., participants just read articles). When participants are asked to
3
In the scanning phase, it is necessary to be exposed to as much diverse information as
possible. However, scanning too many articles leads to heavy cognitive load and overflow
of information. Thus, 100 to 200 articles are suitable for scanning (Washida et al., 2009). In
the present study, the order of presentation of articles was the same for all participants.
 H. Honda et al. / Technological Forecasting & Social Change 119 (2017) 18–26 21

Table 1
Six items about impressions of the article.

Item How do you feel about the article,

A Positive or negative?
(negative = −2, −1, 0, 1, positive = 2)
B Pleasant or unpleasant?
(unpleasant = −2, −1, 0, 1, pleasant = 2)
C Useful or not useful?
(not useful = −2, −1, 0, 1, useful = 2)
D Have seen before or have never seen before
(have never seen before = −2, −1, 0, 1, have seen before = 2)
E Cool or not cool?
(not cool = −2, −1, 0, 1, cool = 2)
F Curious?
(not curious = 0, 1, curious = 2)

rate their impressions of an article verbally, direct comparisons between
articles are highly difficult (i.e., participants would have to rely on
memory).
The whole interview took about 3 h. During the interview, partici-
pants were allowed to take rest whenever they wanted. This interview
was conducted from May to June 2014.4
2.2. Results and discussion
We obtained impression ratings for 151 articles and 36 scenarios (12
participants × 3 scenarios). Using the generated scenarios, we examined
the difference in generated scenarios between experts and non-experts.
Impression ratings were used to evaluate the two hypotheses.
Fig. 3. Mean generated scenario ratings for the three evaluation items. Error bars denote
standard error.
and feasible though there were no significant differences in usefulness
and feasibility between experts and non-experts.
Accordingly, these results indicate that experts tend to weigh
uniqueness more while generating scenarios. In contrast, non-experts
tend to weigh usefulness and feasibility more. Maybe, there is trade-
off between uniqueness and usefulness as well as feasibility. That is,
the more (or less) unique a scenario is, the less (or more) useful or fea-
sible it may be. Thus, it is noteworthy that experts generated more
unique scenarios than non-experts did, while there were no significant
differences in their evaluations on usefulness and feasibility.

2.2.2. Difference in subjective impressions for articles between experts and

2.2.1. Generated scenarios
non-experts
First, we evaluated the difference in generated scenarios between
Next, we examined differences in subjective impressions for articles.
experts and non-experts. The generated scenarios were evaluated by
As discussed in the introduction, experts may, while non-experts may
three independent evaluators (two university professors in cognitive
not, realize that a certain article is informative for projecting the future.
or marketing science, and one postdoctoral researcher in computer
Therefore, there may be differences in subjective impressions for arti-
science).5 Each evaluator individually evaluated 36 scenarios using a
cles between experts and non-experts.
computer. For each generated scenario, the title and detailed summary
First, we examined individual differences within each group for each
were presented on the computer screen, and the evaluator was asked
item. In particular, we examined whether the direction of the impres-
to evaluate the scenario on a five-point scale in terms of uniqueness
sion rating was the same or different. For questions (A), (B), (C), (D),
(1 = not unique at all, 5 = very unique), usefulness (1 = not useful
and (E), when all six participants (i.e., all experts or non-experts) gave
at all, 5 = very useful), and feasibility (1 = not feasible at all, 5 =
ratings greater (or less) than or equal to 0, the six participants were con-
very feasible). Previous studies argued that these three items are useful
sidered to be consistent. Otherwise, the rating pattern was assumed to
for evaluating generated ideas (e.g., Amabile et al., 2005; Franke et al.,
be inconsistent. For question (F), ratings were categorized as 0 or 1
2006; Kristensson et al., 2004; Moreau and Dahl, 2005; Wajima et al.,
and 2. Rating patterns were considered consistent when ratings for
2013).6
more than three participants were in the same category. Otherwise,
We examined the consistency of evaluations among the three eval-
the rating pattern was regarded as inconsistent. Using these criteria,
uators using Kendall's coefficient of concordance. The coefficients
we examined whether impression ratings were consistent or inconsis-
were 0.579 (p b 0.01) for uniqueness, 0.589 (p b 0.01) for feasibility,
tent within each group (experts and non-experts) for each of the 151 ar-
and 0.767 (p b 0.01) for usefulness. These results indicate that evalua-
ticles (see the upper panel of Fig. 4).7
tions were not statistically inconsistent among the three evaluators.
Using these within-group consistency categorizations, we examined
Therefore, we used the median evaluations across the three evaluators
the consistency of impressions between experts and non-experts. We
as the evaluation values. Fig. 3 shows the mean evaluation values for
calculated the proportion of times each of five patterns was observed
the three items for experts and non-experts. We compared evaluation
in the 151 articles. The lower panel of Fig. 4 shows the five patterns.
values between experts and non-experts for each item. There was a sig-
To examine frequencies of observed patterns, we conducted a simula-
nificant difference in uniqueness (t = 2.30, df = 34, p b 0.05, d = 0.77),
tion in which simulated twelve participants were assumed to provide
but not in usefulness (t = 1.28, df = 34, p = 0.21, d = 0.43) and feasi-
random impression ratings for each question, and calculated the pro-
bility (t = 0.62, df = 34, p = 0.54, d = 0.21). These results indicate that
portions each of the five patterns was observed. The mean of 1000 sim-
experts generated more unique scenarios than non-experts did, and
ulations was used for calculating the proportion produced by simulated
that non-experts tended to generate scenarios that were rather useful
participants.
4
Fig. 5 shows the observed and simulated proportions for the five
The interviewer was the same for all participants.
5 consistency patterns. For each question, we tested whether the
Three evaluators did not know who (expert or non-expert) generated each scenario.
6
In Kristensson et al. (2004), the three items were named as originality, value, and real-
7
ization. They argued that these items capture the qualities of a creative performance. In the For question (F), consistency categories 1, 2, 4, and 5 within a group in Fig. 4 were as
present study, we modified the three items (uniqueness, usefulness and feasibility) as follows: Categories 1 and 4 correspond to ratings 1 and 2, and categories 2 and 5 corre-
they were appropriate for the evaluation of foresight. spond to rating 0.
22 H. Honda et al. / Technological Forecasting & Social Change 119 (2017) 18–26
Consistency type between experts and non-experts
Same
Diff
C-I
I-C
I-I
Fig. 4. Within-group consistency (upper panel) and five impression rating consistency patterns for experts and non-experts (lower panel).
consistency patterns were significantly different between the observed
and simulated data. The observed data was significantly different from
the simulated data for all questions (Fisher's exact tests, p b 0.001 for
(A), (B), (C), (D), and (E), and p b 0.05 for (F)). Next, we conducted
pair-wise comparisons to examine which response patterns were sig-
nificantly different between the observed and simulated data for each
question. Table 2 shows the results of the pair-wise comparisons. For
items (A), (B), (C), (D), and (E), common differences between observed
and simulated data were “Same,” “Diff,” and “I-I.” These results indicate
that although the proportion of “Diff” was significantly higher in the ob-
served than in the simulated data, the proportion of “Same” was much
higher in the observed than in the simulated data. Thus, rating direction
was generally consistent between experts and non-experts. For item
(D), “I-C” was also significant. Because item (D) asked about familiarity
with the article, there may be more individual differences for this item
compared to other items. For item (F), although the proportion of “C-
I” was lower in the observed than in the simulated data, response pat-
terns in the observed data were generally similar to those in the simu-
lated data. This indicates that although there were individual
differences in curiosity, there were no systematic differences between
experts and non-experts.
Taken together, the results indicate that there were no systematic
differences in article impressions between experts and non-experts,
1–4, 2–5
1–5, 2–4
1–6, 2–6
3–4, 3–5
3–6
suggesting that subjective impressions for articles cannot account for
differences in generated ideas between experts and non-experts.
2.2.3. Diversity of referenced articles in the scenario generation phase
Next, we examined the diversity of articles referenced when gener-
ating scenarios. We defined diversity of referenced articles using the
simple matching coefficient (SMC) between impression ratings. The
SMC is calculated as:
number of matching rating
SMC ¼ ð1Þ
number of compared rating
The number of items where the ratings are the same is the numera-
tor, while the number of items compared is the denominator (i.e., six).
The SMC takes a value between 0 and 1. It equals to 1 when ratings
for the six questions are identical. In contrast, it equals to 0 when ratings
completely differ across the six questions. For example, if a participant
used three articles, Art 1, Art 2, and Art 3, to generate Scenario X, and im-
pression ratings for the six items were [2, 0, −1, 1, 0, 0], [2, 1, 0, 1, 0, 0],
and [2, 2, 0, −1, −1, 1], the SMC between Art 1 and Art 2 (SMCArt1, Art2)
would be calculated as follows: the numerator (the number of matching
ratings) would be 4 (A, D, E, F) and the denominator (the number of
compared ratings) would be 6. Thus, SMCArt1, Art2 would be 4/6.
 H. Honda et al. / Technological Forecasting & Social Change 119 (2017) 18–26 23

Fig. 5. Observed and simulated proportions (out of 151 articles) for the five impression consistency patterns.

Likewise, the SMC between Art 1 and Art 3, and that between Art 2 and (B) shows the distribution of the number of referenced articles. Evident-
Art 3 was 1/6 and 2/6, respectively. The diversity of articles referenced ly, experts referred significantly more articles than non-experts did
when generating Scenario X (DivX) is defined as: (t = 2.23, df = 34, p b 0.05, d = 0.71).
These results show the possibility that experts referenced more arti-
4=6 þ 1=6 þ 2=6 cles in generating scenarios than non-experts did, and this resulted in
DivX ¼ 1− ¼ 0:611 ð2Þ
3 the difference in the diversity of the referenced articles. We then exam-
ined whether only the number of referenced articles explained the dif-
Thus, we assumed that a participant referenced more diverse articles ference in the diversity of referenced articles between experts and non-
when the scenario generation was based on articles with more different experts. We therefore conducted the following regression analysis. We
impression ratings. first compared the three models (Table 3). In these models, we used di-
First, we compared the diversity of the referenced article between versity (Div) as the dependent variable, and we used different indepen-
experts and non-experts. Fig. 6 (A) shows the distribution of diversity dent variables depending on the models. In Model A, we used the
(Div) for articles referenced in generating scenarios. It was found that number of articles and a dummy variable discriminating experts (1)
experts referred to more diverse articles in generating scenarios than and non-experts (0) as the independent variables. In Model B and C,
non-experts did (t = 2.78, df = 34, p b 0.01, d = 0.93). Since the number we used either number of articles or the dummy variable. We compared
of referenced articles may have been related to the diversity, we then these three models in terms of fit and parsimony using the AIC. Findings
compared the number between the experts and non-experts. Fig. 6 revealed that Model A was the best among the three models. According-
ly, we discussed the estimated parameters for Model A. Results of the re-
Table 2 gression analysis were as follows: the estimated coefficients for the
Results of pair-wise comparisons. intercept, the number of articles, and the dummy variable were 0.26
(p b 0.001), 0.02 (p b 0.01), and 0.07 (p = 0.07), respectively, and the
Significant differences based on Bonferroni-corrected pair-wise
Question comparisons. adjusted R-squared was 0.346 (see Fig. 7). Note that the dummy vari-
able was marginally significant. These results indicated that experts
A Same (p b 0.001), Diff (p b 0.001), I-I (p b 0.001)
B Same (p b 0.001), Diff (p b 0.001), I-I (p b 0.001), tended to reference more diverse articles than non-experts did, even
C Same (p b 0.001), Diff (p b 0.001), I-I (p b 0.001), when the number of referenced articles was the same. We note that
D Same (p b 0.001), Diff (p b 0.001), I-C (p b 0.001), I-I (p b 0.001) the estimated parameter for the number of referenced articles was
E Same (p b 0.001), Diff (p b 0.001), I-I (p b 0.001) very small and the adjusted R-squared was rather modest. Thus, al-
F C-I (p b 0.05)
though the number of referenced articles was significantly related to
24 H. Honda et al. / Technological Forecasting & Social Change 119 (2017) 18–26

Fig. 6. Statistics regarding the generated scenarios. (A) denotes the diversity of articles referenced in generating a scenario. (B) denotes the number of articles referenced in generating a
scenario.

the diversity of referenced articles, the relationship may have been es-
sentially small.8
Taken together, these results indicate that there was a difference in
the diversity of the referenced articles between experts and non-ex-
perts. Experts tended to refer more diverse articles in generating ideas
than non-experts did. This was partly explained by the number of refer-
enced articles since experts referenced more articles in generating ideas
than non-experts did. However, the results of the regression analysis
suggested that even when the number of referenced articles was the
same, the diversity was higher for experts than it was for non-experts.
3. General discussion
We examined the difference in foresight between experts and non-
experts using the scanning method. Our findings can be summarized
as follows. First, experts generated more unique scenarios than non-ex-
perts did, while non-experts tended to generate scenarios that were
rather useful and feasible though there were no significant differences
in usefulness and feasibility between experts and non-experts. Second,
in the scanning phase, experts and non-experts had similar subjective
impressions of the articles (see Fig. 5). Third, in the scenario generation
phase, the diversity of the referenced articles differed between experts
and non-experts. It was found that experts referenced more diverse ar-
ticles in generating scenarios than non-experts did. This difference was
partly derived from the difference in the number of articles referenced
in generating scenarios (see Fig. 6). However, even when the number
of referred articles was the same, the articles referred by experts were
more diverse than those referred by non-experts were (see Fig. 7).
8
One scenario by a non-expert was generated based on only one article. This outlier
may have distorted the results of the statistical analyses. Hence, we conducted the same
analyses by excluding this data. We found that this outlier did not distort the results: Di-
versity of the referenced articles was higher for experts than for non-experts (t = 2.65,
df = 33, p b 0.05, d = 0.70), and number of referenced article was marginally higher for
experts (t = 1.84, df = 33, p = 0.07, d = 0.64). As to the results of the regression analysis,
Model A was the best among the three models tested (see Table 3). The estimated coeffi-
cients for the intercept, number of articles, and dummy variable in Model A were 0.32
(p b 0.001), 0.02 (p b 0.05), and 0.06 (p = 0.05), respectively, and the adjusted R-
squared was 0.251 after the exclusion of an outlier.
These results indicate that the differences in the generated scenarios be-
tween experts and non-experts can be attributed to the diversity of ar-
ticles referenced during scenario generation.
In the scanning phase, there were no significant differences in sub-
jective impressions between experts and non-experts. Both experts
and non-experts did not differ in terms of their ages and education
level. In contrast, their work contents varied. This difference may have
produced some differences in knowledge about new technology, the
latest fashion, customs, and social issues. However, it is noteworthy
that the impressions of articles were highly similar among the partici-
pants of the two groups, and that there was a significant difference in
the generated scenarios between experts and non-experts. This sug-
gests that diversity of articles referenced in the scenario generation
phase plays an important role in producing individual differences in
generated scenarios. Previous studies on creativity have identified
many cognitive factors that affect creativity. One of the most important
factors is divergent thinking (Guilford, 1968, 1982), which is the ability
to extend original ideas and generate multiple solutions to a problem.
Hence, previous and present findings together suggest that diversity
of referenced article in the scanning method promotes divergent think-
ing and produces individual differences.
Finally, we note three limitations in the present study. First, the du-
ration of the experiment (3 h) was so long that some adverse effects on
results may have been produced (e.g., fatigue). In future research, it will
be better to revise the experimental procedure and shorten the duration
of the experiment. Second, we have to examine the features of generat-
ed scenarios from different perspectives. In the present study, we eval-
uated the scenarios using three items; uniqueness, usefulness, and
Table 3
The three regression models tested for explaining the diversity of referenced articles.
AIC AIC
Independent variable (all data) (excluding one outlier)
Model A (N) + (D) −57.1 −65.7
Model B (N) −55.5 −63.6
Model C (D) −49.0 −62.2
Note. (N) denotes number of articles and (D) denotes dummy variable discriminating ex-
perts and non-experts.
 H. Honda et al. / Technological Forecasting & Social Change 119 (2017) 18–26
Fig. 7. Diversity (Div) as a function of number of referred articles. The lines denote the
regression lines.
feasibility. We used these three items according to previous studies
(e.g., Amabile et al., 2005; Franke et al., 2006; Kristensson et al., 2004;
Moreau and Dahl, 2005; Wajima et al., 2013). However, some dimen-
sions that affect idea generation may exist. Based on these dimensions,
the evaluations of the generated scenarios may change. From this anal-
ysis, we will obtain more clues about how people can take advantage of
the scanning method for generating insightful ideas. Thus, it is neces-
sary to evaluate the generated scenarios from a wider perspective.
Third, although our findings suggest that diverse perspectives play an
important role in generating scenarios about the future, we have to ex-
amine the link between divergent thinking and the diversity of refer-
enced articles. Here, we discuss two ways to enable people to
generate more diverse perspectives in the scanning method. (A)
While using the scanning method, the instruction about diverse per-
spectives should be emphasized upon. Based on the present findings,
the following instruction may be effective: “In generating a scenario,
refer to articles regarding which you have different impressions.” By en-
couraging the use of articles that seem different, people may explicitly
have diverse perspectives. (B) People may implicitly have diverse per-
spectives after scanning diverse articles in the scanning phase. Thus,
while using the scanning method, it may be effective to collect “diverse”
articles. For example, by collecting articles from as many fields or topics
as possible, “diversity” of articles may be achieved. In sum, we believe
that there are two ways (i.e., explicit and implicit approaches) to facili-
tate the diverse perspective. In future research, these two possibilities
should be empirically examined, and the link between divergent think-
ing and the diversity of referenced articles needs to be examined.
In conclusion, we found differences in foresight between experts and
non-experts in the use of the scanning method. The most important dif-
ference was the diversity of articles referenced during scenario genera-
tion. This may be related to the fact that divergent thinking plays an
important role in producing individual differences in the generated
ideas. These findings contribute to clarifying the factors that promote
the generation of insightful ideas.
Acknowledgement
This research was supported by Hitachi Solutions, Ltd. and Grants-
in-Aid for Scientific Research (A, 16H01725) and Young Scientists (B,
16K16070). Yuichiro Wajima is now in Office of Management and Plan-
ning at Osaka University. We appreciate the insightful comments made
by the anonymous reviewer.
References
Aguilar, F.J., 1967. Scanning the Business Environment. Macmillan.
Amabile, T.M., 1979. Effects of external evaluation on artistic creativity. J. Pers. Soc.
Psychol. 37:221–233. http://dx.doi.org/10.1037/0022-3514.37.2.221.
25
Amabile, T.M., Barsade, S.G., Mueller, J.S., Staw, B.M., 2005. Affect and creativity at work.
Adm. Sci. Q. 50:367–403. http://dx.doi.org/10.2189/asqu.2005.50.3.367.
Ansoff, H.I., 1975. Managing strategic surprise by response to weak signals. Calif. Manag.
Rev. 18, 21–33.
Brown, V., Tumeo, M., Larey, T.S., Paulus, P.B., 1998. Modeling cognitive interactions dur-
ing group brainstorming. Small Group Res. 29:495–526. http://dx.doi.org/10.1177/
1046496498294005.
Fahey, L., King, W.R., 1977. Environmental scanning for corporate planning. Bus. Horiz. 20:
61–71. http://dx.doi.org/10.1016/0007-6813(77)90010-6.
Förster, J., Friedman, R.S., Liberman, N., 2004. Temporal construal effects on abstract and
concrete thinking: consequences for insight and creative cognition. J. Pers. Soc.
Psychol. 87:177–189. http://dx.doi.org/10.1037/0022-3514.87.2.177.
Franke, N., Von Hippel, E., Schreier, M., 2006. Finding commercially attractive user inno-
vations: a test of lead-user theory. J. Prod. Innov. Manag. 23:301–315. http://dx.doi.
org/10.1111/j.1540-5885.2006.00203.x.
Friedman, R.S., Förster, J., 2001. The effects of promotion and prevention cues on creativity.
J. Pers. Soc. Psychol. 81:1001–1013. http://dx.doi.org/10.1037/0022-3514.81.6.1001.
Galinsky, A.D., Moskowitz, G.B., 2000. Counterfactuals as behavioral primes: priming the
simulation heuristic and consideration of alternatives. J. Exp. Soc. Psychol. 36:
384–409. http://dx.doi.org/10.1006/jesp.1999.1409.
Guilford, J.P., 1968. Intelligence, Creativity, and Their Educational Implications. Knapp,
New York, NY.
Guilford, J.P., 1982. Cognitive psychology's ambiguities: some suggested remedies.
Psychol. Rev. 89:48–59. http://dx.doi.org/10.1037/0033-295X.89.1.48.
Kiyokawa, S., Washida, Y., Ueda, K., Peng, E., 2010. Can diverse information improve idea
generation? Cogn. Stud. 17, 635–649.
Kristensson, P., Gustafsson, A., Archer, T., 2004. Harnessing the creative potential among
users. J. Prod. Innov. Manag. 21:4–14. http://dx.doi.org/10.1111/j.0737-6782.2004.
00050.x.
Kuhn, K.M., Sniezek, J.A., 1996. Confidence and Uncertainty in Judgmental Forecast-
ing: Differential Effects of Scenario Presentation. J. Behav. Decis. Mak. 9:
231–247. http://dx.doi.org/10.1002/(SICI)1099-0771(199612)9:4b231::AID-
BDM240N3.0.CO;2-L.
Lesca, N. (Ed.), 2013. Environmental Scanning and Sustainable Development. John Wiley
& Sons.
Loveridge, D., 2008. Foresight: The Art and Science of Anticipating the Future. Routledge.
Maddux, W.W., Galinsky, A.D., 2009. Cultural borders and mental barriers: The relation-
ship between living abroad and creativity. J. Pers. Soc. Psychol. 96:1047–1061.
http://dx.doi.org/10.1037/a0014861.
Markman, K.D., Lindberg, M.J., Kray, L.J., Galinsky, A.D., 2007. Implications of counterfactu-
al structure for creative generation and analytical problem solving. Personal. Soc.
Psychol. Bull. 33:312–324. http://dx.doi.org/10.1177/0146167206296106.
Meissner, P., Wulf, T., 2013. Cognitive benefits of scenario planning: its impact on biases
and decision quality. Technol. Forecast. Soc. Change 80:801–814. http://dx.doi.org/
10.1016/j.techfore.2012.09.011.
Moreau, C.P., Dahl, D.W., 2005. Designing the solution: the impact of constraints on con-
sumers' creativity. J. Consum. Res. 32:13–22. http://dx.doi.org/10.1086/429597.
Mueller, G., Smith, J.B., 1984. Six “commandments” for successful futures studies for cor-
porate planning null. J. Bus. Strategy 5:88–92. http://dx.doi.org/10.1108/eb039063.
Nijstad, B.A., Stroebe, W., 2006. How the group affects the mind: a cognitive model of idea
generation in groups. Personal. Soc. Psychol. Rev. 10:186–213. http://dx.doi.org/10.
1207/s15327957pspr1003_1.
Nijstad, B.A., Stroebe, W., Lodewijkx, H.F.M., 2002. Cognitive stimulation and interference
in groups: exposure effects in an idea generation task. J. Exp. Soc. Psychol. 38:
535–544. http://dx.doi.org/10.1016/S0022-1031(02)00500-0.
Phadnis, S., Caplice, C., Sheffi, Y., Singh, M., 2015. Effect of scenario planning on field ex-
perts' judgment of long-range investment decisions. Strateg. Manag. J. 36:
1401–1411. http://dx.doi.org/10.1002/smj.2293.
Schoemaker, P.J.H., 1993. Multiple scenario development: Its conceptual and behavioral
foundation. Strateg. Manag. J. 14:193–213. http://dx.doi.org/10.1002/smj.4250140304.
Sternberg, R.J., 1999. A propulsion model of types of creative contributions. Rev. Gen.
Psychol. 3:83–100. http://dx.doi.org/10.1037/1089-2680.3.2.83.
Stoffels, J.D., 1994. Strategic Issues Management: A Comprehensive Guide to Environmen-
tal Scanning. Pergamon.
Wajima, Y., Washida, Y., Tominaga, N., Ueda, K., 2013. Improvement in quality of business
ideas by introducing user perspective. Jpn. Soc. Artif. Intell. 28:409–419. http://dx.doi.
org/10.1527/tjsai.28.409.
Washida, Y., 2007. Foresight. NTT Publish CO., Ltd., Tokyo.
Washida, Y., Mitsuishi, S., Horii, H., 2009. A future scenario generation experiment in
socio-technological problems using scanning method. SOCIOTECHNICA 6:1–15.
http://dx.doi.org/10.3392/sociotechnica.6.1.
Wright, G., Goodwin, P., 2009. Decision making and planning under low levels of predict-
ability: enhancing the scenario method. Int. J. Forecast. 25, 813–825 (doi:0.1016/
j.ijforecast.2009.05.019).
Hidehito Honda is a postdoctral research fellow in Graduate School of Arts and Sciences at
the University of Tokyo. He received his M.A. and Ph.D. in cognitive science from Tokyo In-
stitute of Technology. His current research interest includes human judgment and decision
making.
Yuichi Washida: Ph.D. and Professor at the Graduate School of Commerce and Manage-
ment, Hitotsubashi University. Washida has written several books regarding information
technology and society. He conducted the joint research project with Toshiba, Ricoh, and
Ericsson Consumer Laboratory. He also participated in the Japanese Governmental re-
search project at Ministry of Economy, Trade and Industry.
26 H. Honda et al. / Technological Forecasting & Social Change 119 (2017) 18–26

Akihito Sudo is an assistant professor in the institute of industrial science at the University
of Tokyo. He received his M.Eng. in applied physics from Waseda University and Ph.D. in
intelligent system from Tokyo Institute of Technology. His current research interests in-
clude human mobility prediction and computational creativity using Bayesian statistics
and machine learning techniques.
Yuichiro Wajima is an assistant professor in Office of Management and Planning at Osaka
University. He received his Ph.D. in cognitive science from Tokyo Institute of Technology.
Kazuhiro Ueda is a professor at the University of Tokyo. He received M.A. and Ph.D. de-
grees in cognitive science from the University of Tokyo in 1990 and 1993, respectively.
His current research interests include cognitive analysis on creative thinking and decision
making, behavioral economics, cognitive analysis of skill acquisition in Japanese tradition-
al arts such as Bunraku and Nogaku, cognitive neuroscience approach to social cognition
and perception such as eye-gaze perception and animacy perception, as well as cognitive
interaction studies.

Keigo Awata is a director in Future design Lab at The Japan Research Institute, Ltd. His cur-
rent research interest includes support for innovation generation using the scanning
method.