A FORAGING MODEL FOR HUMAN BEHAVIOR

1
A FORAGING MODEL FOR HUMAN BEHAVIOR IN
E-COMMERCE Dissertation Proposal by Bjarne
Berg-Saether Committee members Dr. Antonis
Stylianou, Chairperson Dr. John Brzorad Dr.
Heather Lipford Dr. Sungjune Park Dr. Dmitry
Shapiro Dr. Kexin Zhao
2
Agenda
Introduction Approach Literature review Optimal
Foraging Theory Human Cognitive Factors The
Economic Perspective Developing a Behavioral
Foraging Model Theoretical Model and
Hypotheses Theoretical Components Methods The
Experiment Instrument Sample Population Question
s and Answers
3
Introduction

• Human interaction with the Internet has been
  studied from a variety of standpoints. Some have
  looked at technology adoption, optimal usage and
  cost/benefit functions, while others have
  examined the Internet from social, psychological,
  and interface standpoints. However, few have
  examined Internet usage at the individual level
  from a combined biological and economic
  perspective (Pirolli, 2003).
• It is likely that research into human foraging
  behavior on the Internet will lead to improved
  understanding of the reasons why some e-commerce
  sites succeed and other fail, how participants in
  the marketplace select a certain number of
  patches (web sites) as their local foraging area
  and often limit their searches to these patches.
• It may also provide better understanding of how
  humans expand and contract their foraging areas
  (collection of web sites) and how they decide
  when to stop foraging and start executing
  purchase transactions.

4
Introduction Example of Information Foraging
In this example we are searching for tickets
between Washington and Paris. We want to make
the round-trip in less than 24 hours total
traveling time, and as cheap as possible.
There are 689 available flights/routes. Sorting
by Duration, we find that there are 237 flights
that that has a total travel time of less then 24
hours (round-trip).
5
Introduction to Information Foraging
We found that ticket prices range from 719 to
4,384 for our flight. The order which the 237
flight prices are listed is illustrated in this
graph.
Charnov's Marginal capture Theorem (Pirolli, 2007)
6
Agenda
Introduction Approach Literature review Optimal
Foraging Theory Human Cognitive Factors The
Economic Perspective Developing a Behavioral
Foraging Model Theoretical Model and
Hypotheses Theoretical Components Methods The
Experiment Instrument Sample Population Questions
and Answers
7
Overall Approach

• The proposed research consists of
• 1. Building a theoretical model
• 2. Test the proposed relationships
• 3. Build a simulation model
• 4. Simulate changes in characteristics and
  explore asymptotes and economic aspects of the
  foraging model

8
Agenda
Introduction Approach Literature review Optimal
Foraging Theory Human Cognitive Factors The
Economic Perspective Developing a Behavioral
Foraging Model Theoretical Model and
Hypotheses Theoretical Components Methods The
Experiment Instrument Sample Population Questions
and Answers
9
Optimal Foraging Theory

• In 1966 the field known as Optimal Foraging
  Theory (OFT) was established through the
  publication of Emlens article on foraging
  behavior of birds and by MacArthur and Piankas
  work, published the same year, on optimization
  models.
• In general, the models established over the next
  ten years focused on four core areas that became
  known as elements of a micro-ecological theory.
  Micro-ecology is defined by Merriam-Websters
  dictionary (2003) as a branch of science
  concerned with the interrelationship of organisms
  and their environments. For foraging theory it
  refers to the study of
• 1) What to eat (optimal diet)
• 2) Where to eat (optimal patch choice)
• 3) Optimal allocation to each patch (time)
• 4) Optimal patterns and speed of movements.

10
Optimal Foraging Theory

• Combined as a whole, the micro-ecological theory
  forms the platform for macro-ecological theory,
  which has far reaching implications.
• Presumably, over time through natural selection,
  parents that are successful foragers will pass
  some inheritable traits to their offspring and,
  thereby, evolve better foragers adapted to the
  environment.
• This also implies that in a stable environment,
  such an evolution has already occurred in the
  past and that the average observed foraging
  behavior of the current foragers is already close
  to the optimal foraging behavior.
• If the environment changes, as it always does,
  the unsuccessful tails of the distribution will
  become the new norm in theory. This is known
  as saturation of the behaviors.

11
Agenda
Introduction Approach Literature review Optimal
Foraging Theory Human Cognitive Factors The
Economic Perspective Developing a Behavioral
Foraging Model Theoretical Model and
Hypotheses Theoretical Components Methods The
Experiment Instrument Sample Population Questions
and Answers
12
Human Cognitive Factors

• Classification of HCI research (Zhang and Li,
  2005). Based on our literature review of
  Cognitive traits and Demographics we look at a
  set of HCI factors that relate to foraging at
  the individual level.

(Pirolli, 2007 Newell, 1990)
13
Demographical factors - Education

• Education is a significant factor in technology
  system interaction. A study by Argarwal and
  Prasad (1999) found education to be closely
  related to the beliefs about both technology and
  actual usage.
• Others have found that a low level education not
  only acts as an access barrier to technology, but
  also impacts what services are being used and how
  they are used (Larsen and Rainie, 2002).
  Specifically, higher education has been found to
  be linked with more technology usage (duration),
  more exploration (search), and earlier adoption
  of new technology products and services. These
  findings have held very consistent over time.
• For example, in an experimental study of 100
  individuals, the education level was found to be
  a significant factor measuring user competence
  (Munro et al., 1997).
• A correlation between a users general
  educational level and technology usage has been
  confirmed in longitudinal observations. In a
  study of Internet usage over a five year period,
  Losh (2003) found that while Internet usage
  increased for all education levels in the period,
  higher educated individuals tend to use
  technology for more diverse tasks and for longer
  periods.
• These research findings infer that higher
  educated users tend to exhibit increased
  technology awareness, increased computer skills,
  and is a factor that should be included in our
  research effort.

14
Demographical factors - Gender

• Gender is one of the validated measures of the
  UTAUT (Venkatesh et al., 2003), and has been
  consistently validated in HCI in areas such as
  TPB. In a study on specific gender differences in
  system usage, using TPB as their framework, the
  authors conclude their findings by stating
• Clearly, gender shapes the initial decision
  process that drives new technology adoption and
  usage behavior in the short-term, which in turn
  influences sustained usage, thus establishing
  that early intentions formed by women and men
  will have a lasting influence on their usage of
  the said new technologyit is critical to
  recognize that the underlying drivers of these
  stable early intentions are different for women
  and men.
• Gender differences were observed even when key
  potential confounding variables (i.e., income,
  organization level, education, and computer
  self-efficacy) were taken into account
  (Venkatesh et al., 2000, p. 50).
• Studies have also found gender differences in use
  of technology. Colley and Matlby (2008) found
  that females are more likely to buy travel
  on-line (6 for females vs. only 1.5 for men),
  and to conduct general information research
  on-line (36 for females vs. only 25.5 for men).
  
• Due to these consistent findings in the
  literature, we will propose to examine gender as
  one factor in our study.

15
Demographical factors - Age

• Age is an important factor in many research
  papers published. Age has consistently been found
  to have an impact on behavior in online settings
  (Zaphiris and Sarwar, 2005). The reasons for
  differences in computer behavior of older users
  were found to be
• a) Vision (decline in static acuity, dynamic
  acuity, contrast sensitivity, color sensitivity,
  sensitivity to glare, decrease in visual field,
  and decrease in ability to process visual
  information).
• b) Psychomotor abilities (mouse movements and
  typing)
• c) Attention (declines in selective and divided
  attention),
• d) Memory and learning (decline in the ability to
  process items from working memory into short term
  memory decline in episodic memory (memory for
  specific events) and decline in procedural memory
  (memory for how we carry out tasks).
• e) Intelligence and expertise (decline due to
  memory loss)

16
Demographical factors - Age

• In a follow-up to these research findings on age,
  Zaphiris et al. (2007) also noted that cognitive
  impairments were often subtle in nature, gradual
  and often not visible (i.e., gradual memory loss
  and gradual processing capability loss).
• Furthermore, aging is also sometimes accompanied
  by significant changes in personality,
  independently of changes in cognitive abilities,
  and these may represent barriers to the
  productive use of information technology
  (Cuttler and Graf, 2007).
• The personality changes manifest themself by a
  propensity to have smaller social networks online
  and in person (Pfeil et al., 2008), access fewer
  on-line sites and have more hierarchical
  organized on-line communication with few central
  sites and few people when engaged in
  communication (Zaphiris and Sawar, 2006). In
  short, a smaller breadth of experiences and
  contacts.
• Research has also revealed that there are
  differences not only among teenagers and very old
  individuals but also within the middle age-range
  group as well. When examining thirty young adults
  (age 20-29) and thirty middle-aged adults (46-59)
  and their interaction with portable multimedia
  players (radio, audio, video), it was found that
  the middle aged group had much more rigid
  exploration of new features, made more mistakes,
  demonstrated less exploration, and reported a
  higher workload for the same tasks as those
  completed by the younger group (Kang and Yoon,
  2008).
• Since the age findings in the HCI literature are
  consistent, we include age as a factor in our
  study.

17
Agenda
Introduction Approach Literature review Optimal
Foraging Theory Human Cognitive Factors The
Economic Perspective Developing a Behavioral
Foraging Model Theoretical Model and
Hypotheses Theoretical Components Methods The
Experiment Instrument Sample Population Questions
and Answers
18
Do people have rational expectations?

• Foragers make purchasing decisions without
  perfect knowledge of prices and price
  distributions and continuously estimate what the
  probabilities are of
• a) Finding a similar items (that meets the need)
• b) Finding it in a reasonable time (foraging
  costs)
• c) Finding the item at a lower cost that offsets
  the foraging costs.

19
Do people have rational expectations?

• In general, the Rational Expectations Hypothesis
  (REH) assumes that the average expectation of
  foragers provides the best point estimate of
  price, uses all available information and is
  non-biased (Muth 1961).
• Specifically, REH suggests that the expectations
  are identical to the best guess of the future
  (Sargent, 1993), and that peoples rational
  expectations do not differ systematically. That
  means that any errors in prediction of prices and
  foraging costs are random.
• The work on REH gained recognition when the
  future Nobel laureate Robin Lucas published his
  paper on the effects of unexpected changes to
  monetary policies (1972). Specifically, he found
  that it was not the change that mattered is was
  the change relative to the expectations of
  participants that had the greatest influence on
  markets (expectations drive behaviors).
• For our research, all else held constant, the
  prevailing price of an item in a market should be
  equal to the local sales price plus the foraging
  cost of acquiring the item at another location.
  Foragers should have rational expectations that
  leads them to make rational foraging decisions
  (any errors are due to mistakes and occurs
  randomly).

20
Why Forage more than once?

• The core benefit to continued foraging after an
  item that satisfies a need has been identified,
  is the ability to find better items, or items at
  a lower cost.
• It is this variability of costs and benefits that
  drives the foragers behavior.
• If the variability of prices was zero, the
  forager would not forage at all and word rely on
  the preciously gathered information for all
  events.

21
Price distributions can be calculated
(Pirolli, 2007)

• Predicted price distribution using Pirolli's
  lognormal probability density function (graphed
  3,666 prices, in increments of one from 719 to
  4,319)
• Actual observed prices in example (graphed the
  237 ticket prices, from our example, from 719
  to 4,319) - note that makes the graph
  non-continuous

22
Do people have rational expectations?

• Using Piroelli's equations, we can find the
  expected value for the plane ticket we started
  this session with. Based on our data the
  predicted mean price of this plane ticket is
  1,302.44
• Notice that the predicted price is extremely
  close to the actual average price of the 237
  plane tickets provided in our search 1,322
  (98.5 accurate).
• The simulation part of our experiment will
  examine how good people are at estimate prices
  and price distributions (needed to optimize
  foraging and decide when to buy and when to
  abandon a web site).

23
Agenda
Introduction Approach Literature review Optimal
Foraging Theory Human Cognitive Factors The
Economic Perspective Developing a Behavioral
Foraging Model Theoretical Model and Hypotheses
24
The Proposed Model
25
Summary of Hypotheses
26
Cognitive Absorption and Exploration Behavior

• From the HCI literature we define Cognitive
  Absorption (CA) as a state of deep involvement
  with software (Agarwal and Karahanna, 2000 p.
  673), while exploration behavior is defined as
  an individuals motivation to investigate his
  surroundings (Houghton Mifflin, 2008).
• Since cognitive absorption can manifest itself as
  exploration within a site (many page visits) or
  exploration in larger contexts (more sites), we
  measure the exploration behavior as both the
  number of sites an individual visits (exploration
  breath) as well as the number of pages the
  individual is accessing (exploration depth).
• Curiosity (cu), a key component of CA is also a
  key element of exploration. We therefore expect
  that a high CA score is positively related with
  the exploration behavior construct. Other CA
  components also suggest this relationship. For
  example, Kao et al. (2008) found that individuals
  engaged in a concrete task (instead of abstract
  issues) exhibited a high degree of focused
  immersion (fi). This is important since Huang
  (2003) found that the immersed attention of an
  individual is positively linked with utilitarian
  web performance and task performance such as
  buying (as opposed to hedonic performance which
  refers to other usage such as entertainment).
  Huang also found close links between CA
  components such as control (cd), curiosity (cu)
  and utilitarian/task performance (i.e., searching
  for an item). In addition, Liaw and Huang (2006)
  found that another CA component, heightened
  enjoyment (he), as reported by 116 individuals,
  was closely linked with their perception and use
  of search engines (exploration behavior). These
  consistent cognitive findings lead us to the
  following hypotheses

27
Summary of Hypotheses
28
Cognitive Playfulness and Exploration Behavior
Cognitive playfulness (CPS), describes an
individuals tendency to interact spontaneously,
inventively, and imaginatively with
microcomputers (Webster Martocchio, 1992).
These researchers argued that when left to
their own devices, individuals high in cognitive
playfulness are more likely to explore the
features of computers than individuals low in
playfulness (p. 761). In a follow-up study by
the same authors (1995), the researchers found
that involvement in a playful, exploratory
experience is self-motivating because it
encourages repetition (more exploration). These
findings are supported by Agarwal and Karahanna
(2000) who found that CPS has a positive effect
on usage of information technology since
individuals with high CPS scores also had a high
degree of self-motivation. Specifically, the
researchers found that the more playful you are,
the more likely you are to enter into a state of
deep involvement with software (exploration).
This leads us to hypothesis number two
29
Summary of Hypotheses
30
Personal Innovativeness and Exploration
Behaviors
Innovativeness is a key cognitive characteristic
of individuals. Agarwal and Karahanna (2000)
found that a person with a high Personal
Innovativeness with Information Technology (PIIT)
score, is more likely to enter into deep
involvement with software and be more likely to
explore. Others found that such individuals
were faster to adapt to technology environments
and were more likely to experiment with
technology (Agarwal and Prasad, 1998). One
area of uncertainty is the propensity to explore
different patches (web sites). One could argue
that a person with a high PITT score (above
average) would be likely to explore a higher
number of patches, while an equally strong case
could be made that a person with a high
innovativeness would explore a single site in
more innovative ways (visit more pages),
resulting in fewer patches being visited. For
example, the relationship between innovativeness
and number of patches could manifest itself as
innovation within sites, (resulting in fewer web
sites being visited), or as innovativeness with
the Internet (resulting in more sites being
visited). The relationships between PITT and
exploration behavior therefore lead us to
hypothesis number three
31
Summary of Hypotheses
32
Computer Self-Efficacy and Exploration Behaviors
Computer Self-Efficacy (CSE) is defined as a
judgment of ones capability to use a computer
(Compeau and Higgings, 1995). CSE is basically an
extension from Bandura's (1971) Social Cognitive
Theory (SCT) from psychology that emphasizes the
social origins of human behavior, rather than
environmental influences. One key element of CSE
is that beliefs and self-reflections about ones
own capabilities drive behavior. A person who
believes that they have good computer skills is
more likely to engage in targeted computerized
foraging efforts, while individuals who believe
that they have weak computer capabilities are
likely to extensively engage in exploration by
visiting more web sites and more web pages before
making a purchase The link between CSE and
exploration has been validated many times in
computer training (Compeau and Higgings, 1995
Compeau et al., 1999 Bolt et al., 2001 Hasan
and Ali, 2004), in Internet bank usage (Chan and
Lu, 2004), in cultural studies (Pearson, 2004),
and in technology adoption (Agarwal and
Karahanna, 2000). Others who examined web
searching behavior also found that the propensity
to search was significantly related with an
individuals CSE score (Kuo et al., 2004). It is
likely that an individual with a high CSE has
strong beliefs in his own exploration skills with
computers. Therefore we expect an individual with
a high CSE score to visit fewer sites and also to
examine fewer pages (since the individual
believes that he already knows how to find the
best items), than individuals with low CSE
scores. This leads us to hypothesis number four
33
Summary of Hypotheses
34
Age and Exploration Behaviors
Li and Chatterjee (2005) found that younger
individuals were more likely to explore more
sites when buying than older individuals. The
differences in exploration behavior are sometimes
due to cognitive impairments that are often
subtle in nature, gradual and often not visible
(i.e., gradual memory loss and gradual processing
capability loss), and, thereby, an unwillingness
to learn new sites and explore more web pages.
Aging is also sometimes accompanied by
significant changes in personality, independently
of changes in cognitive abilities, and these may
represent barriers to the productive use of
information technology (Cuttler and Graf, 2007).
The personality changes manifest themselves by
a propensity to have smaller social networks
online (Pfeil et al., 2008), access fewer on-line
sites (less exploration) and have more
hierarchically organized on-line interaction with
few central sites (Zaphiris and Sawar, 2006). In
short, a smaller breadth of exploration and
experiences. Research has also revealed that
there are differences not only among teenagers
and very old individuals but also with middle age
users. Kang and Yoon (2008) found that middle
aged people were much more rigid in their
exploration of new technology and demonstrated
less propensity to explore, than younger
individuals. Research has consistently shown that
age-related differences exist in exploration, and
that increased technology experience is unable to
offset the effects of aging (Czaja and Sharit,
1997). We therefore expect exploration behaviors
to be related to the participants age, leading
us to a set of hypotheses relating to age
35
Summary of Hypotheses
36
Gender and Exploration Behaviors
Purchasing behavior is also different across
genders. Imhof et al. (2007) found that males
tend to search the Internet more intensely than
females, visit more sites and also use the
Internet substantially more for shopping than
females. This research is supported by
Nielsens Marketing Research (2002) which found
that males visited 801 web pages each week, while
females visited only 573 web pages in an average
week. These findings appear to be relatively
consistent in the literature. For example, Ono
and Zavodny (2003) found that females spent
about the same amount of time on-line as males,
but accessed fewer web sites. In general,
females appear to be more task oriented when
on-line and use the Internet to buy (not shop),
while males are more likely to use the Internet
for exploration and entertainment (Shaw and Gant,
2002). This implies that females view shopping
as a social activity, a need that is not met by
the Internet, while males view shopping as an
exploration activity resulting in more web sites,
more web pages being accessed and more shopping
overall (Kennedy et al., 2003). This leads us to
hypothesis number six
37
Summary of Hypotheses
38
Education and Exploration Behaviors
Education is also a factor in exploration
behaviors. For example, Larsen and Rainie (2002)
found that a low level education not only acts as
an access barrier to technology, but also impacts
what services are being used and how they are
used. Specifically, higher education has been
found to be linked with more usage and more
exploration. This link has held true over time.
Munro et al. (1997) found that the general
education level was significantly correlated with
computer usage (breath of experiences and
willingness to explore new technology) and Losh
(2003) found that, while Internet usage has
increased for all education levels over time,
higher educated individuals tend to use
technology for more diverse tasks than users with
lower education levels, indicating that higher
educated users are more likely to exhibit
different exploration behaviors than those with
less education. The reason for the lower
propensity to explore among those less educated
may be grounded in their belief of the usefulness
of the Internet overall. For example, Zhang
(2005) found that users with bachelor degrees
expressed a belief that the Internet was much
more useful than those with only high-school
diplomas. This link between levels of education
and exploration behaviors leads to a set of
additional HCI hypotheses
39
Summary of Hypotheses
40
Exploration Behavior and Likelihood of
Surrender/Acquisition
Naturally, as the number of web sites accessed
increases, the likelihood of buying from any
given site decreases at a rate of 1/N, we
therefore do not believe this is an interesting
relationship despite being part of our model.
However, there is an interesting relationship
between the number of web pages accessed and the
likelihood to buy from a given site. Li et al.
(2002) explored 132,000 web sessions of 300
people and found that the average number of pages
viewed during a retail web session was 12.09, but
the variance was very large. A quarter of the
users viewed two pages or less, while the median
was 5 pages and the standard deviation was 28.23.
This indicates that without more information, we
can only conclude (95 confidence) that the
average person accessing retail sites will look
at between one and 68.5 web sites, hardly a good
predictor. However, when analyzing only the
lowest 75 percentile of the users, we find that
this group will access only 13 or less web pages
in a retail session. In short, there is a large
variance in web page access, but this variance is
mostly due to a few individuals who do extensive
shopping. We suggest that there is a non-linear
relationship between site surrenders and the
number of pages accessed. For example, if a
person accesses only a few sites, it may be an
indication that he is not a committed buyer,
while a person that accesses a very high number
of web pages may simply be gathering information
and is not planning to make a purchase either
(does not have to buy).
41
Exploration Behavior and Likelihood of
Surrender/Acquisition
For example, Suh et al. (2004) explored over 1
million web sessions of 166,794 users accessing
an eCommerce site and found that while 14
different web pages could be accessed in 175
different patterns before purchases were made,
the best predictor for a buyer was only a viewing
of 7 web pages (and seven access patterns). Web
users that accessed fewer, or more web pages,
were less likely to buy anything. These
findings are highly consistent with Li and
Chatterjees research (2005) that examined 11,139
web user sessions at Barnes and Nobles web site
and found that the committed buyer views a
significantly higher number of pages than
non-committed buyers (8.75 pages for committed
buyers). However, they also found that the
higher the number of pages a person had to visit
before a purchase could be executed, the more
likely the shopper was to abandon the site,
indicating a non-linear relationship between web
pages visited and the likelihood of surrender.
This leads us to hypothesis number eight
42
Summary of Hypotheses
43
Perceived Usefulness based on Previous
Experience and Foraging Time
King and He (2006) did an article review and
found 140 relevant technology adaptation and
assessment articles published in 22 journals
between 1989 and 2005 and found that it appears
to be a substantiated relationship between PU and
actual usage. Therefore, we expect to find that
the PU based on prior experiences is related to
the time a person is willing to devote to
reviewing items offered at a web site. For
example, if a person perceives the web site to be
very useful based on prior experiences, we expect
that the user will spend more time reviewing
items at the site, than on sites that are
perceived to be less useful. It may be
important to separate the time that a person has
direct control over from the time that is
controlled by the site. For example, a person has
little control over the time it takes to access
(load) a web page, nor the time that is required
to orient, enter a search, execute the search and
to finalize a purchase (some control can be
exerted if the user choses to abandon the site).
However, a person does have substantial control
over the time he is willing to spend reviewing
items presented by a search. We propose that the
key to understanding how much time a person is
willing to spend on the review of search results
and overall foraging is determined by the
perceived usefulness of the site based on prior
experiences, leading us to hypotheses
44
Summary of Hypotheses
45
Patch Exhaustion and Foraging Time
Patch exhaustion is a process where individuals
use their previous experience to solve a foraging
problem. For example, patch exhaustion predicts
that a person buying an airline ticket will first
access a website that they are most familiar
with. If unsuccessful at this site, they will
proceed to the site they know second best and try
there (Smith and Dawkins, 1971). The sites
(patches) are accessed in order of familiarity
and only when known sites are fully explored
(exhausted) will the person try new sites. It
is important to note that it is not the
experience that matters but the Perceived
Usefulness (PU) based on the previous experience.
46
Summary of Hypotheses
47
Patch Exhaustion and Foraging Time
Previous experience and pre-existing knowledge
has been shown to restrict foraging behaviors of
individuals. Experiments in birds discovered that
they were good at selecting optimal patches based
on controlled food availability. When the best
patch was swapped with the second best (food was
moved), birds quickly identified the new best
patch. However, when the worst patch was made
the best, birds settled for the second best patch
as determined by their previous foraging
experience and took a long time to identify the
new best patch (Smith and Dawkins, 1971).
This demonstrates a learning behavior that can
work against an individual relying on past
experiences (Smith and Sweatman, 1974). While
creating distortion in the foraging surplus (s)
in the short run, the behavior is actually a
foraging optimization method. The birds simply
assume that a previously poor patch would be
unlikely to become the optimal patch in a short
time period. By using this probabilistic model,
the birds simply moved to the second best patch
once food had been removed from the best patch.
Due to the relocation costs (ta and to), the
search costs (ts), and the low probability of
fundamental changes in the environment, birds do
not conduct a full survey of all patches, thereby
optimizing their patch selection through the
reduction of foraging costs (Smith and Dawkins,
1971).
48
Patch Exhaustion and Foraging Time
In a foraging environment where information
technology is used, this is analogous to web site
selections. Once a person is unable to find the
item he is searching for (at a reasonable price),
the person may migrate to another web page that
he is already familiar with. Only when the
familiar sites are exhausted, or substantially
reduced, will the person start a new search of
web sites that may meet his requirements. This
may occur even if a new site is now the optimal
site. The person may simply perceive that the
search costs (ts) or the orientation costs (to)
are too high, or unlikely to exist outside
previously explored patches (p). While this
behavior is predicted by Optimal Foraging Theory,
it has not been tested in an information
technology setting. This is a very important,
since the theory predicts that increased
knowledge and experience may work against a
person in an environment where rapid changes
occur. We propose that e-commerce participants
leverage the same foraging optimization as
animals and use similar implied probabilities
and, therefore, are willing to devote more time
to the sites that are accessed first than
subsequent sites. Again, we propose that the
time a person is willing to spend reviewing the
items at a site may be different from the overall
time a person is willing to spend at a site,
since a person has a significant control over the
first aspect, but little control over the latter.
As a result, we examine these time groupings
separately and expect that time spent at a site
will be related to site exhaustion (the order a
site is accessed is related to the time a person
is willing to spend at the site), leading us to
the following hypotheses
49
Summary of Hypotheses
50
Foraging Time and Surrender/Acquisition
51
Foraging Time and Surrender/Acquisition
52
Summary of Hypotheses
53
Information Load and Surrender / Acquisition
There is an information load perspective to the
concept of foraging. For our purposes we are
examining the information load as the number of
items presented within a site by a given search.
For example, a recent search for the term
airline ticket on Google returned over 23.9
million results. It is unreasonable to expect a
person to review all results before making a
decision. Likewise, it is hard for a person on a
travel site to review a high number of items
returned from a given search. Sweeney and
Crestani (2006) suggest that the maximum number
of items that a person searching for something is
willing to examine is between 10 and 20, before a
new search is entered or alternative search
methods or sites are utilized. In their research
they asked participants to find newspaper
articles in the Wall Street Journal based on a
set of criteria. The search results were
presented on various electronic devices (PDA, PC,
phones) and were summarized differently. The
authors discovered that the device or the
summarization had little effect, but the number
of items presented for a given search had a
significant effect on the participants ability
to find information. A very high number of items
returned by a search may actually be providing so
much information that it is no longer useful to
the audience.
54
Information Load and Surrender / Acquisition
Biology research has established that a major
impact on an individuals likelihood to surrender
a site is the abundance of relevant items at a
patch. The number of available items is a
factor even when the number of items required is
less than the number available (Pyke et al.,
1977, 1984). As a result, Optimal Foraging
Theory predicts that a minimum number of items
should be available if the person is to continue
foraging at a web site, even when those few items
available meet, or exceed, the needs of the
individuals. As noted before, there are also a
maximum number of items that a person is willing
to explore as well. We, therefore, expect a
non-linear relationship between the number of
items provided for review and the likelihood of
surrender. This leads to hypothesis number
twelve
55
Summary of Hypotheses
56
Agenda
Introduction Approach Literature review Optimal
Foraging Theory Human Cognitive Factors The
Economic Perspective Developing a Behavioral
Foraging Model Theoretical Model and
Hypotheses Theoretical Components Foraging
Surplus Usage Consolidation and Surrender /
Acquisition Methods The Experiment Instrument Sa
mple Population Questions and Answers
57
The Foraging Surplus Identity
58
The Foraging Surplus Equation

s Number of searches at a patch te Time to
enter search tf Time to find (execute
search) to Time to orient E cost per unit of
time p number of patches accessed ta
Time to access patch to Time to orient at
patch ts Time to search tz Time to acquire
item q Number of items returned for review
tr Time to review each item
59
Usage Consolidation

• When there is an abundance of items available for
  consumption, research in biology has demonstrated
  an increase in specialization of consumption
  (Emlen, 1975). This preference assists us in
  understanding phenomena such as brand preference
  and site preference. As items are available in a
  variety of patches, previous experiences or brand
  recognition yields less variety in consumption.
• This has been extensively researched in biology
  (Custard, 1976) where the Optimal Diet Theory
  (ODT) model predicts that increased food
  abundance leads to greater food preference (Pyke
  et al., p141). For example, the Koala bear in
  Australia eats only Eucalyptus leafs, not because
  of a lack of other food sources, but because of
  the abundance of Eucalyptus trees in its habitat.
  
• We extend this concept to usage consolidation and
  define this as the number of sites accessed
  divided by the number of available sites for a
  given task, product or service.
• There are several implications of this food
  specialization/user consolidation. In our case,
  ODT predicts that the more web sites are
  available for a given item, the more
  consolidation of use we should see.

60
Usage Consolidation

• Further support for this idea is found in the
  Optimal Foraging Theory (McArthur and Wilson,
  1967 For example, if there are only three travel
  web sites, we would expect little preference in
  use among the users (a relatively uniform
  distribution of the user base). As the number of
  travel web sites increases, the ODT predicts that
  consolidation of usage occurs (compression). The
  detailed nature of this relationship is not
  extensively explored in the IT field, except when
  observed in the aggregate. For example, in
  December 2008, the top ten travel web sites in
  the US accounted for 17.4 of total purchases
  (Hitwise, 2008), despite the availability of over
  46,800 travel sites.
• Early research in Internet usage also supports
  this consolidation. Cokcburn and MacKenzie (2000)
  discovered that users doubled the number of
  website visits overall between 1995 and 2000, but
  the number of revisits to a site increased from
  61 to over 80, meaning that 80 of the sites
  visited by a user had previously been seen. In
  short, users were consolidating their usage to
  fewer web pages despite the 8,000 times increase
  in available web pages during the time period.

61
Usage Consolidation

• There is evidence that the compression is
  on-going in the travel industry. JD Powers and
  Associates (2008), a research firm, reported an
  increase of 15 in on-line travel bookings in
  2008, However, they only found a 1 increase in
  bookings by smaller firms (increasing the market
  share of big firms from 84 to 86), supporting
  the Optimal Diet Theory.
• In short, there is a negative relationship
  between the number of available sites for a given
  item and the number of sites accessed.
• As the number of sites available for a given item
  increase, the number of sites accessed by an
  individual decrease.

62
Agenda
Introduction Approach Literature review Optimal
Foraging Theory Human Cognitive Factors The
Economic Perspective Developing a Behavioral
Foraging Model Theoretical Model and
Hypotheses Theoretical Components Methods The
Experiment Instrument Sample Population Question
s and Answers
63
Overall Approach

• The proposed research consists of
• 1. Building a theoretical model
• 2. Test the proposed relationships
• 3. Build a simulation model
• 4. Simulate changes in characteristics and
  explore asymptotes and economic aspects of the
  foraging model

64
The Experiment
65
The Experiment Text
66
The Experiment Text
67
The Experiment Text
68
Agenda
Introduction Approach Literature review Optimal
Foraging Theory Human Cognitive Factors The
Economic Perspective Developing a Behavioral
Foraging Model Theoretical Model and
Hypotheses Theoretical Components Methods The
Experiment Instrument Sample Population Question
s and Answers
69
The Instrument

• The Survey Instrument is using existing questions
  from validated research papers in the IS field

70
Agenda
Introduction Approach Literature review Optimal
Foraging Theory Human Cognitive Factors The
Economic Perspective Developing a Behavioral
Foraging Model Theoretical Model and
Hypotheses Theoretical Components Methods The
Experiment Instrument Sample Population Question
s and Answers
71
Target Sample Size - Power Analysis
72
Overall Approach

• The proposed research consists of
• 1. Building a theoretical model
• 2. Test the proposed relationships
• 3. Build a simulation model
• 4. Simulate changes in characteristics and
  explore asymptotes and economic aspects of the
  foraging model

73
Agenda
Introduction Approach Literature review Optimal
Foraging Theory Human Cognitive Factors The
Economic Perspective Developing a Behavioral
Foraging Model Theoretical Model and
Hypotheses Theoretical Components Methods The
Experiment Instrument Sample Population Question
s and Answers
74
The Instrument
75
The Instrument
76
Cognitive Absorption - Temporal Dissociation
77
Cognitive Absorption - Focused Immersion
78
Cognitive Absorption - Heighted enjoyment
Control
79
Cognitive Absorption - Curiosity
80
Personal Innovativeness (PITT)
81
Cognitive Playfulness (CPS)
82
Computer Self-Efficacy (CSE)
83
Extra Questions - Future use
84
The Experiment Example