Content uploaded by Timo Saarinen
Author content
All content in this area was uploaded by Timo Saarinen
Content may be subject to copyright.
Relationship between Uncertainty and
Patterns of Pre-purchase Consumer
Search in Electronic Markets
Theresa Lauraeus-Niinivaara, Timo Saarinen, Anne Sunikka and Anssi Öörni
Helsinki School of Economics
theresa.lauraeus@hse.fi, timo.saarinen@hse.fi, anne.sunikka@hse.fi, anssi.oorni@hse.fi
Abstract
Electronic markets are expected to facilitate consumer
information search and product comparison to the extent
that consumers are able to accumulate nearly perfect
information. We present an analysis of search patterns
based on a laboratory experiment on product search
processes. We identified three types of search patterns in
our experiment: sequential, simultaneous, and iterative.
We found that search pattern has an impact on search
costs and the efficiency of search judged by the purchase
price. Sequential search emerged as the still dominant
search pattern even though it leads to the most expensive
purchase. Simultaneous search seems to combine low
search costs with the highest efficiency. Iterative search
pattern was the slowest. We also studied the relationship
between uncertainty and search pattern, because
uncertainty should have an effect on the search pattern
employed. We found that uncertainty is strongly related to
search behavior, but not to the search pattern employed.
1. Introduction
Consumer search is the main method, besides
advertising, for acquiring information necessary to make
purchase decisions. Consumers look for products with
desired qualities and sellers offering these products at
competitive prices in an attempt to decide what, when,
and from whom to purchase. Markets are dynamic, which
results in information becoming obsolete [33]. Changing
identity of sellers and buyers, and also fluctuations in
supply and demand, result in uncertainty. Identification of
prospective products and sellers is often the dominant
motive of search. Another, yet related, cause is
consumers’ inability to ascertain product quality and
seller reliability before the purchase decision [39], [40].
Information search precedes many consumer decisions
[26], [7], [28], [3], [30]. However, while extensive search
may precede some procurement decisions, others are
made routinely with little, if any, search and consumers
are often found to engage in limited search even for high-
ticket durables [25]. The complexity of consumer decision
phenomenon is depicted by a notion that more than 60
determinants have been related to pre-purchase consumer
search [30].
Information search is often costly [33]. The main cost
factor is typically the opportunity cost of the searcher’s
time. Search costs depend on consumer’s ability to search,
which heavily impacts the pattern of search one can
adopt. Search theory is rather uniform in its definition of
the implications of search costs on consumer behavior and
price dispersion. Stigler [33] proposed that high search
costs will lead value maximizing consumers to limit their
pre-purchase search, which results in less than perfectly
informed purchase decisions. Since consumers vary on
their market knowledge and search costs, relatively wide
price dispersions persist in many consumer markets. The
very basis of search theory [33] suggests two of the most
profound measures of search costs: the amount of search
and price dispersion for products of comparable quality.
These are the two key measures that this work examines
in an attempt to determine how electronic consumer
markets have affected pre-purchase consumer search.
Consumer information search has been one of the
most enduring literature streams in consumer research [3].
Marketing and consumer behavior researchers have been
examining consumer’s pre-purchase information seeking
behavior since at least 1917 [10] and even today most
consumer information processing and decision making
models include pre-purchase information search as one of
the key components [5], [7], [12], [14], [27]. There have
been three major theoretical streams of consumer
information search literature [30], [32]: psychological /
motivational, economics, and consumer information
processing approaches.
In recent decades, there have been many
investigations into consumer search behaviour in a digital
Proceedings of the 41st Hawaii International Conference on System Sciences - 2008
1530-1605/08 $25.00 © 2008 IEEE 1
environment [8], [9], [15], [16], [17], [21], [30], [31], [40]
in the context of search attributes [17], [30], [40].
Recently, there has been research into internet-based
market efficiency [23], [40] and search costs [13], [15]. In
a digital environment consumer information pre-purchase
and search behavior is expected to be different from the
traditional search behaviour [16], [17], [39].
The research of consumer behaviour in electronic
markets and consumer choice of distribution channels is
in need of sound theoretical frameworks that enable
researchers to integrate electronic markets research with
adjacent fields of study. Previous research has been
largely explorative and difficult to assimilate to existing
consumer behavioural research for lack of a connecting
theoretical frame.
In this paper, we have demonstrated that search
behaviour together with uncertainty is a concept that can
be used to explain the variation in the extent of consumer
search in electronic markets. In the body of this paper, we
will first discuss the prototypical search patterns
identified in consumer behavioral literature; sequential
and simultaneous search. This work then connects the
pattern of the search process to the outcomes of search,
i.e. price of purchase and time cost of search. We set up a
laboratory experiment in which the subjects searched for
compact discs in contexts with varying degree of purchase
related uncertainties. We have observed the resulting
search process, identified the prototypical patterns of
search, and studied the impact of the patterns on the
outcomes of search. For consumer uncertainty we will
propose a new conceptualization, based on the multiple
criteria decision making lexicon. We will test the effect of
uncertainties and search patterns by constructing a truth
table to find the explanation with the least number of gaps
to account for the different patterns of search. Finally we
will discuss the outcomes of search.
We have three claims in our work: 1) Consumers
employ different search patterns in their pre-purchase
search. 2) Search is shaped by the uncertainties related to
the purchase decision. 3) Search patterns have an effect
on the outcomes of search.
2. Information search behaviour
There have been three major theoretical streams of
consumer information search literature [30], [32]. The first
is the psychological/motivational approach, which
incorporates the individual, the product class, and the task
related variables such as beliefs and attitudes [3], [11] and
involvement [3]. The second is the economics approach,
which uses the cost-benefit framework to study
information search [30], [32]. The economic theory of
search states that consumers weight the cost and benefits
of search when making search decisions. The third one is
the consumer information processing approach which
focuses on memory and cognitive information processing
theory [30], [32].
Search is often characterized by the locus of search
activity. Information search behavior can be defined as
“the motivated activation of knowledge stored in memory
or acquisition of information from the environment” [12].
As the definition suggests, information search can be
either internal or external. Internal search is based on the
retrieval of knowledge from memory. On the other hand,
external search consists of collecting information from the
marketplace [12]. Generally, it is believed that consumers
tend to acquire information as a strategy of certain risk
reduction efforts in the events of identified uncertainty
regarding the outcome of an action [24] and in the events
of identified discrepancy between external information
and prior product knowledge to protect themselves and to
maximize their satisfaction [5], [35]. However, consumers’
information search behavior is likely to be influenced by
the perceived cost of information search. Consumers are
likely to search for information as long as they believe
that the benefits of acquiring information outweigh the
cost of information search as indicated in “the economics
of information” theory [33].
As a measure of search behavior we followed the
example of Urbany et al. [35] based on Kiel & Layton
[18]. Search behaviour can be measured by the actual
shopping time (AST), and we used actual minutes spent
on search, instead of Urbany et al. [31] who asked buyers
to estimate the total time on shopping. In addition, we
noted the price of the product.
Sub-constructs of search behaviour are the width and
depth of search, meaning the extensiveness of search. The
width of search can be defined as the number of
alternatives considered. The depth of search describes
how many attributes of a product or alternative are
evaluated.
2.1 The costs of consumer information search
Cost of information search in the theoretical
framework is presented by three dimensions: financial
cost, time spent and cognitive effort required. Each
dimension of cost represents a different perspective of
cost. Financial cost represents the amount of money spent
to acquire the necessary information. Time spent refers to
the amount of time required for information search. Effort
refers to the amount of cognitive effort required to
process the information. The first dimension of the
proposed cost of information search construct, financial
cost, was first proposed by the Stigler [33] in the
economics of information theory. The other two
dimensions of the cost of information search, time spent,
and effort required, are mostly utilized in consumer
Proceedings of the 41st Hawaii International Conference on System Sciences - 2008
2
behavior studies conducted mostly in laboratory
conditions [6].
Search theory is rather uniform in its definition of the
implications of search costs on consumer behavior and
price dispersion. Stigler [33] proposed, that high search
costs will lead value maximizing consumers to limit their
pre-purchase search, which results in less than perfectly
informed purchase decisions. Since consumers vary on
their market knowledge and search costs, relatively wide
price dispersions persist in many consumer markets. The
very basis of search theory [33] suggests two of the most
profound measures of search costs: the amount of search
and price for products of comparable quality. These are
the two key measures that this work examines in an
attempt to determine whether electronic consumer
markets have positively affected pre-purchase consumer
search.
3. Search patterns
The economics literature is interested in exploring
optimal search behaviour with the cost – benefit
framework. Most studies see sequential search as the
dominant way of searching. Kohn and Shavell [19] even
define search as “sequential sampling from a population
where the samples could be prices, product features etc.”
The Internet has, however, changed the hegemony of
sequential search, since ways of searching, previously
maybe possible, but difficult, are now made easier for
consumers. In this section, different ways of searching,
mainly in the online settings are examined.
In addition to the two prototypical search patterns,
sequential and simultaneous searching, a third pattern,
iterative search, emerged in our experiment.
3.1 Sequential searching
Sequential search is a process whereby a consumer
wishing to buy one unit of commodity obtains quotations
one-at-a–time until a satisfactory price is obtained. In an
online environment, an example of a sequential search is a
consumer surfing through different Web pages, and
visiting various online-sellers [36], [38].
3.2 Simultaneous searching
According to Stigler [33], search takes place when a
buyer (or seller) wishes to ascertain the most favorable
price, and must thus canvass various sellers (or buyers).
Stigler developed the “economics of information” EoI
theory on the assumption of the so called fixed sample
size (FSS) searching, according to which an individual
obtains all samples at once, and the commodity is
purchased from the seller quoting the lowest price. In
other than the economics literature, FSS searching is also
called simultaneous searching. The essence of
simultaneous searching is that a consumer is able to
evaluate available products side by side.
In offline circumstances, a consumer might collect a
simultaneous sample based on either internal information
formed by experience of repeated purchases (internal
search), or by, for example, acquainting her/himself with
special issues of consumer journals that compare products
the consumer is interested in (external search). In online
settings, a consumer can use various tools (for example,
comparison sites or comparison agents) to collect
information that is available on the Internet on a particular
product or service. According to Whinston et al. [36],
price search in a price database is an example of a
simultaneous search in an online environment.
Electronic and simultaneous search is given a
definition by Öörni [39], [40] the characteristics of which
are i) the information channel is electronic, ii) all the
information is retrieved in a single stage iii) no human
interaction is required. In his empirical research Öörni
[39], [40] found out that the use of electronic and
simultaneous search in the context of travel services was
very rare in the beginning of 2000’s.
Manning and Morgan [22] stated that both
simultaneous search and sequential search may be
considered special cases of a general search pattern,
according to which a searcher obtains more than one
sample at a time and then has to decide how many more
times to sample. Agrawal et al. [1] compared
simultaneous and sequential search, and concluded that
simultaneous search allows for information gathering
quickly, though overinvestment in information gathering
may occur (i.e. the simultaneous sample might be too
extensive). Sequential search, on the other hand, is slow,
but avoids unnecessary information gathering. The
optimal search pattern has been suggested to combine the
speed of simultaneous search with the flexibility of
sequential search to avoid unnecessary costs [1].
3.3 Iterative searching
The possibility to return to price / product information
that was previously searched but not chosen can be called
iterative search. Iterative search allows back-and-forth-
movement as consumers compare product and service
offerings. An iterative search begins just as a sequential
query to the product information. The definition of
iterative search might be sequential search with
recall The query results are compared to each other, and
then results are noted. The difference to sequential search
is that after finding the outputs, consumer will make the
query again, and the process is then repeated.
Proceedings of the 41st Hawaii International Conference on System Sciences - 2008
3
4. Uncertainty and search behavior
One reason consumers search for information prior to
purchase is to reduce uncertainty. Information search is
often seen as a mean to lessen decision-related
uncertainty. Therefore, greater uncertainty should lead to
more extensive search behaviour [20]. Some early studies
of uncertainty constructs and uncertainty dimensions of
“knowledge uncertainty” and “choice uncertainty” have
been done in the sixties and seventies [4], [33], [20], [37],
[5]. However, many researchers have argued that there
may be certain conditions, under which uncertainty
reduces, (instead of increases), search behavior [2], [37],
[5].
Urbany et al. [35] have investigated information search
in the context of consumer decisions. They identified
various forms of decision-related uncertainty, which are
likely to influence the information search in many ways.
Consumer researchers have defined uncertainty in many
different ways, for example, as perceived risk. Urbany et
al. [35] defined uncertainty as the amount of information
the buyer brings to the search process. If a consumer
receives more information before shopping, s/he will have
stronger prior beliefs and less uncertainty about which
store to shop in. This definition is consistent with more
traditional conceptualizations of uncertainty [34]. The
dimensions of uncertainty, proposed by Urbany et al. [35]
provide the central ingredients for our study. We will next
define the two uncertainty dimensions found in the
previous literature.
4.1 Knowledge uncertainty
Knowledge uncertainty (KU) captures doubts
consumers have about their own ability to judge sellers
and products well enough to execute rational product
comparisons. Urbany [34] has defined KU as uncertainty
about the knowledge of the alternatives and variables, i.e.
what is known about alternatives. The original construct
of knowledge uncertainty is from Stigler [33]. KU may
arise from the lack of factual information about
alternative choices and / or uncertainty over what decision
rules are relevant [35]. KU may also be related to
uncertainty over how to acquire the necessary information
to make a choice. Several researchers [35], [5], [33] agree
that the lack of (or uncertainty about) product knowledge
increases search costs and therefore may reduce search.
High KU is associated potentially with reduced ability
to comprehend and efficiently use new information,
which makes information search a more difficult process.
The link between prior knowledge or expertise, search
cost, and search intensity has several proponents [2], [8],
[28]. Experts with lower KU and greater prior knowledge
about a product have a greater capacity for learning new
information and therefore are more likely to search than
non-experts [2]. Consumers might be certain about what
model or brand to choose, and at the same time they
might be very uncertain about the knowledge they hold
about a given product class. In fact, consumers with low
knowledge about the product category might experience a
more difficult search task than consumers with a high
prior knowledge of the product class. Those consumers
higher in KU might search less than those with lower KU.
In sum, if high knowledge uncertainty causes limited
search, the result of that will be both high search costs and
a difficulty of assessing the benefits of the search [35].
4.2 Choice uncertainty
Choice uncertainty (CU) means uncertainty about
which alternative to choose [34], [35]. The original
construct of choice uncertainty is from Lanzetta [20]. It is
interesting how the characteristics in a choice set i.e.,
experienced similarities or differences between the
current choices, contribute to CU. Information search will
be greater when the choice sets are similar, because of the
CU generated [20] In conclusion, Lanzetta argues that a
bigger uncertainty, and on the other hand, a similar choice
alternative set, should result in more executive search
[20]. In contrast, Stigler´s EoI theory and cost - benefit
model predicts [33] that a greater similarity between
choice alternatives / choice-set will reduce search, due to
lower expected gains from search and presumably lower
choice uncertainty.
Sieber & Lanzetta [29] predict that consumers with
less complex conceptual structures might be more likely
to apply well-defined rules to make decisions. They
predict that low CU may result from poor knowledge of
the available choice-set (i.e., poor knowledge of what
alternatives are available). According to Sieber &
Lanzetta [29], consumers who utilize simple conceptual
structures perceive less information and, therefore, they
might experience less choice uncertainty than consumers
who utilize complex structures [29].
CU is more influential than KU. Choice uncertainty
might come from different sources, firstly, a high level of
ignorance about the product or the market place or
secondly, a relatively well-informed base of knowledge
that suggests that there may be yet undiscovered
alternatives [35]. Urbany et al. [35] defined choice
uncertainty as uncertainty regarding which alternative to
choose. According to them [35], choice uncertainty covers
questions such as what and where to buy, and exists as a
separate construct in the consumers’ mind. They have
further proposed that uncertainty related to the selection
of the evaluative means may in fact be a separate entity.
In their study, they found that choice uncertainty
increases search behavior.
Proceedings of the 41st Hawaii International Conference on System Sciences - 2008
4
4.3 Impact of uncertainty on search behaviour
Urbany et al.[35] found that CU increases search
behavior while KU reduces search. They found a strong
relationship between CU and KU, and according to them,
consumers can be high in KU yet low in CU and vice
versa [35].. Therefore, it is possible that high KU may not
always lead to high CU and greater search, even though
CU and KU are positively related [35].. High KU is
associated potentially with a reduced ability to efficiently
use new information, which makes information search a
more difficult process. The lack of (or uncertainty about)
product knowledge increases search costs and therefore
may reduce search [35].
According to Urbany et al. [35], three interesting
results emerged in their study; Firstly, KU and CU are
very strongly related. Secondly, the simple correlations
indicate that both CU and KU are positively related to
search behavior, although the correlations for KU are
smaller. Thirdly, the regression and discriminant analysis
results indicate that KU and CU both have significant
effects on search, but the CU X KU interaction does not
[35].
In the light of these findings, it becomes apparent that
it might be useful to study the relationship between
uncertainty measures and consumer pre-purchase search,
and buying behavior.
5. Method
The effect of individual differences and purchase
situations on search behavior is complex, often interactive
and difficult to interpret and generalize about [35].
Therefore, we chose as cohesive a group as possible for
our observation research. Our response group consisted of
12-15 year old teenagers from the same demographic
area. Our observation situation was the same for every
respondent, interactive purchase via Internet without time
limits.
The method used in this study is empirical
observation. We chose this method in order to find out
what people really do in a search and purchase situation,
instead of just asking what they think they would do. The
more specific description of our method is in our working
papers and former paper of the observation research [21].
We observed and interviewed 56 pupils belonging to age
groups from twelve to fifteen years studying in Espoo,
Finland. We chose this target group because we felt that
pupils have not established ways of searching information
on the Internet. We conducted observations during three
days in April and May 2004 on the school’s premises.
There was always one observer present per pupil. All the
observers were experienced researchers briefed of the
research objectives and methods prior to the experiments.
The observational study was conducted in the
following way: The observer explained the objectives of
the experimental tasks to the pupils who were instructed-
to think aloud, i.e. comment on all their moves and
reasons for the choices while they were searching for
information. Background information on the subjects was
gathered with a formal sheet and we used a standardized
form to record the actions of the subjects. After each
interview, the respective observer went through the results
with the one researcher who was responsible for inserting
the data in a database. Having one person to insert data
was meant to ensure consistent interpretation for all
observations.
5.1 The design of the experiments
We designed three assignments to measure the effects
of knowledge and choice uncertainty on the search effort.
The assignments were simple product search and
comparison tasks during which the subjects were asked to
think aloud their actions and the reasons behind them. The
three assignments were worded as follows:
Assignment 1: Buy a Christmas present CD for Your
grandmother.
Assignment 2: Buy the Red Hot Chili Peppers’ “By the
way”- CD for a friend.
Assignment 3: Buy a CD yourself.
In the first assignment, knowledge uncertainty was
high while choice uncertainty was low. The subjects were
unlikely to be familiar with the music categories searched
for the CD, yet, choice uncertainty was low since the risks
related to an adverse choice were low – the subject would
not be stuck with the record. The second assignment was
designed to have both low knowledge and choice
uncertainty. The music category should be familiar to
most subjects and the task was narrowly framed to lower
choice uncertainty. In the third assignment, knowledge
uncertainty was low because the subjects were
knowledgeable about the music genres of their choice.
Choice uncertainty, on the other hand, was high since
they had the chance to win the record and, therefore, were
at some pressure to make a good choice.
Proceedings of the 41st Hawaii International Conference on System Sciences - 2008
5
6. Results
6.1 Identification of search patterns
Three different search patterns were identified during
the experiment thus confirming our first proposition on
different search patterns that consumers employ in their
pre-purchase search. Out of 168 units of analysis
sequential search was employed 110 times (65%),
simultaneous 37 times (22%), and iterative search pattern
21 times (13%).
6.2 Uncertainty as a determinant of search
pattern employed
According to Urbany et al. [35] KU and CU both have
significant effects on search. According to our second
proposition, search is shaped by the uncertainties related
to the purchase decision. According to this proposition,
KU and CU embedded in the tasks should have an effect
on the search pattern employed. According to the results,
various levels of uncertainties do not seem to have any
impact on the pattern of search employed since the
sequential search pattern is the most usual search pattern
in all tasks, as depicted in Table 3.
Table 1. Number of search patterns used in
different uncertainty tasks
Uncertainty
Task 1
High
KU
Task 2
Low
Uncertainty
Task 3
High
CU
Simultaneous 13 13 11
Sequential 33 38 39
Iterative 10 5 6
6.3 The effect of uncertainty on outcomes of
search
High knowledge uncertainty affects the search effort.
It appears that in some settings high KU promotes search
while in other contexts high KU inhibits search. We did
not take a prior stand on the issue, but accepted that KU
has an effect on the extent of the search effort. We
designed our experiment to include one assignment with
high knowledge uncertainty (1st assignment) and two
assignments (2nd and 3rd) low on knowledge uncertainty
in an attempt to control the effect of choice uncertainty on
search.
We propose to operationalize the effect of KU as the
time spent for search, since time captures the total effort
of search better than most other measures. We claim that
search is shaped by the uncertainties related to the
purchase decision, and formulated more specifically as
follows: Consumers spend more time on search under
high knowledge uncertainty, as follows:
3211
3210
:
:
µµµ
µµµ
=≠
==
H
H
We scrutinized the data to ensure that the tested
variables were normally distributed or did not depart too
markedly from normality. For the paired samples t-tests,
the α-risk was controlled at 0.05 when µ1= µ2= µ3.
Table 2: The effect of uncertainty on shopping
time.
Time
Uncertainty
Task 1
High
KU
Task 2
Low
Uncertainty
Task 3
High
CU
Avg. 6.339 4.188 4.464
Std.Dev. 3.589 3.423 2.593
N 56 56 56
T-tests
Pairs t-value p-value
(2-tailed)
1 & 2 4.583 0.000
1 & 3 3.815 0.000
2 & 3 -0.628 0.533
The actual shopping time varied between 4.19 minutes
to 6.3 minutes. On average, the pupils spent the most
amount of time on the first assignment (6.3 minutes), and
the least on the second assignment (4.2 minutes). The
third assignment took them, on average, 4.5 minutes to
complete.
These figures suggest that KU increases the amount of
time spent on search, for KU was high in the first
assignment. The t-tests suggest that the first assignment
deviated markedly from the later assignments judged by
the amount of time the subjects used to search. This
supports the hypothesis (H1) that high knowledge
uncertainty leads to extended search. The difference in
average times for the second and third assignments was
too small to be statistically significant, which is congruent
with the low KU for the two assignments.
It is also noteworthy, that the standard deviation of the
time spent on the assignments steadily decreased during
the test. We interpret this as a sign of a learning effect.
Proceedings of the 41st Hawaii International Conference on System Sciences - 2008
6
Table 3: The effect of uncertainty on purchase
price.
Price Uncertainty
Task 1
High
KU
Task 2
Low
Uncertainty
Task 3
High
CU
Avg. 15.64 15.68 13.66
Std.Dev. 5.106 15.363 5.604
N 56 56 56
T-tests
Pairs t-value p-value
(2-tailed)
1 & 2 -0.04 0.9651
1 & 3 1.96 0.0530
2 & 3 1.95 0.0533
6.4 The effect of search patterns on outcomes of
search
Our third proposition suggests that search patterns
have an effect on the outcomes of search. We propose to
operationalize the effect of search pattern as the time
spent for search, since time captures the total effort of
search better than most other measures.
We formulated hypothesis 3 so that both search time
and the price at which consumers were able to find a
suitable product are dependent on the search pattern. Our
null hypothesis is that observed shopping time and best
prices found were equal over the tasks, i.e. the amount of
uncertainties did not affect either.
H0 = µ1= µ2= µ3.
H1 = µ1 ≠ µ2 ≠ µ3
Table 4. The effect of search pattern on
shopping time.
Time
Search Pattern
Simul-
taneous
Sequen-
tial Iterative
Avg. 4.49 4.76 7.19
Std.Dev. 2.59 3.41 3.52
N 168 168 168
T-tests
Pairs t-value p-value
(2-tailed)
Sim & Seg -0.50 0.6180
Sim & Iter -3.08 0.0042
Seg & Iter -2.92 0.0068
The actual shopping time varied between 4.49 minutes to
7.19 minutes (see table 5). On average, the subjects spent
the most amount of time when using iterative search
strategy (7.19), and the least when searching
simultaneous (4.49 minutes). The difference in time used
was statistically significant for iterative search when
contrasted to simultaneous or sequential search. Thus, we
conclude that iterative search is the most costly strategy
while sequential and simultaneous search do not
necessarily diverge in this respect
Table 5. The effect of search pattern on
purchase price.
Simul-
taneous
Sequen-
tial Iterative
Avg. 11.70 16.20 14.00
Std.Dev. 5.46 4.93 5.18
N 168 168 168
T-tests
Pairs t-value p-value
(2-tailed)
Sim & Seg -4.67 <.0001
Sim & Iter -1.89 0.0662
Seg & Iter 1.79 0.0837
The actual purchase price in different search patterns
varied from 11.70 euros to 16.20 euros. The pupils spent
the most money when searching with a sequential pattern
(16.20 euros), and the least when searching simultaneous.
The difference in purchase price was statistically
significant for simultaneous search against sequential
search strategy. The difference between iterative search
against the simultaneous and sequential strategies was
statistically non-significant.
Judging by the result of our t-tests, the students spent
the least amount of money when using simultaneous
search strategy, both sequential and iterative strategies
leading to more expensive purchases.
Proceedings of the 41st Hawaii International Conference on System Sciences - 2008
7
6.5 Search performance by uncertainty and
search pattern
We examined next the combined effect of uncertainty and
search pattern on the outcomes of search.
Table 6. Search performance time by uncertainty
and search pattern
Time Uncertainty
Search
pattern
Task 1
High
KU
Task 2
Low
Uncertainty
Task 3
High
CU
Simultaneous 5.54 3.54 4.36
Sequential 5.97 4.14 4.41
Iterative 8.60 7.00 5.00
We found that iterative search pattern is the slowest
under any uncertainty, whereas the differences in time
spent on search between simultaneous and sequential
patterns are not substantial. However, simultaneous
search is somewhat more efficient than sequential search.
Next, the combined effect of uncertainty and search
pattern on price of the purchased product is examined.
Table 7: Search performance price by
uncertainty and search pattern
Average price
Uncertainty
Search
pattern
Task 1
High
KU
Task 2
Low
Uncertainty
Task 3
High
CU
Simultaneous 14.62 11.00 9.09
Sequential 15.64 17.40 15.21
Iterative 15.40 14.00 11.50
The sequential search pattern seems to lead to the
most expensive purchase under any uncertainty. It is
surprising that this search pattern is so common
(employed 65% in our experiment) even though it results
in the most expensive purchase.. Simultaneous search, on
the other hand, leads to most inexpensive purchase but it
was used only 22 % in the experiment. The effect of
search pattern was independent of the related
uncertainties, suggesting that the possible effects of
uncertainty on the outcome of search are relatively small
compared to the effects of the search pattern employed.
Our observation may explain, at least in part, why
electronic markets have not increased market efficiency
as much as expected, i.e. narrow price dispersion and low
average prices. While the benefits of simultaneous search
strategy seem quite apparent, the majority of the subjects
did not exploit them. Consumers must adjust their
behavior to the new environment to realize the potential
benefits.
7. Discussion and Conclusion
In this paper, we have demonstrated that search
pattern employed and search related uncertainties are
concepts that can be used to explain the variation in the
efficiency of consumer search in electronic markets.
Uncertainty has been established as the motive of
consumer search [33]. That is a concept that can be used to
link electronic markets research to economic, consumer
behaviour, and decision-making research facilitating the
creation of a fuller picture of the effects electronic
markets may have on consumer behaviour. The concept
of uncertainty provides us with a coherent theoretical
frame to explore consumer search in electronic markets.
Previously the patterns and extent of consumer search
have been explained by using concepts such as price,
brand, and loyalty. While these concepts are valid, as
such, they share little theoretical ground and it is not clear
how they could be fitted into a framework encompassing
the essential factors of consumer search. In addition to the
two prototypical search patterns, sequential and
simultaneous searching, we used iterative searching
pattern in our study. The sequential searching is still the
predominant way of searching, even among the
youngsters.
Uncertainty, on the other hand, is a concept well
established as the foundation of consumer search. It is
also the prime concept linking consumer search and
decision-making theories. As decision-making is central
to consumer search, it is hoped that uncertainty could be
conceptualized further to create a theoretical frame that
could be used to analyze different decision making stages
in consumer purchasing behavior in electronic markets.
We have operationalized uncertainty with two constructs,
knowledge uncertainty and choice uncertainty. In the
following, we summarize the results of our experiment.
Uncertainty as a determinant of search pattern
employed: According to our data, most people used
sequential search pattern (65%). Theoretically, knowledge
and choice uncertainty should have an effect on the search
pattern employed, but as noted in results, the search
patterns used by the pupils seem not to be dependant on
the uncertainties of the tasks. It is possible that even the
relatively young consumers have pre-existing, well
developed search patterns, and they are reluctant to adjust
their behavior to the new environment to realize the
potential benefits. We found that even relatively young
people tend to adhere to sequential search rather than
simultaneous search pattern.
Proceedings of the 41st Hawaii International Conference on System Sciences - 2008
8
The effect of search patterns on outcomes of search:
Our t-test shows that the actual shopping time varied
between 4.49 minutes to 7.19 minutes according to the
different search patterns. The actual purchase price
according to the different search patterns varied from
11.70 euros to 16.20 euros.
Our observations seem to support the hypothesis (H3) that
selection of search pattern has an impact on the efficiency
of search judged by the amount of time the subjects used
to search.
Uncertainty and search pattern as determinants of
search performance measured by time: Our results
suggest that knowledge uncertainty increases the amount
of time spent on search. As the effects of uncertainty on
search patterns are examined in more detail, the following
results emerged: Iterative search is the most costly search
pattern while sequential and simultaneous search do not
necessarily diverge in this respect. When search patterns
are taken into account, simultaneous search seems to be
the most efficient search pattern when measured by time
spent on search.
Uncertainty and search pattern as determinants of
search performance measured by purchase price:
Simultaneous search seems to be the most efficient
strategy when measured by or purchase price. It would
seem a prudent choice as simultaneous search is the least
costly, judged by the time spent, and tends to lead to
inexpensive purchases. The effect of the pattern employed
was also independent on the related uncertainties,
suggesting that the possible effects of uncertainty on the
outcome of search are relatively small compared to the
effects of the search pattern employed.
Simultaneous search seems to combine low search
costs with high efficiency. The fact that it wasn’t the
strategy of choice for most of our subjects raises the
question of the necessary preconditions to simultaneous
search. It is possible that electronic markets are less
transparent when it comes to search related meta-
information: where to find a suitable search engine and
how to use it. This knowledge must, for the large part, be
extracted at a cost through on-going search.
8. References
[1] Agrawal Manish, Hariharan Govind , Kishore Rajiv , Rao
H.R. (2005). Matching intermediaries for information goods in
the presence of direct search: an examination of switching costs
and obsolescence of information. Decision Support Systems, 41
(1), 20 – 36.
[2] Alba, J.W. and Hutchinson, J. W. (1987), “Dimensions of
consumer expertice”, Journal of Consumer Research, Vol.13
March, pp.411-54.
[3] Beatty, S.E. amd Smith, S.M: (1987), “External search
effort: an invertigation across several product categories”,
Journal of Consumer Research, Vol. 14 June, pp. 83-95.
[4] Berlyne, Daniel E. (1960), Conflict, Arousal and Curiosity,
New York: McGraw-Hill.
[5] Bettman, J. R. (1979a). An Information Processing Theory of
Consumer Choice, Reading, MA: Addison-Wesley.218
[6] Bettman, J. R., Sujan, M. (1987): Effects of Framing on
Evaluation of Comparable and Noncomparable Alternatives by
Expert and Novice Consumers. Journal of Consumer Research
14, 141-154
[7] Bettman, J. R., Johnson, E. J. and Payne, J. W. (1991).
Consumer decision making, Handbook of Consumer Behavior,
1991
[8] Brucks. M, The Effects of Product Class Knowledge on
Information Search Behavior, Journal of Consumer Research,
June 1985, vol. 12
[9] Chiang K-P., (2006), Clicking Instead of Walking:
Consumers Searching for Information in the Electronic
Marketplace, Bulletin, December/January 2006
[10] Copeland, Melvin T. (1917). Relation of consumers buying
habits of marketing methods.Harvard Business Review, 1
(April), 282-289.
[11] Duncan, Calvin P. and Olshavsky Richard W. (1982)
“External search: The role of consumer beliefs,” Journal of
Marketing Research, 19 (Feb), pp. 32-43.
[12] Engel, J.F., Blackwell, R.D. and Miniard, P.W. (1990)
Consumer Behavior, Chigago: The Dryden Press, 6th edition
[13] Hoque, A. Y., & Lohse, G. L. (1999). An information
search cost perspective for designing interfaces for electronic
commerce. Journal of Marketing Research, 36, 387-394.
[14] Howard John A. and Jadish N. Sheth (1969), The Theory of
Buyer Behavior. New York, NY: Wiley.
[15] Jansen B. J. Paid Search, IEEE Computer, July 2006
[16] Jansen, B. J., & Pooch, U. (2001). Web user studies: a
review and Framework for future work. Journal of the American
Society of Information Science and Technology, 52(3), 235-246.
[17] Johnson, E. J., Moe, W., Fader, P., Steven, B., & Lohse, J.
(2004). On the depth and dynamics of online search behavior.
Management Science, 50, 299-308.
[18] Kiel, G.C. & Layton, R.A (1981), “ Dimensions of
Consumer Information Seeking Behavior”, Journal of
Marketing Research, 18 (May), pp.233-239.
Proceedings of the 41st Hawaii International Conference on System Sciences - 2008
9
[19] Kohn M.G., Shavell S (1974). The theory of search,
Journal of Economic Theory 9, 1974, pp. 93– 123.
[20] Lanzetta, John T. (1963), “Information Acquisition in
Decision - Making,” in Motivation and Social Interaction-
Cognitive Determinants, ed., O. J. Harvey, New York:Ronald,
pp.239-265
[21] Lauraeus-Niinivaara Theresa, Saarinen Timo, Öörni Anssi
(2007), Knowledge and Choice Uncertainty Affect Consumer
Search and Buying Behavior, HICSS Conference 3.-7.1.2007
[22] Manning R, Morgan P (1982). Search and Consumer
Theory. Review of Economic Studies 1982, XLIX, 203 - 216
[23] Marmorstein, H., Grewal, D. and Fishe, R. P. H. (1992).
The value of time spent in price comparison shopping: Survey
and experimental evidence. Journal of Consumer Research, 9
(June), 52-61.
[24] McCleary, K. W. and Whitney, D. L. (1994). Projecting
western consumer attitudes toward travel to six Eastern
European countries. Journal of International Consumer
Marketing, 6 (3/4),239-256.
[25] Moorthy S, Ratchford B, Talukdar D. Consumer
information search revisited:theory and empirical analysis. J
Consum Res 1997;263 – 77 (March).
[26] Newman, J. and Staelin, R. (1971), “Multivariate analysis
of differences in buying decision time”, Journal of Marketing
Research, Vol. 8 May, pp. 192-8.
[27] Olshavsky, R. W. (1985). Towards a more comprehensive
theory of choice. In E. Hirschman and M. T. Holbrook (Eds.),
Advances in Consumer Research, Vol. 12. (pp. 465-470).
[28] Punj,G. and Staelin,R. (1983): A Model of Consumer
Information Search Behavior for New Automobiles, Journal of
Consumer Research, March1983, vol. 9
[29] Sieber, Joan E. and John T. Lanzetta (1964), “Conflict and
Conceptual Structure as Determinants of Decision making
Behavior, “Journal of Personality, 32 (4), 622-641.
[30] Smith, J. B. and Spreng, R. (1996), “A proposed model of
external consumer information search”, Journal of the Academy
of Marketing Science, Vol. 24 Summer, 246-56.
[31] Spink, A., Koshman, S., & Jansen, B. J. (2005).
Multitasking on the Vivisimo Web Search Engine. IEEE ITCC
International Conference on Information Technology,.2005,LA
[32] Srinivasan, N. (1990). Pre-purchase external information
search for information. In V. E.
[33] Stigler, George J. (1961), “The Economics of information,”
Journal of Political Economy, 69 (June), 213
[34] Urbany, Joel E., (1986), “An Experimental Examination of
the Economics of Information”, Journal of Consumer Research ,
Vol 13, Sep 1986
[35] Urbany.J, Dickson.P, Wilkie.W: “Buyer Uncertainty and
Information Search”, Journal of Consumer Research,
September 1989, vol. 16
[36] Whinston, A. B., D. O. Stahl, et al. (1997). The Economics
of Electronic Commerce. Indianapolis, Macmillan Technical
Publishing.
[37] Wilkie, William L.(1975),”How consumers use product
information: An assessment of research in relation to public
policy needs, Washington,D.C.:National Science
Foundation.1985
[38] Wu, D., Ray, G., Geng, X., & Whinston, A. (2004).
Implications of reduced search cost and free riding in e-
commerce. Marketing Science, 23, 255-262.
[39] Öörni, A, Consumer Search in Electronic Markets,
European Journal of Information Systems, (2003) 12, 30-40
[40] Öörni, A ,(2002) Dominant search pattern in electronic
markets: Sequential or simultaneous search, HSE Working
papers, W-31
Proceedings of the 41st Hawaii International Conference on System Sciences - 2008
10
