114

Chapter 4
How People Approach
Information

ABSTRACT
Information interaction and interpretation will vary dependent on people’s goals and information needs.
These differing goals and information needs shape what information is actually deemed important, and
how hard people will work to understand it. As a result, information importance can also differ radi-
cally between people. In any particular situation, even after the information has been found and is being
considered, factors inherent in human nature come into play to influence the interpretation. This chapter
considers some of those factors influencing the ways people approach and interpret information. The
white area in Figure 1 shows the area of the HII model relevant to this chapter.

BACKGROUND a very active to a very passive approach to

interpreting it. Some people question all the
My definition of usability is identical to my information and some people prefer to avoid
definition of plain language, my definition of the information altogether.
reader-focused writing, my definition of document Effort: How much effort people will put into un-
design… We’re here to make a the product work derstanding the information and what factors
for people —Janice Redish influence their decision to exert that effort.
Sufficing: People stop looking for and interpret-
ing information when they are happy with
The main areas covered in this chapter are: the answer because it fits their information
needs or understanding of the situation. The
Reacting to Information: How people actually answer is not always sufficient, but people
react to information. This can range from have a hard time judging sufficiency.
Multitasking: Multiple tasks pull people in differ-
DOI: 10.4018/978-1-4666-0152-9.ch004
ent directions and they can rarely give full

Copyright © 2012, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
How People Approach Information

Figure 1. HII model—approaching information

attention to interacting with information. INTRODUCTION

Dealing with multiple simultaneous tasks
creates its own problems. A common design idea is to create a web site
Salience: Not all information carries the same level (especially intranet sites) that contains a large col-
of importance. The most salient information lection of cases which people can use to share their
needs to be identified and processed. experiences and prevent similar problems in the
Information Quality: Information must meet a future. However, while most people see this idea as
certain level of quality before people will a potential benefit, research calls into question how
trust it and people must trust it before they much it is actually used. The provided information
will use it. People judge quality based on often gets viewed as a big pile of text people must
their perception of its accuracy, complete- search through and then convert/reshape for an
ness, authority, usefulness, and accessibility. answer to their situation. In general, people usu-
Age: Age and its associated changes in cognitive ally feel competent to go about their work without
ability change how people interact with needing of additional information; they only resort
information. to information searches when they don’t know an
Novice to Expert: Level of experience and back- answer (Kwan & Balasubramanian, 2003). Note
ground knowledge change how people the use of the word “feel” at the beginning of the
interact with information. previous sentence. Unfortunately, people “feel”
they know more than they do; they tend to overrate
their ability since they don’t realize the extent of

115
 How People Approach Information

Box 1.
what they don’t know about a topic. These same
feelings of overrating ability and competence Defining content with fixed rules
relate to the issues of people sufficing and mini- Effective design needs more than a fixed set of rules or simple
knowledge what has worked before. Too often, though, good
mizing effort when performing a task, which will ideas in one context are copied in another context in ways that
be discussed in this chapter. People stop looking prove inapplicable because the person only knows the rule, but
not the logic behind it. Without understanding both the rule
for information when they feel the information
and the logic leads to poor design, or worse: it leads to simply
they have adequately addresses a question, with copying existing (potentially poor or unusable) design ideas
factors such as prior knowledge and desired level because “that’s the way it’s done.”

of knowledge influencing when they stop look- For example, consider the design differences between two
ing (Albers, 2004a). As long as they know some living rooms.

way of accomplishing a task, they see no reason • The furniture is arranged differently.
to expend effort finding a better way, assuming • The room shapes are different.
• The paints and fabric are different.
they even know that a better way might exist. • One has two love seats and no sofa, the other has just
Factors such as these affect how people approach a sofa.
• Only one contains a television.
information, how they interpret it, and how they
connect it to their situation. Yet, for all of these differences, people have no trouble
recognizing each as a living room. And even if the description
were totally verbal, most people would be able recognize the
room as a living room. The basic characteristics of living room
versus dining room or bedroom are apparent.
REACTING TO INFORMATION
If, on the other hand, we defined living room as having green
sofas, then most houses suddenly cease to have a living room.
Different groups of people, or the same person Likewise, following interior design rules for making paint con-
at different times, will approach information dif- trast wonderfully with the upholstery, rug, and room accents
is an exercise in futility unless the fundamental room layout
ferently. What they want and how they expect to fits the owner’s needs and the floor joists properly support the
acquire information differs. Understanding what weight. The structural needs for an antique grand piano are dif-
information is desired and audience expectations ferent from a sofa, as are the traffic flow paths around them.

are essential to meeting their information needs

and communication goals. Therefore, design
teams need to ensure that the information source even acknowledge that they are good ideas, but
conforms to both the situational context and the that they just don’t see how to apply them to their
way people approach the information. situation. This can have major economic impacts,
A design team often makes the assumption such as a failure to transfer best practices between
that the information can be captured in one situ- work sites or rejecting a recommendation that
ation and used across many others. Although it would improve a process. Bechky (2003) points
greatly simplifies content creation, the mindset out that communication problems, such as those
that “information is information,” and that what which occur between production workers, engi-
applies to one situation applies to all, seriously neers, and management, arise from how they view
impedes HII. As may be expected, that assump- the information differently. She also describes
tion has proven problematic. Szulanski (1996) how factors as simple as where the interaction
and Tyre and von Hippel (1997) have looked at occurs (within an office/conference room or on
transferring information internally within a corpo- the production floor) have a strong impact on how
ration and found causal ambiguity to be a major the information is approached and interpreted.
impediment. Causal ambiguity puts information Information is useful only when it fits within
in terms or concepts which people cannot easily the contextual constraints of the current situation
place within their current situation. People may and the reader can understand how it fits. Many

116
How People Approach Information

Box 2.
post-incident reports find that people had the
required information but failed to act on it. They Approaches to health information
either because they didn’t realize its importance, The Internet has become the major source of health informa-
tion for non-medical people. In 2004, 56% of Internet users
they didn’t perceive its relevance to the situation, reported accessing the Internet for healthcare information, a
or else too much effort was required to analyze number which is by now undoubtedly very low.
Access to information about disease ailments, their treatments,
and reshape the information. All of which di-
and their consequences for patients is clearly important, but
rectly result from poor HII and not “human error.” how people actually approach that information varies.

• The information a patient privileges is different from

In explaining (or excusing) their failures to foresee that of a medical provider. Patients want information
crises in their businesses, executives and direc- about how to care for themselves, not disease-specific
medical details relevant to medical providers.
tors of large corporations have emphasized the • Patients’ ways of accessing information change over
difficulty of “knowing everything.” With the pace time. Hospital-linked, professional sources of informa-
tion are generally used around the time of diagnosis.
of corporate life today, even the best intentioned But within 6 months, many patients had begun using
must rely on knowledge that has been extracted, non-professional sources such as television, magazines,
and newspapers (Booth et al., 2005).
summarized, and reorganized. No business leader • Information on the Internet supplements a physician’s
would claim “PowerPoint made me do it!” But the advice. Essays from the early days of the Internet
expressed fear that people would ignore medical advice
reduction— and occasional oversimplification— and try to define their own treatments via online means.
of a complex matter to a few bullet points on a This fear has not materialized. Most people use the
Internet for finding general information and getting a
PowerPoint slide epitomizes the way technology second opinion, but not as a means of self-diagnosis.
has changed how we know things (Gorry, 2005). Online health information complements, but does not
replace, interaction with a physician (Zeng et al., 2004).

A significant factor that distinguishes good All of these different approaches are valid and all are used by
from poor information is how well the presentation non-medical people looking for healthcare information. From
the very beginning of considering the HII, design teams need
takes into account contextual constraints and fits to consider how different sections of each audience group will
the information into a given situation. Consider approach the information, how that approach affects the HII,
and how to ensure that the design supports it.
how a report may explain why a change would be
good for one department but ignores the affects of
the recommended change on other departments.
& Rogers, 1999). The effectiveness of different
It has overly focused on one aspect of the situ-
approaches varies, of course, and a design team
ation and not the entire situation including how
should strive for designs which lead the reader
the department fits within the corporate structure.
along the most effective paths. Of course, this
Designing information which conforms to
complicates design teams’ jobs since they cannot
these constraints is difficult. The situational con-
assume that any specific strategies will be used.
text of a person is a dynamic, continually evolv-
They need to understand what information is
ing and open-ended process which contains all
needed and how it will be used, but this does not
the cultural, social, and historical aspects of the
mean restricting people to a predefined, “correct”
situation. Any new information that people obtain
way (Albers, 2004a).
must be transformed from its current form into
Usability testing is the current best method of
one relevant to all of those aspects. Fortunately,
determining if the HII is achieving project goals.
people are adaptive in transforming their infor-
Such testing can best determine whether people
mation (Payne, Bettman, & Johnson,1993). The
approach information as the design team expected
strategies they use are taken from a mixed bag,
(Redish, 2007). However, Russell (2005) consid-
with frequent strategy shifts; people do not adhere
ers how difficult it is to determine whether a design
to just one strategy (Navarro-Prieto, Scaife, M.

117
 How People Approach Information

fits how people want to approach the information. Active View: In the active view, people intend to
He claims that testing must go beyond the typical actively interact with and manipulate infor-
observations of usability testing and include deeper mation and expect to exert some level of
techniques, such as eye-tracking, to capture the effort to fully extract information from it. A
subtleties of people’s interaction with the content. business analyst sitting down with the
Some of the major problems requiring thought, monthly reports knows that the main points
apart from poor design or content, are that: that senior management needs are not listed
verbatim in those reports. The information
• People can approach information with must be manipulated to reveal the answers.
either an active or passive intent, which Passive View: In the passive view, people expect
changes how it needs to be presented. the information to be provided in a fully
• People tend to avoid information. Dealing digested manner that requires no effort for
with the content takes a lot of effort and understanding. An example is a couch po-
people try to minimize that effort. tato sitting down in front of a television and
• People need to transform the text into thinking “I’m here; entertain me.” From a
a form that fits their current situation. documentation or web-content viewpoint,
Information may be of high quality by the passive approach could be the static
most standards, but if it does not lend itself presentation of a news site or any text which
to easy transformation, the content will not provides a fixed set of content with no provi-
be successfully communicated. sion for finding more or filling in missing
details. A passive approach could also simply
provide a fixed sequence of steps to accom-
Active versus Passive Approaches plish a task with no explanation. Although
to Content people can use this to accomplish the task,
next time they will be forced to follow the
People can approach information with either same steps again.
an active or a passive view. The active-passive
spectrum (Figure 2) forms a continuum with most Older and mostly discredited design ideas
people falling somewhere in the middle, and with viewed people as passive input/output devices
their exact placement depending on the situation, that were empty containers into which informa-
their individual personality, and the external fac- tion could be poured or from which it could be
tors they bring to the situation. retrieved. Many data entry programs still take
this view, that people are themselves nothing
Figure 2. Active-passive spectrum with examples more than data entry devices. As long as the only
of the various points (adapted from Retzinger, function of the program is data entry, the concept
2009, p. 249) of people as data entry devices is marginally
workable. People’s control over an interaction
is a key factor in active approaches (Retzinger,
2009). However, control of the interaction needs
to be real, not just a false sense of control gained
via superficial interactivity.

A user can be presented with options that allow

them to be in control of a text, yet those options

118
How People Approach Information

can be seen as features that designers create and than learning new methods so they can efficiently
force users to use. Thus, a false sense of control interact with information, people are content with
can be seen as being created. This false sense their current approaches. In reflecting on this
of control can be seen specifically when a user paradox, Krug (2000) advocates against passive
uses a website where a user has control over how approaches to design and for active approaches
she navigates through a website, but she does which do not hinder the reader. He points out that
not necessarily control the content of a website the idealized rational human that a design team
(Retzinger, 2009, p. 247). may think does not reflect the reality of a real
person using the interface. Design teams must
When the HII requires interaction with or ma- design for the real person, not an idealized one.
nipulation of the information, problems can arise. Carroll and Rosson (1987) have observed that
Issues of information salience and minimizing the paradox of the active user is composed of a
cognitive effort, which are discussed later in this production bias and an assimilation bias.
chapter, can impact any conclusions drawn from
the information. Production Bias: The observation that people
Data dumping of corporate reports and other are often not discovering and using func-
information, with the attitude of “here’s the stuff, tions that could have been more efficient.
you sort it out,” is an example of assuming active People have a goal to complete a task and
recipients, but failing to meet their needs. Data they approach the task focused on achieving
dumping views information as a commodity to that goal. Learning new functionality which
be dumped on people. But effective HII requires may be useful later is not a current goal.
that information be shaped and focused for each People want to create a table of contents in
audience group to maximize comprehension. a document. They may realize that learning
Mirel, Feinberg, and Allmendinger (1991) found and applying styles will make generating the
that when active users are somewhat uncertain table of contents easier, but refuse to invest
about their task goals, they seek instruction, but the time since they want this documents
not in the form of rote, step-by-step procedures. table of contents now.
Rather, they desire elaborated discussions on the Assimilation Bias: The observation that people
implications of various commands and the condi- often figure out how to use what knowledge
tions and cases that are associated with specific they already possess to achieve new goals.
courses of action. A consistent disconnect is seen People use their prior knowledge as a basis
between design teams viewing the content itself for interacting with information. Assimila-
as the important “thing” while people interact- tion bias is using what worked in the past
ing with that content viewing their goal as the to accomplish a new task. A person may
important “thing.” People see information only center a heading with spaces rather than a
as a means of achieving their goal and judge it on center function because in the past that is
how well it helps they achieve that goal. Software what they used.
training people consistently encounter users who However, assimilation bias introduces a
do not wish to learn about software, but only to significant chance of negative transfer (see
do something meaningful with it. They don’t care chapter 3: What People Bring with Them,
about the software itself; they care about ways to “Transfer effects”), which can frustrate
accomplish their job. people. For example, learning to use a layout
Carroll and Rosson (1987) first described what program, such as Adobe InDesign, which
they called “the paradox of the active user.” Rather uses a paste-up board metaphor, conflicts

119
 How People Approach Information

Box 3.
them, but simply relied on system feedback while
Approaching healthcare information performing the action (Payne, 1991). Interpret-
A significant amount of the work on how people (patients) ing this approach as more than just minimizing
approach information has occurred within the realm of health
care. Many of the brochures that patients receive assume a pas-
mental effort, Fu and Gray also found that people
sive recipient receiving information as a commodity (Payne, justified this approach as giving better control
2002). The writing style and information content level seems
over performance and errors. It allowed them to
to assume that the patient only wants a minimal amount of ma-
terial, an assumption which is not supported by the research. accomplish a task and ensure they were progress-
For example, “pregnant women either represented themselves ing in the desired direction.
as active information seekers or provided compelling reasons
why they could or should not actively seek information”
(McKenzie, 2002, p. 43). Strong Tendency to Avoid Reading
Part of the design problem is the different mindsets of patient Information
and healthcare provider. Patients want information about the
disease. The healthcare provider wants to treat the disease and
privileges information relevant to making treatment decisions Design teams try to create a highly usable design
(as well as already understanding the disease prognosis). containing all of the information people need.
People want healthcare information. They want to understand
what is wrong with them and how it will affect their life; how- However, many people don’t really want to access
ever, they do not necessarily want to make treatment decisions a web page or use a manual. Ignoring information
(Jenkins, Fallowfield, & Saul, 2001). The two types of infor-
mation needs—the patient’s and the provider’s—are distinct,
has a long history. The syndrome is often summed
but people writing the information often take the healthcare up comically as follows: “When all else fails,
provider’s point of view.
read the manual.” Penrose and Seiford (1988)
found less than 30% of the people in their survey
looked at the manual before using the software.
with prior knowledge from word processors Those people viewed manuals as something to
based on a typewriter metaphor. The negative ignore until they had a problem. Only then would
transfer makes it difficult to learn to add text they search for information. The movement of
and, since it is not part of a typewriter, and information to the Internet has not changed this
people may never learn that objects can be finding; it has only shifted the search venue from
moved off the page and onto the paste-up manuals to web pages. Nielsen (2001a) sums
board for later reuse on different pages. up this phenomenon by stating, “Nielsen’s First
Law of Computer Documentation is that people
Fu and Gray (2004) looked deeper at the don’t read it.”
concept of how these biases affect the learning Simply providing information to people has
of new procedures and found that inefficient never worked. It must match their mental models
procedures continue to be used even when people and situational needs (Mirel 1988). However, the
seem to know about more efficient procedures. way people avoid reading information raises pro-
They explained their findings in terms of people found implications design teams must consider.
offloading cognitive effort. People avoid using a
procedure which forces them to keep information • Design teams cannot assume that people
in their head and prefer procedures which offload have read any previous information, neces-
that memory effort to the display. A series of inef- sitating highly modular design.
ficient steps with good feedback is preferred over • Design teams cannot assume that all of the
a shorter/faster series that requires mental effort. A text in the section will be read. For exam-
study that looked at how well people understood ple, people routinely ignore written proce-
software functions they used constantly found dures and only follow examples (Lefevre
many people could not describe how they used & Dixon, 1986).

120
How People Approach Information

• Design teams cannot assume that any text Depending on how closely the original informa-
not deemed essential to the current task will tion matches the reader’s needs, the transformation
be read—what the reader deems essential, can run from minimal changes to extensive reshap-
not the design team, which can often differ. ing. A potential issue with how people transform
Unfortunately, poor design can lead people information is that they will change the problem
to misjudge how essential a block of text is to fit information presentation (Johnson, Payne,
for achieving a goal. & Bettman, 1988). Thus, the wrong presentation
• Design teams cannot assume that people can cause difficulty in relating the information
read before they get into trouble. When a to the situation. However, a later section of this
problem occurs, they want only enough in- chapter considers the amount of effort people put
formation to get out of trouble and back on into tasks, and discusses that if the information
track. requires extensive transformation, it will probably
be ignored as mentally too demanding.
Perversely, just when information can be most Software-based documentation requires sig-
useful to people, they avoid it because of various, nificant transformation. The content explains
often personally unpleasant factors. Johnson, An- how to perform a task with the software, but does
drews, and Allard (2001) discuss an information- not relate it to the situation. When the real-world
seeking paradox in which people are less likely to goal requires using multiple software functions,
look for cancer information as their proximity to people are forced to develop their own model of
cancer increases. This appears irrational because how to relate the software tasks to achieving the
the proximity to cancer patients increases the in- goal (Mirel, 1992; Redish and Schell, 1989). For
formation need and that information could result example, inserting a pull quote into a Word docu-
in reduced morbidity and mortality. Although ment requires using a textbox. Although Word’s
design teams know that people need information, documentation explains how to insert a textbox,
they must acknowledge that for various reasons, it never provides any clues about why a textbox
people will desire to avoid the content. The design would be used. If people did not know to start
must draw them into the content. with a textbox, they would be unable to transform
the content into part of their “insert a pull quote”
Transforming Information goal statement. One of the distinguishing differ-
ences between vendor manuals and third-party
Information as written is rarely directly usable books found in bookstores is that the content of
for people; instead, it needs to be transformed a third-party book requires less transformation.
and synthesized to be directly applicable to the The writer has better fit the information to the
situation. People are constantly faced with huge reader’s needs.
quantities of information of varying relevance The difficulty of transforming content can be
and quality. However, they may find it difficult found in Dye’s (1988) research, which found that
to mentally transform a collection of data be- over 70% of user comments about a product’s
cause this transformation carries a high cognitive documentation arose from errors in interpreting
workload. As a result, poor design carries the risk the documentation. Further analysis found that the
that people will shed the workload and not do the documentation did correctly explain the software
transformation (Herbig & Kramer, 1992; Webster but failed to relate it to real-world tasks. System
& Kruglanski, 1994). In simple terms, they ignore content might well contain the answers, but if
the information. poor HII intrudes—if people cannot transform
that information into a form relevant to the current

121
 How People Approach Information

Box 4.
goal—the information will not be successfully
communicated. Transformation of numbers versus graphs
Another example: people find it easier to One of the basic tenets of numerical design is to use tables for
exact numbers and graphs for trends. Using the wrong presen-
remember phone numbers formulated as words tation means that people will have a hard time transforming
(555-CALL), but they also find them harder to the information which is readily apparent in one into the other.
Given a set of numbers in a table, people find it hard to see
dial than a full set of numbers. The problem is that
trends, other than trivial trends such as “increasing.” It is also
people do not dial letters, but numbers; they have very difficult to discern trends within non-linear functions, or
to transform CALL by searching the touchpad for in data with high noise levels. Likewise, most graphs do not
allow easy reading of exact values of various data points.
the corresponding digits. In both of these cases, the proper presentation of the graphic
Factors affecting information transformation: minimizes the required transformation.

Credibility: How much credibility the informa-

tion is perceived to have determines how assembled out of these odds and ends of
much effort people are willing to put into partially relevant and partially extraneous
the transformation. Content deemed only generalization. And these ‘theories’ are used
marginally credible will not be transformed. for further prediction” (Carroll & Rosson,
Credibility is based on perceived value, not 1987, p. 81). Klein (1999) found that people
actual value; with poor HII, the two can be evaluate information based on the order
substantially different. in which they receive it. This accentuates
Task Relevance: The content must be relevant to the importance of information salience,
the current task. People need to be able to and that information be presented in an
distinguish relevant from irrelevant infor- order relevant to situational understanding.
mation and to group information elements Random data dumps of information fail to
together. Information viewed as secondary communicate since they require too much
will not be transformed. Research on in- effort to transform.
structions finds that people skip blocks of Readability: Readability considerations such as
explanatory text (failing to transform it for font choice and page layout make a strong
their situation) since it does not help them initial impression. Content will be ignored
complete the task. Relevant text must be if it is presented in an uninviting design or
cued so that readers know to transform it. unreadable fonts, even if the reader admits
Availability: People tend to work only with infor- that the information is important.
mation that is easily available. Rather than
hunting down higher quality information,
they will use what is at hand. Information Which is Not
Systematic Arrangement: Information must be ar- Understood is Ignored
ranged to support connecting it to the current
situation. People always impose structure on People tend to ignore information they do not
information, and they will interpret it and understand, or if they do not understand how it
draw some type of conclusion. If the content applies to the current goal. Content written at the
fails to fit the context, that interpretation wrong level, especially at a more expert level than
and conclusion can be wrong. “If something the reader can handle, is typically skipped. People
can be interpreted (no matter how specious reading healthcare information will not read sec-
the basis for this interpretation), then it will tions containing medical jargon. Since they don’t
be interpreted. Ad hoc theories are hastily understand the terms, the text makes no sense to

122
How People Approach Information

them. Likewise, most people don’t understand information will be ignored and irrelevant infor-
mathematics or statistics and will skip sections mation brought to the fore.
containing equations or statistical information. Relevant information gets ignored because
Interestingly, text written at an inconsistent people suffice. Rather than devising ways to ef-
level, whether intentionally or unintentionally, ficiently handle the cognitive load, people often
bears the risk that readers will skip the entire simply reduce cognitive load by dumping parts of
text. After ignoring (skipping) some paragraphs, the problem and reverting to previously learned
readers who have a hard time getting reoriented conventions. Rather than working to understand
will just stop reading. Text which intentionally information, people reduce their cognitive effort
contains content written at various levels must and ignore it because it isn’t understood (Bettman,
clearly cue the level of each text block. Johnson, and Payne, 1990; Wickens, 1992).
Interestingly, as the information flow increases
Information Which is Not in complexity, contrary to intuition, cognitive
Relevant is Ignored overload prevents people from increasing the
complexity of their mental integration. Instead,
Based on experience, people use their activated they start to ignore increasingly larger amounts
mental model to begin the search for information. of information. Thus, while this seems to support
The mental model provides the basic skeleton, that highly compact information presentation,
which they fill in with information specific to Rubens and Rubens (1988) found that making
the situation. The mental model provides the the information too compact or concise hindered
filter which allows them to sift through all of the performance.
information available and only focus on the rel- Good mental representations of the overall
evant information. The mental model can lead to situation, and commanding a clear understanding
an expectancy bias (seeing what you expect to see of the current situation, are very important when
rather than what is there) that exerts a strong effect dealing with complex or open-ended questions,
on what information gets processed. Expectancy since they lack any clear stopping point for ac-
bias helps explain why, in many disaster or major quiring information (Redish, 2007; Rouet, 2003).
accident situations, people made seemingly bad The design and analysis phases of development
choices and ignored the correct data that was right require working closely with the real users and
in front of them (Klein, 1988; Woods, Patterson, understanding how they develop their understand-
& Roth, 2002). They were using a wrong mental ing of a situation.
model and did not see the relevance of informa-
tion which turned out to be critical. Known Information versus
People mentally process only the information Need to Know Information
they view as relevant to the current situation.
Most situations contain an overabundance of data, Casner (1994) found that with most commercial jet
mostly irrelevant to current goals, that complicates pilots, a large proportion of the mental processes
the work of comprehending the content. People used to fly a known route were environmental.
need to sift through the available text, decide what With experience gained by repeated flying the
is relevant, and use it to achieve their goal. Un- route, the pilots learned what to look for and what
less the presentation effectively supports cueing to expect next. The commercial pilot knows what
of information to allow people to make proper information is needed and knows when and where
choices, there is a strong risk that some relevant to look for specific information. Before looking at
the specific information, the pilot will not be able

123
 How People Approach Information

to state its value, but will know it is needed, why Information Relationships and
it is needed now, and how it will fit into the cur- Information Interaction
rent situation model. Likewise, a business analyst
knows what to look for in the monthly financial Connecting different goals and their associated
or production reports. Good HII makes it easier information needs is key to understanding a situ-
to find and analyze that information. ation and knowing what information is salient.
Some uses of online information systems are Effective design focuses on providing information
no different. People with experience know what that allows people to achieve their goals. Build-
information is needed as well as how to find it ing an understanding of the information requires
(such as by reading monthly reports, or booking forming the proper relationships between the
an airline flight). They simply need to access the salient information elements.
proper information. People new to the situation Understanding information relationships
lack this ability because they lack experience, provides a solid foundation for understanding the
but through interaction with a particular site (for cause of a problem and how to solve it (Casaday,
instance, a corporate intranet) they build up a 1991). Understanding these relationships places
base of experiences which help them to remain the entire situation in context (Endsley, 1995).
oriented and to find information quickly. The Helping people achieve goals and comprehend
faster and easier a design supports acquiring this information requires providing contextually
experience, the more efficient the HII becomes relevant information that fits their goals and in-
across multiple interactions. Experience helps formation needs, allowing for efficient interac-
them build a good mental representation of the tion with that information, and making clear the
overall site and, consequently, lets them easily find interrelationships within that information. Designs
the information which fits into their mental model must provide integrated information that demon-
of the current problem situation. The specifics of strates these relationships, not just a collection of
the desired information and their prior knowledge information (Albers, 2009, 2010). The design goal
affect the coherence of the mental representation, behind showing the information relationships “is
which in turn, affects the quality of the final result to help people solve problems, rather than directly
(Rouet, 2003). to solve problems posed to them (e.g., question-
Finding information takes effort. Both the answer systems)” (Belkin, 1980, p. 134). People
physical acts of interacting with a computer or must be made aware that the information exists
paper manual, and mentally processing the text and that they have easy access to it; but in addi-
to integrate it into the current situation model, tion, the overall information structure must permit
require effort. In general, people try to minimize easy construction of information relationships.
the effort they put into searching for information. Presenting too much information leads to cogni-
So if they already possess some knowledge, they tive overload, and trying to guide people along
will be willing to use imperfect knowledge they one path only frustrates them because it presents
already have, rather than interact with a system information they either already know or consider
to find better information (Gray & Fu, 2004). irrelevant, while failing to show the information
Going with previously acquired information is relationships that fit the current situation.
more error-prone, but less taxing on time and The fundamental problem of many existing
cognitive resources. designs is poor handling of information relation-
ships. The systems contain the information, but
lack clear connections between information ele-
ments. The interrelationships that require a high

124
How People Approach Information

Box 5.
salience that are often not clearly evident. Further
complicating the design, these interrelationships Information relationships and improving writing
can change as people’s situational awareness im- A major pharmaceuticals manufacturer wanted to improve
the ability of their chemists to write reports. A typical report
proves or the situation evolves. Yet, design teams is an incident report, which describes where something went
cannot be faulted for the lack of good design; we wrong, such as when part of the chemistry control went out
of specifications. The report must document what happened
still lack clear methods of how to perform that
and what corrective actions were taken and whether it affected
design. Today we still lack easy ways of tracking, the production batch quality and how such events would be
manipulating, and refinding multiple pieces of prevented in the future.

information on intranets and the Internet. A fun- The managers saw a need to “improve tech writing” and “have
damental problem to overcome is that people have chemists write clearly” since the reports were not accomplish-
ing what they desired. However, the writing consultants they
a hard time integrating information and relating hired saw the problem not as a lack of writing ability, but as a
various data points to each other. But even more lack of building relationships within the report. The chemists
were not having problems with sentences and grammar (the
importantly, they have a hard time remembering or “improve technical writing” part), but with thinking through
considering subtle cause-and-effect, goal-oriented what happened, connecting the basic plant data with unin-
tended results, and building those relationships within the text
relationships that exist between the information to present a well-formed argument on what occurred and what
being viewed and other relevant information. needed to change (Bernhardt, 2002). The chemists understood
the problem but did not understand how to communicate it to
For design teams to provide support for infor- management.
mation relationships requires understanding both
what the situation is and what it means to people:
how do they view the situation, what previous EFFORT EXPENDED TO
knowledge do they bring, and what do they want UNDERSTAND INFORMATION
to accomplish? Achieving the design goal of sup-
porting people in understanding the relationships Effort has been mentioned multiple times in earlier
between goals involves: sections of this chapter. This section takes a closer
look at what is meant by effort and the different
• Finding the logical relationships between types of effort people engage in when approach-
different goals and available information. ing information.
• Defining the information needs to make When people engage in a task or read informa-
those relationships both easily visible and tion, they engage in an effort trade-off between
clearly understood. obtaining the desired information and expending
• Defining how people will have to interact the least acceptable effort. Depending on the situ-
with the information to fully comprehend ation, the least acceptable effort varies based on
it. the perceived value of the information sought.
• Defining how the social and cognitive as- People call the help desk first because they think
pects of interactions inherent in a situation it requires less effort. A consideration of this ef-
affect both goal relationships. fort trade-off is a significant factor missing from
• Showing why or when the connections be- many design analysis reports.
tween goals are relevant or when they are A misguided approach in many information
irrelevant (with the latter being just as im- systems is a focus on ensuring that the maximum
portant as the former). amount of accurate content is available. Then,
when the system is released, design teams are
surprised to find that people are not extracting
the maximum amount of information possible.

125
 How People Approach Information

Box 6.
Instead, support requests how people asking for
content which is already in the system. People are Failure to use system features
failing to interact with the system as intended. And Most people only use a small subset of the features in a prod-
uct, which is interesting since they buy the product on the basis
often it is that design teams find people request- of the whole feature set. One explanation of the scant use of
ing information that is already available in the most features is the effort required to learn how to use them.
system. Because of poor design, they can’t find it. Consider learning to use a new camera. The camera itself is the
This closely relates to reluctance to use help files size of a deck of cards. Its accompanying documentation con-
(Scanlon, 1996; Spool 1997). Likewise, help desk tains three manuals: a “Direct Print User Guide” at 77 pages;
a “Camera User Guide” at 225 pages; and a “Software Starter
research reveals that a majority of the answers for Guide” at 80 pages.
help desk questions are within the user-available
The manuals explain all of the 23 shooting modes with an
online help and documentation. However, those additional 34 functions that could be used within the various
people either couldn’t find that information or, shooting modes. Although people will acknowledge that they
may be able to take better pictures if they learn how to use
more likely, didn’t bother looking since it seemed those modes, it is simpler and adequate for their needs to just
easier to call the help desk. People are reluctant figure out how to point and shoot while relying on autofocus
and autoflash. The extra effort required to learn the 23 modes
to look for information because they believe their and 34 functions is not worth their time.
search will not succeed. Such a view may stem
from a learned response because of past use of
poor documentation. The information was not Since people try to minimize the amount of
available in the past; why would it be there now? physical and mental effort they expend, an effec-
Part of the problem is a design assumption tive and efficient design must minimize both the
that providing a maximal amount of informa- physical and mental exertion required to interact
tion will result in people’s extracting all of that with the information. Many interfaces, espe-
information, or more typically, that they will have cially those which provide information in complex
no trouble extracting specific, relevant points. In situations, impose a high cognitive demand. Of
general, that doesn’t happen because people don’t course, at times the high cognitive demand was
want to expend the mental energy to transform the result of poor design decisions (which are
that information into content applicable to their often easier for the development team)—products
situation. Instead, people use strategies based on such as data dumps for reports. The goal for qual-
a trade-off between time and the amount of effort ity HII is to minimize both cognitive and physical
required to gain accurate and sufficient infor- interaction demands, which means people get a
mation (the next section, on sufficing, explains better design. Design teams need to consider three
how obtaining complete information is rarely different types of effort:
people’s goal). People attempt to minimize both
cognitive resources and time wasted by estimat- Perceived Effort: People make decisions to mini-
ing the amount of effort required for a task and mize effort based on their estimates of the
making strategic choices toward accomplishing effort (what they perceive the effort will be)
the task based on that evaluation (Fennema & rather than the actual effort.
Kleinmuntz, 1995). At the same time, they tend to Physical Effort: Clicking and typing on an inter-
make choices in terms of immediate, rather than face and manipulating information requires
long-term, efficiency and effort (Payne, Howes, physical effort.
& Reader, 2001). In turn, choices which lead to Cognitive Effort: The mental effort required to
minimizing effort lead to sufficing and accepting synthesize and transform the content into
inadequate information. Either way, people do not information applicable to the situation.
extract all of the information that they should or
that they need.

126
How People Approach Information

Box 7.
effort level exceeds the potential return, people
Online prescription writing will not read the page.
Physicians work under high time pressure and want to accom- Unfortunately, the correlation between antici-
plish their tasks as fast as possible. Thus, the slower time for
using an online clinical order entry system (online prescrip-
pated and actual effort is low; people find it very
tion writing) gets translated to requiring extra effort, causing hard to judge accurately how much effort a task
initial resistance to using the system. Plus, the major benefits
will require (Fennema & Kleinmuntz, 1995). This
are to the hospital system as a whole and not to the individual
physician. leads them to pick strategies which seem low-effort
but which in fact might be highly sub-optimal,
Studies of online clinical order entry have found a few consis-
tent aspects associated with the effort expended by physicians. ultimately requiring more effort than was really
The use of the system is typically mandated by hospital ad- necessary had a wise strategy been followed from
ministration, leaving the physician little choice about using the
system. In general, online prescriptions take longer than writ- the start. For people to make a reasonable effort
ing by hand, with only a modest offset in time savings from estimate requires that information have a coher-
looking in charts and the decrease in pharmacy inquiries from
unintelligible handwriting (Bell et al., 2004). There tends to be ent structure which they are able to distinguish
a strong emotional reaction to using the system—probably the (Thuring, Hannemann, & Haake, 1995). People
only response available since hospital administration mandates
that physicians use it. On the other hand, speed does improve must be able to grasp the relationships between
with experience. In the longer term, after becoming reliant on the information elements (Albers, 2010). Simply
the system, any downtime causes chaos and greatly slowed-
down responses. After becoming proficient with the system,
having a coherent structure from the design team’s
the physician has trouble writing prescriptions any other way perspective is not enough if people can’t relate it
(Ash et al., 2007). The physical and cognitive effort required
to hand-write prescriptions is revealed to be greater than using
to their situation.
the system, but only after the system is well-learned. People’s perceptions of effort depend on task
complexity, or on how difficult they assume a task
will be to accomplish—a judgment that varies
Perceived Effort depending on prior knowledge. Kieras and Polson
(1985) showed that perceived task complexity cor-
When starting a task, people select a strategy and relates with people’s knowledge of the task, not
make a mental estimate of the amount of effort it the actual complexity. This means that a task about
will require. Estimates which fail to closely fit real- which people know very little will be perceived
ity can misdirect people’s efforts. If the estimate as harder (requiring more effort) than one about
calls for too much effort, they may forgo the task which they have an understanding. People who
and make do with other information which they are unfamiliar with a task will probably grossly
have estimated to result in lower effort. However, over-estimate the actual effort. To complicate the
the effort used for strategy selection is not the situation, as task complexity increases, accurate
actual effort required, but their perception of the judgment of expected effort decreases, regardless
anticipated effort (in other words, how hard does
the task seem to be if a particular strategy is used).
Page design guidelines say to use lots of white Box 8.
space and graphics—often summed up as making Fine print or all caps print
the page look inviting—with the intent of making The design technique of putting blocks of text in either fine
the page appear to require less effort to read. A print or all capital letters strives to take advantage of perceived
effort. Reading text written in either one is more difficult
page that does not look inviting will cause people than reading normal-sized, mixed-case text. As a result, the
to form an opinion that the text requires a lot of writer can count on people either to skip or to skim that text.
This amounts to hiding information in plain sight by using
effort to read and understand. If the perceived formatting that discourages people from reading it, which is an
ethical issue. The basic logic informing perceived effort needs
to be considered in any design.

127
 How People Approach Information

Box 9.
of people’s experience. The findings of Fennema
and Kleinmuntz (1995) of low correlation between Perceived effort of Word styles
anticipated and actual effort reinforce the notion I know Word very well and can handle making changes to
styles at essentially an automatic level. On the other hand,
that estimates of the difficulty of the HII necessary even after teaching students how to use styles and requiring it
for a task are often nearly independent of actual in assignments, some will ignore them as too confusing or not
worth the effort.
task difficulty.
Improving the correlation of perceived effort Defining the styles, which does require a rather unwieldy
to actual effort requires that the design address sequence of clicks and dialog boxes, can easily be perceived as
more effort than the potential gain. The productivity loss does
those factors which can increase the accuracy of not occur until later in the project when changes, which could
the estimate. Some factors influencing the ac- be accomplished quickly with some style updates, require go-
ing through the entire document and making the same format
curacy of perceived effort are: alterations repeatedly.

Of course, students are rarely faced with the need for the large-
• The accuracy of a mental estimate of per- scale changes which make styles worth using, which greatly
ceived effort depends on how well the task reduces styles perceived relevance and their perceived need
to learn the material. All of the paragraphs in this book have
fits into people’s overall mental model. styles applied since I can see their advantage.
• When the design does not effectively assist
in resolving the question of effort, people
make effort estimates and strategy deci- • Perceived effort is very much based on
sions based on simple heuristic rules or just short-term gains, rather than long-term im-
using what they already know (Fu & Gray, provements. In many situations, this short-
2004). Unfortunately, the use of simple term gain comes at the expense of learning,
heuristic rules often leads to non-optimal which would result in better longer-term
or incorrect solutions. performance (Payne, Howes, & Reader,
• High perceived effort estimates lead peo- 2001). This can be seen with people who
ple to suffice (Simon, 1979) and to be con- have the attitude of “I just need to get this
tent with lesser amounts or poorer-quality job done; I don’t have time to learn a dif-
information. ferent way, even if it might be faster.”
• Different display designs can also affect
the perceived effort (and the actual effort) Physical Effort
for finding and using information. The dis-
play format can actually change the way Physical effort is the amount of physical work
people view the information. Based on the required to accomplish a task. This could range
amount of effort they judge it will take to from walking down the hall to visiting a library
understand the information, they change to drilling down through a list of links because
how they view the information. Slovic the required information is spread across multiple
(1972) found that people tend to change pages. In informational situations where people
their assessment strategy to fit the presen- know what they want and know how to get it, the
tation method. physical effort can be the major factor in judging
• Rather than mentally scheduling tasks in overall effort. For instance, people know a web
proportion to their significance and cog- page contains the information they want, but they
nitive demands, people tend to use equal- have to decide whether they need the information
scheduling, which assigns each task the enough to open a web browser and go to that URL.
same amount of mental effort (Langholtz, Or, if they have the web page open, they have to
Gettys, & Foote, 1995).

128
How People Approach Information

decide whether it is worth making several mouse for imperfect knowledge in-the-head.” (Gray &
clicks and waiting during slow page downloads. Fu, 2004, p. 163). Studies which manipulated
Or, to make it more extreme, they might have system response time by varying the delay time
to decide whether the information is worth their after clicking on a link show that as the response
turning on a computer. time increases, people take more time to analyze
In many situations, a tradeoff exists between the situation and make a choice (Gray, Schoelles,
minimizing physical and cognitive efforts. Any & Sims, 2005). When clicking a link brings an
choice that decreases one increases the other. almost immediate response, there is no time-price
How people handle this tradeoff depends on the to pay for making the click, and people are willing
situation, but the question is very sensitive to the to do more searching and to follow questionable
amount of physical effort. Wickens (1992) reports paths. But with a higher delay, only paths which
that small differences in the amount of physical are judged as high-quality are followed. The
effort affect people’s problem-solving strategy physical effort of obtaining the information was
and performance. In order to save physical effort, considered not worth it. Of course, this strategy
people would engage in inefficient strategies or depends on people’s ability to judge link quality,
would accept slower task times. which itself is suspect.
Duggan and Payne (2001) explored different Design teams need to include usability testing
reading strategies in the design of written proce- which considers the physical/cognitive tradeoff
dural instructions (for a simulated VCR). When and how to minimize it. A common situation is
people were forced to read several instruction when multiple windows are open on a computer
steps in a chunk before executing those steps, desktop, or a browser window has multiple tabs.
they learned the procedures better. The extra Systems that use multiple tabs, such as electronic
physical cost of accomplishing the task forced medical records (Gray & Fu, 2004), are especially
the people into finding a more efficient method. affected. From a design perspective, all of the
Furthermore, when people were allowed to read information people need is available. But, from
and accomplish the procedure as they chose, they the perspective of people needing that informa-
produced a spontaneous chunking strategy which tion, physical actions are required. Windows may
improved procedure retention (Payne, Howes, & have to be moved to allow for comparing differ-
Reader, 2001). In both cases, people worked to ent sets of data. Or, as with browser tabs, some
minimize physical effort at the expense of some values need to be memorized before they can be
extra cognitive effort. compared to values on a different tab. The effort
Gray and Fu (2004) found that performance people will exert to view this information depends
differences can be attributed to individual differ- on its value to them. If the information crosses a
ences in willingness to either memorize or access low-value threshold, people will not expend the
information. People seem willing to trade spending physical effort to view it.
some cognitive resources in exchange for physical Minimizing the physical effort typically re-
effort and waiting. When people had to access quires some up-front planning before engaging
information on a covered screen, even in a well- in a task. O’Hara and Payne (1998) argued that
known location, the effort and trade-off between planning actions continue until the estimated ben-
clicking on the window and using information efits are outweighed by the costs of the cognitive
they already knew depended on the speed of re- effort involved. They performed a study which had
sponse. “Milliseconds matter in that differences in people copying information between files. Typi-
effort measured in milliseconds suffice to induce cally, people want to complete each destination
users to ignore perfect knowledge in-the-world document before going onto the next; however,

129
 How People Approach Information

Box 10.
they are engaged. Each mental task consumes some
Effort of using external aids of those cognitive resources. Once the resources
Studies that look at the performance of student designers are exhausted, people suffer from cognitive over-
found they performed significantly better when their design
process included a knowledge-based system (Antony, Batra,
load, which leads to a high number of errors, or to
& Santhanam, 2005). However, there is an effort trade-off that skipped information. As a result, people strive to
depends on the ease of obtaining useful information from the
minimize the amount of cognitive effort (amount
knowledge-based system.
of cognitive resources) used for a task.
Deadlines were very important in forcing a team of engineer- The best design method to reduce cognitive
ing students engaged in a design project to think critically
about what information was most important and to focus the effort is to ensure that the system design maps
information they sought to support their design efforts. One directly onto people’s view or mental model of a
interesting result was that they tended to look for information
in topics on which they felt knowledgeable and avoided look- specific task (Benbasat, Dexter, & Todd, 1986).
ing up topics unfamiliar to them. For example, they looked up If this fails, people tend to use incremental ap-
information on performing the mathematical analyses required
for the design (such as stress and torque), but they avoided proaches that off-load the cognitive effort and
searching for information related to developing the project’s use general procedures that start with interactive
budget. This can be attributed to the difficulty of determining
how much effort would be required since, not understanding actions. When given a choice of different paths to
budgeting, they were faced with estimating efforts of unknown a solution, people tend to choose the one which
magnitude for an unknown benefit (Hayes & Akhavi, 2008).
is composed of a sequence of interactions. They
preferred working in an incremental fashion so
that they can receive incremental feedback to
in the study design, this was the most inefficient ensure they are making progress. One possible
way. When slow time response was added to the explanation is that incremental performance re-
operation (it took several seconds to copy each quires less cognitive effort as people can depend
piece of information), people were much faster on the displayed information rather than having
at finding more efficient ways of performing the to remember information while making greater
task. In other words, with fast responses, little leaps toward accomplishing the task (Fu & Gray,
effort went into planning; with slow responses, it 2004). Likewise, in a study of note-taking, Cary
was worth the time to plan a better method. Since and Carlson (2001) found that people tried to mini-
people don’t know how fast a system may respond mize cognitive effort and load in working memory
or how quickly information may be found, they by distributing the information between internal
will often continue to modify the effort they put (mental) and external (the display) resources.
into a task. How people work to minimize physi- In a study about people looking for information
cal effort can change during a task, even when on their computer (email or files), even when it
the task performed is routine, such as making was possible to jump straight to the desired infor-
photocopies of a book chapter (Gray, Schoelles, mation, people tended to use various navigation
& Sims, 2005). A slow copier or one which makes methods (Alvarado, Jeevan, Ackerman, & Karger,
it hard to position the book will cause people to 2003). This was explained as a result of minimiz-
skip pages, minimizing the overall physical effort ing cognitive effort since going directly to the
of making the copies. file required people to mentally process up-front
all of the metadata about the file. By navigating
Cognitive Effort to it, they only had to deal with that metadata in
small pieces and could offload much of the effort
Chapter 3: What People Bring with Them discussed to the display.
how people have very limited available cognitive Design teams should not make procedural
resources to allocate to all mental tasks in which information too easy. Some level of cognitive

130
How People Approach Information

effort is required to learn new tasks since the Krug (2000, p. 28) puts it, “Once we find something
material must be encoded in long-term memory. that works—no matter how badly—we tend not to
Step-by-step training methods, with each action look for a better way.” In an information search,
carefully described in the training manual, are people accept the first information which they
problematical since they don’t force the reader deem good enough for their needs. For example,
to do any evaluation or to exert any mental ef- in one study, when using a site map, subjects
fort before performing the step. Each step is fully focused on one or two hierarchies and quickly
presented and only needs to be followed exactly. developed a search strategy. Despite being sub-
People can easily minimize cognitive effort by optimal, these search strategies showed little or
reading the step and remembering it only long no improvement over time, possibly because they
enough to complete it. However, learning a were adequate to the task and required no further
procedure requires that people mentally encode elaboration (Nilsson & Mayer, 2002). People may
both the individual steps and their relationships admit a search is incomplete, or even somewhat
to the entire process and to store this information inaccurate, but as long as they consider it close
in long-term memory (Einstein et al., 2000). By enough, then it’s accepted as an adequate answer.
failing to build relationships, people can perform In fact, people are often highly resistant to
well in a course but be unable to apply the material being shown better or easier ways of performing
in a real-world situation. actions after they find a way that works. Many
times, along with essentially admitting they are
satisficing, people refuse to consider learning
SATISFICING a new way and respond along the lines of, “I
know this way. I don’t have time to learn a better
Another way of viewing how people minimize cog- way.” Rather than looking for an optimal path,
nitive effort is to consider it as only doing enough people satisfice and settle for less-than optimal
work to get a result which is “good enough,” an performance; instead of maximizing output, they
action which Simon (1979) defined as satisfice. minimize cognitive effort and attempt to produce
In information search and interaction, people stop satisfactory output with minimal exertion. For an
when they feel they have enough information to infrequent task, this attitude is acceptable, but
make a decision. With satisficing, rather than people will also exhibit it for tasks they perform
looking for the optimal choice, as in classical deci- several times daily.
sion models, people make choices that are good A potential HII problem with satisficing is
enough (Klein, 1999, 1993; Orasanu & Connolly, that people often do not know whether they have
1993). Most rational models of thought assume enough knowledge to evaluate an answer’s qual-
that people collect all relevant information and ity, but although this does affect the quality of the
then make a decision. However, for a non-trivial decision, it does not factor into their decision.
problem, there is no way of knowing in advance When searching for information in areas with
what all of the information is, and in addition, the which most people are not familiar, such as health
time required to collect the information is typically care, inherently poor judgment in deciding what
excessive. Thus, people satisfice by accepting is “close enough” can impair overall knowledge
the first answer which is deemed close enough acquisition. In these cases, where people lack
or good enough. adequate background knowledge, sufficing can
Satisficing also appears in the persistence of be dangerous because they can’t make good
inefficient strategies for interacting with informa- judgments on when to stop or on the quality of
tion (Rosson, 1983; Bhavnani & John, 2000). As the information they have found.

131
 How People Approach Information

Sufficing is also apparent in how most people most of the time. People only shift to a different
learn to use software programs; people stop learn- procedure when the local data of the specific
ing at whatever level allows them to get by. Once situation forces them to. Part of this issue can be
people figure out how to perform a task, they addressed by providing information which allows
continue to always use that same method, often people to understand and build relationships in
even while admitting that they know there is an the layers of context between the specific tool
easier or quicker way. If that means they only and the situation, an area in which most people
know a few functions, then that is acceptable if it are very weak. They understand very well either
meets their information needs, even if their opera- the situational context or the tool, but not both
tion of the software is highly inefficient. Carroll (Bhavnani & John, 2000). For example, designers
and Rossons (1987) call this the “paradox of the might be very good with creating CAD drawings
active user.” Users of computer systems are so or with understanding the engineering situation
consumed with immediate goals that they are not being designed, but they rarely can do both at an
motivated to take time out to learn better ways of expert level.
accomplishing their tasks. Grayling (2002) uses
sufficing as a potential explanation as to why many Superstitious Behavior
self-identified experts do not perform better in
usability tests than novices. Since their methods Closely related to sufficing is the superstitious
were sub-optimal, they did not perform any bet- behavior people display with software interfaces.
ter than people having to figure out the problem. This does not refer to the many long-standing social
Of course, a self-identified experts may not be as superstitions such as warnings against walking
expert on the system as they think. Instead, they under a ladder, but to consistent (although highly
may understand only a subset of functions which inefficient) ways people have of interacting with
they routinely use and may even have sub-optimal software. Superstitious behavior is a confusion
use of those. of correlation and causality; people tend to repeat
When people engage in satisficing with proce- actions that are reinforced.
dural tasks, there are some predictable elements In figuring out how to perform actions with
design teams can look for. Fu and Gray (2004, software, people overwhelmingly use trial and
p. 901) found that the procedures which people error (Spool 1996, 1997). In the process, they
followed bore two characteristics: find something that works and continue to use
that sequence. Thus, they have performed an ac-
• “The preferred procedure is a well-prac- tion and have received reinforcement. In many
ticed, generic procedure that is applicable situations, the method found by trial and error
either within the same task environment in works, but is highly inefficient, as Zimmermann
different contexts or across different task and Vanderheiden (2005) explain:
environments,
• The preferred procedure is composed of By superstitious behavior, we mean that a person
interactive components that bring fast, in- continues to do something in a certain way that
cremental feedback on the external prob- does not correlate with necessity -- just because
lem states.” it worked that way before. For example, a phone
user might always turn off his phone at the end
Fu and Gray argue that the preference for a of a call because his previous telephone required
general, generic procedure privileges the global him to turn off the phone in order to hang up. Or,
data in the situation and is used because it works certain computer users might always fully delete

132
How People Approach Information

and reenter information into a new record rather havior, modifications to the design that usability
than just editing the old record -- a vestigial tests show increase interface efficiency will not
behavior associated with older technology. Or, improve performance since they will not be used.
someone who does page layouts might always use It can also prevent communicating the content
a certain format because she received it as a rule since people will not work to manipulate the
of thumb from another designer—not because she information, but will accept the one presentation
understood the basis for that designer’s decisions. they know how to produce.
In other words, people engage in superstitious
behavior when they continue to do what they
used to do without questioning or understanding MULTITASKING AND
why—even if the action is no longer necessary TASK SWITCHING
or sufficient within the current context (Note 10).
Consider this sequence of events:
After finding something that works, people
continue to perform the action in the same manner A professor sits at a computer, attempting to write
and refuse to consider a different approach. Raskin a paper. The phone rings; he answers. It’s an
(2003/2004) points out that the real problem administrator, demanding a completed “module
follows after they have found an answer. Since review form.” The professor sighs, thinks for a
people don’t understand how they made the system moment, scans the desk for the form, locates it,
work, all they can do is repeat the sequence they picks it up and walks down the hall to the ad-
discovered without any variance. ministrator’s office, exchanging greetings with a
colleague on the way. Each cognitive task in this
No wonder we tend to act as if computers are quotidian sequence—sentence-composing, phone-
run by magic. Many of us (including me) use the answering, conversation, episodic retrieval, visual
exact sequence of operations for a task because search, reaching and grasping, navigation, social
it worked once and we don’t dare to vary it (even exchange—requires an appropriate configuration
when somebody suggests a different method) (p. of mental resources, a procedural “schema” [men-
12). tal model] or “task-set” (Monsell, 2003, p. 134).

As with sufficing, people have found some- Each of the tasks described in the quotation
thing that works and see no reason to change. Or has a separate mental model which the professor
they may have tried performing the task differ- must switch to before performing the task and
ently and found it didn’t work, thus reinforcing then switching back to a previous one. In this
the superstitious behavior. Typically, they do not example, each event happened in a long, linear
understand why the known sequence works, so series, but typically multitasking involves per-
they cannot figure out what can be changed or forming multiple tasks at once with rapid switches
modified without breaking it. between them. Multitasking can be compared to
This can lead to sub-optimal performance, the vaudeville act of balancing spinning plates.
even on well-designed systems that support very The performer must run from pole to pole to keep
efficient methods of accomplishing the task. This the plates spinning and to avert disaster. A good
type of behavior also shows why design teams performer can keep many plates spinning; a poor
need to understand the whys behind usability test one can spin only a few. And if any plate proves
results, rather than just the actions themselves. difficult to keep spinning and requires extra time,
When people are engaged in superstitious be- then they all crash.

133
 How People Approach Information

Multitasking involves performing several clicking on links or manipulating windows) are

different tasks at the same time. When people two independent tasks, which must be performed
multitask, they have to constantly shift cognitive simultaneously. Additional tasks are typically
resources around (Figure 3). But limited resources required because of the reason why people are
do not allow people simultaneously to keep all of reading the information. For example, using a
the tasks in working memory. Instead, a substan- help system forces multitasking on several levels
tial time penalty is incurred as task information since, besides working with the original interface
is moved in and out of long-term memory. It will and its information, people must also contend with
take longer to multitask than to perform the tasks manipulating the help interface and understanding
individually. The balance of cognitive resources the help information. In addition, factors such
between the tasks has strong influence in how as the help window blocking part of the original
effectively and efficiently the multitasking is application, forcing people to remember informa-
performed. tion, add to the cognitive load and multitasking
Burgess et al. (2000) list three major skills difficulty.
required for multitasking: Although it may appear that people are per-
forming multiple tasks simultaneously, one task
• The ability to create and mentally schedule is receiving the central focus, while the other
future intentions. tasks are rapidly cycled into that central focus. It
• The facility to remember those intentions requires cognitive resources to perform that mental
and to prioritize them. shift, resources in addition o those required for
• The ability to switch from one intention to the task itself. As a result, operations which force
another. people to multitask are not as efficient as they
would be when performed alone. It typically can
Computer-based operations always force take 20-40 percent longer to accomplish a task
people into two tasks: reading content and manipu- via multitasking than by doing the separate tasks
lating the interface to show that content. The acts singularly because of the overhead of mentally
of reading or evaluating contextual information switching between tasks (Rubinstein, Meyer, D.
and interacting with a computer interface (such as, & Evans, 2001).

Figure 3. Multitasking involves moving task information rapidly between working and long-term memory.
In this example, a person is working on tasks 1-4. Since task 2 has the current focus, it is in working
memory while the other three tasks have been shifted to long-term memory. When the focus shifts to task
3, then task 2 is moved to long-term memory and task 3 moves to working memory. Lack of cognitive
resources in working memory prevents keeping all four tasks in working memory at once.

134
How People Approach Information

Each task requires attention to different in- of switching-time costs, in both the physical
formation elements (such as content or interface and cognitive sense, has been done for various
widgets), or retrieval from memory of prior types of tasks (Kieras, Meyer, Ballas, & Lauber,
knowledge, to integrate into the current task. How 2000). There is a switch cost, which increased
well people can handle these three factors and the the response time relative to non-switched tasks;
mental switching requirements determines how individual steps of the task can be performed faster
well they can multitask. People show a wide range with a single task than a multitask. Task times are
in their abilities, from being very efficient and able slower and error rates higher immediately after a
to truly multitask, down to handling the multiple switch, but these effects are seen long-term, prob-
tasks in a linear fashion (Zhang, Gonetilleke, ably because of issues of shifting mental models
Plocher, & Lang, 2005). (Monsell, 2003) and knowing another shift will
Cognitive resources are limited and how ef- be occurring. A preparation effect has been seen
ficiently people use them is a major determinant where knowing the task switch was coming and
of multitasking ability (Baddeley, 1986). Effi- having time to prepare will decrease (but not
cient multitasking performance in complex task eliminate) the switch cost effects. Of course, in
environments is dependent on skillful allocation most multitasking situations of interest it HII, there
of resources (Vu et al., 2000). For example, Sch- is minimal preparation time available.
neider and Fisk (1982) had people perform two The switch costs vary based on how complex
visual search tasks concurrently: one for which and familiar the tasks are. As would be expected,
processing should be automatic and the other it is harder to switch from one complex task to
requiring attention. They found performance another complex task. The level of familiarity with
in the task requiring attention suffered because the task also affects switch cost; it also takes more
people devoted excess resources to the auto- time to switch from a familiar to an unfamiliar
matic task. When the people were instructed to task than from an unfamiliar to a familiar task
strongly emphasize the attention-requiring task, (Kushleyeva, Salvucci & Lee, 2005; Rubinstein,
they performed much better with no decrease in Meyer, D. & Evans, 2001).
performance of the automatic task. The sidebar A potentially serious impairment to multitask-
discussing the problems of laptops for the San Jose ing is that it requires people to balance their limited
police force illustrates an example of this type of cognitive resources between the tasks. However,
dual task. Driving is learned to automaticity, but rather than mentally scheduling cognitive tasks
working with a laptop requires attention. In this in proportion to their significance and cognitive
case, it required too much attention and diverted demands, people use an equal-scheduling method,
needed resources from the driving task. assigning each task the same amount of cognitive
Unfortunately, design commonly focuses on effort (Figure 4). The result is a mismatch of cog-
helping people accomplish one task at a time, nitive resources to demand. Some tasks receive
providing little help in arranging the tasks into insufficient cognitive resources while others get
effective work orders, or supporting switching too many (Langholtz, Gettys, & Foote, 1995).
between multiple tasks (Cypher, 1986). Virtually
every study that looked at switching between Deciding When to Switch between
tasks (from task A to task B to task A) found a Tasks
significant increase in time and errors as com-
pared to performing the same tasks sequentially. When people are multitasking, a simplistic view
People cannot just switch between tasks with no is that they are performing several tasks simulta-
cost in either mental effort or time. The analysis neously. However, the human mind cannot really

135
 How People Approach Information

Figure 4. Mental scheduling of cognitive resources. The circle size corresponds to the amount of cogni-
tive resources the task has allocated. People tend to use equal-scheduling, rather than optimal schedul-
ing. In equal-scheduling example all three tasks receive the same resources so task 3 receives too many
resources and task 2 receives too few.

Box 11.
perform next and when. They describe this internal
Teenagers and multitasking
monitoring in terms of an increasing “pressure”
Many parents are amazed at how their teenagers can multitask, over time that people working on one task feel
such as being able to text-message multiple friends, listen to
music, and study at the same time. However, the research into
as the need to switch to another task increases.
this ability shows that these teens suffer the same decreased The internal pressure arises because in any
performance as seen in older people; they are, however, able sequence of tasks, some are more important than
to shift between tasks better. The overall comprehension and
learning of new material (the textbook they are studying) others to achieving the overall goal. People want to
shows a marked decrease compared to single-task perfor- stay on the tasks they perceive as most important.
mance. Also, most of the tasks they perform do not require a
long-term memory component. Texting with multiple friends, In many tasks, such as assembling a bicycle or
watching a movie, and listening to music produce little infor- completing an online form, the desire to complete
mation which must be commented to memory. On the other
hand, if they are also studying history, then they will show the task overrides the need to find information
comprehension of the material. about properly completing it, which provides one
clue for why people don’t read documentation.
Reading the document and assembling the bike
focus on multiple tasks at once. Instead, people requires people to switch between tasks: read step,
rapidly shift between tasks, with each task receiv- do action, read step and relate to previous action,
ing the bulk of attention for a short time. The do action. If the effort to figure out the step and
previous section considered the time and error relate it to a previous step is too difficult, then
costs of making that switch; this section looks they tend to minimize cognitive effort by ignor-
at what makes people decide to switch between ing the instructions and focusing on figuring out
tasks. The switch between tasks does not happen (guessing at) the actions. Difficulties in refinding
automatically. Something in the situation cues the proper point in the instructions caused by poor
people to expend the cognitive resources to make design can also make people not want to use them.
a switch. Design teams need to recognize those Design teams must understand what tasks
factors and ensure that the design provides cues. people are performing and how simple or com-
The motivation for a time-to-switch signal, plex they view them to be. The design may have
especially for information web sites, is internal, to signal that a change is needed (that something
not driven by external factors (Burgess et al., on the second task requires attention).
2000). Emerson and Miyake (2003) found that
people are silently telling themselves what task to

136
How People Approach Information

Task Interweaving facilitate multiple activities, they will be more

comfortable for the user (p. 244).
Normal task analysis and most designs assume
people accomplish one task at a time in a linear Part of the problem can be traced to divorc-
manner: they start a task, finish it, and then move ing design requirements from their situational
on to the next task. Although people may describe context. The requirements call for a task and the
their task performance in this linear sequence, system performs it, but it fails to fit into the flow
research shows that they tend to use an opportu- of related tasks. People typically have multiple
nistic organization and interweave tasks (Visser information needs which do not necessarily appear
& Morals, 1991). Cypher (1986) found that while in any fixed or linear sequence. The design needs
people say they perform tasks in simple sequences, to consider how to support multiple undefined (at
in reality they interweave multiple tasks. Often design time) paths for information acquisition.
they stop one task and perform another simply
because it is convenient to perform it right now. Task Interruptions
People interweave tasks in a “while I’m at it”
manner (while I’m going to the grocery store, In addition to considering multitasking situa-
I’ll stop for gas on the way). This can range from tions where people voluntarily take on tasks and
working with another file because that folder hap- switch between them until they are complete,
pens to be open to doing a task because they will task interruptions must also be considered. Task
be in a room down the hall. Information search interruptions break the task switching cycle and
and analysis is often filled with many “while-I’m- force people to consider a completely different
at-it” kinds of activities. These can range from task. Task interruptions include phone calls, pop-
the normal “guess I’ll do this now” to task shifts up chat boxes, and pinging sounds of emails. A
that relate to the information itself, such as when phone must be attended to, while the email ping
people stop to look up a definition or decide to can be ignored (McFarlane & Latorella, 2002).
get more information about a sub-topic to help Storch (1992) found that pop-up windows were
understand the current topic. highly disruptive and led to increased errors. In
Cypher (1986) claims that user dissatisfaction general, their effect on decreased performance
results from a failure of the system to support lasts much longer than the interruption itself.
opportunistic work of interweaving tasks. Al- In a business-based study, each employee only
though the system supports the task, the task is spent 11 minutes on any given project before be-
not performed in the same manner people want. ing interrupted and took an average of 25 minutes
Unfortunately, current design methodologies to return to the original task and often performed
give little emphasis to helping arrange the tasks more than just the task causing the interruption
into effective work orders or supporting multiple before returning to the original task (Gloria Mark,
tasks at once. Cypher puts the problem squarely Gonzalez, & Harris, 2005). Then, upon returning
on the designer: to the main task, they had to mentally return to
their location at the point of the interruption and
Program designers put a great deal of effort into resume the task. With highly complex tasks, this
allowing users to perform single activities well, but proves difficult.
considerably less effort goes into allowing users This was clearly brought out in the work by
to arrange those activities. If computer systems Czerwinski, Horvitz, and Wilhite (2004).
are designed so that they actively support and

137
 How People Approach Information

The findings suggest that methods for capturing can give to the overall text. The fundamental
and remembering representations of tasks may be problem is that since attention is driven by limited
valuable in both reminding users about suspended cognitive resources, when people improve their
tasks, and in assisting users to switch among the knowledge of one aspect of a situation, it often
tasks. Examples of such methods include time- occurs at the cost of not gaining information about
centric visualizations and tools that can record, another aspect (Endsley, 1995; Wickens & Hol-
and reconfigure upon demand, the layout of lands, 2000). In other words, people cannot and do
multiple windows of content and applications not view and interpret everything on a display or
that comprise a task (p. 7). page. Instead, they select the items which seem to
have the highest salience and focus their cognitive
Interruptions have three phases: before switch, processing on just those items. When a mismatch
during, and after. Design teams need to acknowl- occurs between the most salient information, the
edge that interruptions occur frequently and ensure most important information for the situation,
that a design supports them. System design should and their desired information, they have to work
support making a smooth switch to and from. harder to decipher which information is of current
Having the ability to easily save and return to the interest, creating a risk that the focus will go on
current work is vital. It should also help to allow the wrong information.
people to switch when it is convenient, rather than A typical misplaced salience problem occurs
forcing an interruption. when the display contains lots of easy to collect/
Interruptions are going to occur as people in- display information which is not highly relevant
teract with information. The system design needs to understanding the situation. To comprehend
to allow for those interruptions and minimize the situation properly, people have to know what
their impact. information they need and pull it from the noisy
background of irrelevant information. With low
prior knowledge, the person lacks the ability to
SALIENT INFORMATION sort relevant from irrelevant noise. Another ex-
ample of misplaced salience is the use of bold text
Information salience is defined as the prominence to make a word or phrase more salient. People’s
given to an information element. The most impor- eyes are drawn to a bold word immediately upon
tant information should have the highest salience. looking at a page. However, if the word does not
Any information system has an overabundance deserve to be bold, then that eye movement is
of information that can quickly lead to cognitive distracting and interferes with comprehension. At
overload (Thuring, Hannemann, & Haake, 1995). the extreme, the overuse of bold text causes the
The raw information about a the situation is often non-highlighted text on the page to gain salience
at salience levels which may not match its im- because the eye is drawn to the text which looks
portance. Design teams need to transform those different. With too much bold, the non-bold text
salience levels into what people require based now looks different from the general page. Like-
on their information needs. Design teams must wise, information at the top of a page receives a
ensure that all of the information the reader sees higher salience value than one lower on the page.
has proper salience to maximize communication The information placed at the top must deserve
and to minimize the risk of information overload. its higher salience.
The importance of the design team having Information salience plays a major factor in
clearly defined a hierarchy of information salience people’s missing information or forming incor-
comes from the limited amount of attention people rect conclusions. People gain an understanding of

138
How People Approach Information

the situation by following “a long and recursive decision they want to make (Ganzach & Schul,
process with backtracking and erratic switching 1995). In most cases, all of the factors will not
among the following activities: thinking about be viewed. Rather, they have a solution in mind
ideas, production, reorganization, modification, and will only look for confirming information.
and evaluation” (Nanard & Nanard, 1995, p. Because people look for confirming information,
50). Thus, supporting erratic methods of under- not disconfirming information, specific elements
standing requires providing salient information that can provide disconfirming information must
when people need it, and in the proper form. The be uncovered in the analysis and given a high
important information must be clear enough so salience level.
that they do not overlook it. Incorrect conclusions Expectancy bias leads people to see the
can be caused by focusing on some information expected answer (Klein, 1999) even when it is
while ignoring other information which would not really there. As a result, they find it easier to
disconfirm the assumption. It could also be caused extract confirming information while ignoring
by improper information salience, which causes disconfirming information. In addition, people
partial or irrelevant information to dominate tend to change their assessment strategy to fit to
people’s view, causing them to reach an incorrect the presentation method, rather than transform
conclusion. Plus, as they gain an understanding of the information to fit a better assessment strategy
a situation or the situation evolves, the informa- (Johnson, Payne, & Bettman, 1988). In other
tion which deserves the most salient presentation words, people will not mentally highlight the
can change. important information they know should have a
People think they use more information than high salience to give it the salience it requires if
they actually do. They tend to underestimate the the presentation makes it difficult to do so.
weight placed on important cues and overestimate Unfortunately, many designs fail to follow
the weight placed on unimportant cues. The salient good practice for presenting salient information
information that grabs attention can unduly influ- and end up with one of the following problems:
ence a decision. Experts are just as susceptible to
this as novices (Andriole & Adelman, 1995). Thus, • All information receives equal salience.
in a retrospective interview during a usability test, When the design team does not understand
it is hard to get clear answers from people about the information, they cannot determine ef-
the salient information and how they used it. fective salience, so everything is presented
If most people are asked what type of infor- equally. Essentially, they take the view of
mation they consider most salient, they answer “the user knows what they want, so they
something along the lines of “information to can figure it out.”
show I’m making the right choice.” Contrary to • Overuse of salience indicators, such as
that information want, the information people bold fonts or colored text. Overuse results
need is not that which shows they made the right in the non-salient information becoming
choice, but rather is the information which might more prominent.
reveal that they made the wrong choice. In other • Minor but easy-to-present information oc-
words, too much confirming information is actu- cupies an excessive amount of the display
ally noise, although people think they want it. The area and consequently diminishes the sa-
design importance for disconfirming information lience of important but harder-to-present
comes from the way people make choices. As the information. Providing large sets of easy-
amount of information increases, people increas- to-obtain numbers or images can overpow-
ingly ignore information incompatible with the er the presentation of smaller sets of more

139
 How People Approach Information

Box 12.
define what are the signal and noise elements.
The original voice spoken into the transmitter is
Information salience of tag clouds
Tag clouds provide information salience by varying the size of a known quantity and anything else that comes
the text, with the largest size being the most important. Using out the speaker is noise that should be eliminated.
www.tagcrowd.com, the following tag cloud was created for
this section on information salience. The word size is based
In an information system, “signal” consists of
on word frequency. On a webpage, links could be created to anything that people need to know but don’t, and
connect the cloud text to relevant text. The actual usefulness of
tag clouds is still being debated.
“noise” is anything that they don’t need because it
is irrelevant or already known and is just clutter-
ing up the page or screen to the detriment of the
signal. Tufte (1983) was referring to essentially
the same concept when he discussed data/ink ratio
and chart junk. Basically, noise consists of all the
stuff that interferes with communicating what
people need. Any time noise receives salience,
the comprehension of information is impaired.
In the hindsight provided by investigations of
bad decisions (the type that led to millions of
dollars of losses or damage), the excesses and ir-
relevancies of information are obvious, but they
are never obvious to the person while making the
relevant numbers. The screen becomes full decisions. The difficulties people have finding or
of information that is easy to capture and comprehending information appear repeatedly in
display, but has no bearing on the current these studies. In many cases, they come down to
user goal (Hsee, 1995). For example, the low SNR with high levels of extra information
long tables giving the technical specifica- blocking the communication of the information
tions for a product when most people do which was really needed.
not need or understand that information. A huge number of documents on corporate
• Design decisions become driven by corpo- intranets were placed there “in case someone wants
rate politics, such as increasing the salience to read it.” McGovern (2006a) points out that a
of features or upgrade options, which dis- couple of studies of web sites removed over 30%
tract people from the information they re- of the pages and no one noticed. Pages that no one
ally want. looks at waste both the author’s time and system
resources, but as long as no one looks at them,
they do not interfere with interpreting relevant
Signal to Noise Ratio information. On the other hand, a communica-
tion problem occurs when excessive or irrelevant
The signal to noise (SNR) ratio is a common information is provided. Unfortunately, people
concept in radio communications and electronic often cannot easily see that they either have too
communication in general. For a radio, the static much or irrelevant information. Instead, they try
is the noise. Too much static and a storm report to process all of it, which leads to problems. The
gets drowned out, or at least you must listen useful information—the signal—is swamped by
closely to understand the announcer. Electrical the non-useful information—the noise.
engineers working on a voice circuit can easily Unfortunately, design teams do not possess a
clear-cut set of techniques, such as a radio circuit

140
How People Approach Information

band-pass filter, available to electrical engineers. information they need, but they have difficulty if
For information systems, taking the raw data in they do not know what they need or whether the
a system and deciding what is signal and what is information even exists. Factors contributing to
noise is extremely difficult. A major problem is a low signal to noise ratio include:
that each person has different needs, which means
that different people perceive a different noise level • Lots of information exists online (espe-
in the same text. For example, many documents cially in corporate intranets) which was
carefully define their terms; yet, if people already placed there “because someone might
know those definitions, then they become noise. Of want it.” The result is lots of information
course, for a more novice reader, the definitions are of minimal relevance to anything. The
essential to understanding the document. Because trend has been to make more knowledge,
of the different goals and information needs each such as best practices, success stories, pol-
person brings to the system, there will never be icy books, and training manuals available
a clean method of resolving the signal to noise online. However, successfully using that
ratio. Rather, task and audience analysis need to information also means that people must
refine the understanding of what is signal, what have access to the relevant knowledge in
is noise, and most importantly, what contextual the context in which they need it (Schwartz
situations cause noise to become signal and vice & Tefieni, 2000). “Contextualized access
versa for each of the audience groups or personas. means access to relevant task knowledge
An excess of noise can occur from either too is immediate and within the problem-solv-
much or too little information. Too much infor- ing context rather than via searching in a
mation causes information to be ignored since separate context. Thus, there is no need to
it takes too much effort to extract it. Too much break the continuity of task performance to
information can result when the design strives seek relevant domain knowledge” (Mao, &
for completeness over clarity, gives information Benbasat, 2001, p. 788).
that is not salient to the current situation, or gives • Graphics tend to out-compete text for peo-
information people consider irrelevant (or perhaps ple’s attention. The eye is naturally drawn
doesn’t understand that it is important). On the to graphical images and graphics tend to
other hand, the problem of not enough informa- be remembered better. Of course, that does
tion can arise when the initial analysis didn’t not mean they are more important.
identify the needed information. It can also arise • Many computer systems present lots of
in a “one-size-fits-all” design. The information easy-to-obtain/display information which
is constrained so it is appropriate for a subset of hides the much smaller amount of impor-
the potential readers, such as low-literacy read- tant but hard-to-display information.
ers. While an important design criterion for the • Poor design can give high salience to irrel-
low-literacy group, it fails to provide adequate evant information or give all information
information to other groups that want or need the same salience, which forces people to
more information. work harder to identify what they need.
People often have a difficult time sifting out
relevant from irrelevant information. Low signal The real HII difficulties in communicating
to noise ratio makes it hard to distinguish and information do not fall within the technical realm.
mentally rank the appropriate information (Albers, Rather, they fall within the people realm, which
2004b). People are very good at pulling informa- revolves around the contextual aspects of the in-
tion out of a noisy environment if they know what formation. The technical realm solves individual

141
 How People Approach Information

problems by placing widgets on the interface and becomes noise for the supervisor and proper signal
developing algorithms for an efficient back end. In for the supervisor is inadequate for the technician.
the people realm, the goals and information needs
must fit within people’s cognitive capabilities and SNR in Interface Design
social environment. They have goals they want to
address and problems they want to solve; helping System interface design provides a striking ex-
to achieve those goals and solve those problems ample of decreasing SNR as viewed over time.
requires systems that provide salient information Each new release of a product contains more
appropriate to the situation. In other words, they functionality but this functionality often comes
need a strong signal with minimal noise. at the price of lower usability. Norman (2002)
Maximizing the signal means understanding calls this phenomenon creeping featurism and
both people’s terminology (versus system termi- suggests that the complexity of use increases as
nology) and their situation. Quite simply, anything the square of the increase in functionality. In other
written in unfamiliar terms will be considered ig- words, twice the functionality means a product
nored, regardless of its importance. But more than is four times as hard to use. Shneiderman (1997)
just the proper terminology, the text must fit within considered the issue of screen density and how the
the situation. For example, medical information amount of information displayed relates to how
and business balance sheets have formats fixed by easily people can use it. It is not a straightforward
long tradition. Changing the format could actually issue of low density equaling ease of use, but rather
cause someone to ignore the entire page with a that all of the information must be relevant. People
logic of “this doesn’t look like the XX report, so it like very dense screens if the content is relevant
doesn’t have the information I want.” The change and well organized. In other words, screens with
renders the entire page as noise. a high SNR can be very dense and screen with a
The difficulty of resolving the SNR occurs at low SNR fail regardless of density.
many different levels. For example, it is reflected The inherent problem in screen density is that
in the basic ideas of the differences in writing for any use of an interface distracts readers from ac-
novice and experts or writing for high and low
literacy levels. The extra explanation required for Box 13.
the novice becomes noise for the expert, which
impairs communication. Interestingly, readers Customer information screen
I once worked for a film processing company where we were
with a high knowledge level in the general subject building new customer service software. The system had a
tend to have poor comprehension of text writ- customer information screen. For this company, the customers
ten for a low-knowledge reader (McNamara & were the retail stores where people dropped off a roll of film
for processing. The customer information screen contained all
Kintsch, 1996). The extra detail bogs them down of the information pertaining to the customer and was 3-4 full
and results in less focused reading, resulting in screens long. However, the first part of the screen was basic
customer information such as address, phone number, etc.
missed information. The reverse also occurs:
without more detail, a written-for-an-expert text The design problem was that all of this information was noise.
The customer called the customer service representatives with
degenerates into noise since it can’t be understood questions. The customer knew their address and the customer
by a non-expert. A similar situation also occurs service representative didn’t normally care. Instead, they had
to scroll down multiple screens to get to the invoice informa-
in the fundamental information requirements for tion relevant to a typical call.
a job. A technician needs specific details about a
From a data entry view and setting up a new customer, the
task while a supervisor wants a higher-level, less overall screen design made sense, but from a customer service
detailed view. As a result, signal for the technician view, the important information was hidden.

142
How People Approach Information

complishing their goals. People don’t want to use of current interest, increasing the risk that they
the interface; they are forced to use it as a means will focus on the wrong information.
of reaching their goal. In reality, they are willing Essentially any formatting which makes text
to work with an interface since that is the simplest different from the surrounding text will increase
path to obtaining information. Thus, they should its salience. The use of colored text in reports to
be given a high SNR to support their information direct manager’s eyes to high or low values is
interactions. one example (Table 1). Skimming the table and
figuring out which values are high or low and may
Designing for Information Salience require further investigation can be difficult since
the values are different for each product. Flagging
A single salience value cannot be assigned to an the text in some manner increases the salience of
information element. Information salience is not the values of interest.
“one-size-fits-all.” The amount of salience it de- Complex sets of information can offer many
serves depends on a wide set of factors. Improper different types of information of interest, which
information salience can cause partial or irrelevant a design team may attempt to flag with different
information to dominate people’s view, causing indicators. Readers must figure out on the fly,
them to make the wrong conclusion. Besides the from context, the specific meaning of the empha-
obvious point of audience needs, information sized text (bold, italics, square brackets, or other
salience also varies over time. As a situation typography) and how to relate the flagging to its
changes, information importance increases or salience level. However, people cannot mentally
decreases. Usability and comprehension problems figure out and keep track of more than two to
occur when information salience remains fixed. three formatting methods without extensive train-
For best comprehension, information needs to be ing. Overuse of many methods of signaling sa-
provided only when people need it. Presenting it lience levels causes all of them to be ignored.
too early or too late is a distraction: what could Having many different types of emphasis on the
have been valuable information simply reduces page imposes too high a cognitive load to be worth
the signal to noise ratio by adding to the noise. the effort to figure out the meaning of each. In
Information salience plays a major factor in general, if people don’t understand why text is
people’s missing information or forming incorrect typographically different, that difference will be
conclusions. The important information must be ignored. In contrast, dense information displays
clear enough so that people do not overlook it when which people use all day long can have many
it is crucial. An issue with good design for effec- different types of indicators, such as fonts or
tive information salience is that design teams’ and colors, which help increase the information sa-
readers’ definitions of important information can lience in specific conditions. But in this instance,
differ greatly. As a result, their expectations differ the people are very familiar with the display, have
about which information should receive the high- time and training to get past the learning curve,
est salience. In addition, design decisions driven and are motivated to learn the different meanings.
by corporate requirements, such as emphasizing However, for people engaged in occasional system
features, upgrade options, or legal issues, shift the use, these same indicators become noise and will
emphasis away from people’s information needs be ignored since they do not understand their
and may actually distract them from the informa- meaning.
tion that they do need. The result is that they have For the complex information relevant of many
to work harder to decipher which information is HII situations, simple signaling of information
does not is not fully address people’s needs. The

143
 How People Approach Information

Table 1. Report with out-of-spec values flagged. Typically, the values would be printed in color. But
other flagging for salience can be used for black and white printing, as done with this example with *
and ** for high and high-high values and < and << for low and low-low values.

Monthly Sales Volume

Broadman Mathison Smith Wagner
Product A (180-220) 203 225 190 173<
Product B (170-210) 193 168< 194 253**
Product C (30-37) 34 22<< 39* 35
Product D (850-950) 903 860 920 884

use of color or bold for single numbers or words Box 14.

does not highlight the information relationships
which are vital for understanding the information. Over-signaling for design conventions in a manual
Many software manuals have a table in the front matter which
Design teams need to ensure that they place the describes the conventions used in the manual, such as:
salience at the proper level and highlight informa-
Bold type Word or characters you type
tion and information relationships rather than data.
The proximity of two information elements to italic type Specialized terms. Placeholders for information you
supply, such as filenames. So you may see cd filename.doc
each other also strongly impacts their relative
salience. Pomerantz and Schwaitzberg (1975) monspaced font Examples of field syntax

demonstrated that relating or comparing informa- small caps All key names are in small caps.
tion requires both information elements to be close
The problem with using too many of these conventions (I’ve
together and increasing the distance between ele- seen more than these four in some manuals) is that people
ments decreases the information salience. Thus, can’t figure them out or easily remember them. People don’t
read the front matter; instead, they have to determine the cod-
information which needs to be processed to- ing on the fly. Having too many conventions creates a ransom
gether, must be positioned as a single unit. Wick- note effect in the text. Instead of figuring them out, people just
ignore all of them.
ens & Carswell (1995) have defined this need for
proximity of information as grouping “like with
like,” essentially matching Gestalt theory. How-
single design as meeting a “like with like” group-
ever, exactly what is meant by “like with like”
ing requirement. Current trends toward adaptive
depends upon the mental model and the informa-
and dynamic information may help, but design
tion needs of the reader. What is considered the
teams still need to understand audience needs
most salient information can vary immensely with
from the beginning of the project.
different mental models. For example, a program-
Here are some design considerations for help-
mer may want to group all items from a particu-
ing ensure information receives proper salience:
lar database table. A decision maker, on the other
hand, may want items which should fluctuate
• Highly salient cues get more attention un-
together, or which influence each other, to be
less people are forced to learn to ignore
grouped together. Furthermore, as people interact
them. Bright colors or flashing text draws
with a situation and conditions evolve, informa-
the eye and makes those elements more
tion salience and the information which should
salient. However, the overuse of these el-
be grouped together changes. In all of these
ements for advertising on web pages has
cases, only some readers would consider any

144
How People Approach Information

resulted in banner blindness, with people ible information in decision-making (Fu &
mentally tuning them out. Attempts to Gray, 2004).
draw attention to highly salient informa- • People evaluate information based on
tion using these methods will often back- the order in which they receive it (Klein,
fire, with people purposely refusing to look 1999). Changing the order of presentation
at them. Nielsen’s (2007) study found that can change the relative salience of infor-
people didn’t see the large red numbers mation viewed later. Note that the desired
for the US population on the US Census order can vary depending on the situation,
Bureau’s website (Figure 6-13). reader knowledge, or point of time within
• Top locations and larger presentations are a dynamic situation. Systems that strive to
more salient and over-weighted, especially always provide complete information often
for people under time pressure (Wallsten & violate ordering because completeness can
Barton, 1982). come at the expense of providing a subset
• Repetition of information increases its sa- of information in the proper order for the
lience. If information is repeated in mul- current situation.
tiple places, people will give it a higher • Complex displays or visual presentations
priority. with many elements competing for atten-
• Images, especially those of people, have a tion cause all of the elements to have a
higher salience than text. Readers tend to lower salience (Janiszewski, 1998). Visual
always look at pictures of people first, with complexity could distract viewers from the
the gaze going directly to the eyes. The im- intended messages.
portance of the text then gets judged based • The information requires good visual mo-
on the image. Irrelevant images or eye- mentum (Woods, 1984).
candy images can cause a reader to regard
the text as irrelevant or less than credible. Timing of Presentation and
• Information which is not visible has a Information Salience
lower salience (Hallgren, 1997). Thus,
any information not currently being dis- As the proceeding discussion makes clear, a single
played receives a lower priority than what design that will maximize salience cannot be
people are currently looking at, regard- defined. The amount of salience an information
less of whether the unseen information element deserves depends on a wide set of factors
has scrolled off the screen or is on another which change, both between situations and within
page (print or online). People find it dif- a situation. Information salience is not “one-size-
ficult to remember or form goal-oriented fits-all.” Only a thorough situational analysis can
relationships between information being define when and how information salience needs
displayed and relevant information not be- change for various audience groups.
ing displayed (Rubens & Rubens, 1988). Besides the obvious differences between
Screen size limitations, especially with people’s information needs, salience also varies
mobile displays, are problematical since with time. As a situation changes, the informational
all relevant information cannot be dis- importance for any particular element increases
played simultaneously. The human trait of or decreases (Klein, 1999). For best comprehen-
minimizing cognitive effort often causes sion, information needs to be provided only
people to go by memory for non-visible when people need it, and as a result, usability and
information and to over-emphasize the vis- comprehension problems occur when information

145
 How People Approach Information

salience remains fixed (which occurs with most has been performed by Stanford’s Persuasive
web information and with printed text). Informa- Technology Lab, (http://credibility.stanford.edu/).
tion must not be presented until people need it, In an interview discussed in Paiva (2000),
and then only the information they need should be Rosalind Picard reports on an empirical evaluation
presented. If it is presented too early or too late is where people were asked to rate trustworthiness of
a distraction, and what could have been valuable an interface in relation to its design. In particular,
information is reduced to noise and lowers the she cites one participant’s statement: “If it looks
signal to noise ratio of the overall communication. pleasant, I just trust it!” Research has consistently
Another complication in the timing of informa- found that attractive designs and presentations
tion salience is that many people refuse to look work better (Müller, 2007; Norman, 2004). As
for information until they feel they are in trouble would be expected, the use of appropriate color
or have exhausted other options. As a result, they has a significant effect on the quality ratings
often receive information too late to effectively people assign to websites (Cyr, Head, & Larios,
help them understand the situation. In this case, 2010). In a study of healthcare information, people
information salience becomes of minimal use. For separately rated the visual design and information
example, in a study of how information affects credibility, and the correlation between these two
prostate cancer screening, most men received factors was statistically significant (Robins, Hol-
information focused on helping them make a mes, & Stansbury, 2010). In other words, people
screening choice after they had already made a strongly judge the quality of information based on
screening decision and, thus, it was of little value its appearance, rather than on its content.
in their decision (Sheridan et al., 2004). On the Moving beyond appearance, a study by Stvilia,
other hand, the same information, given earlier, Mon, and Yi (2009) found five factors that people
may have been extremely helpful and driven the use to judge information quality as well as some
decision process. In this type of situation, the of criteria which go into making those judgments
proper information may appear to receive appro- (Table 2). The relationships among these five fac-
priate salience from a simple usability perspective, tors depend on the type of webpage; how people
but the overall situation does not reflect it since judge information quality varies based on the
the information was provided too late for any general type of information and does not remain
noticeable effect. the same across all web pages.

Accuracy: Accuracy means having correct infor-

INFORMATION QUALITY mation. More content influences lower-
knowledge readers into believing the infor-
Any information must meet a certain level of qual- mation is more accurate.
ity before people will trust it. People must trust a Completeness: Completeness means the quantity
source before they will be believe it, as Salvador, of information. The contributing criterion of
Barile, and Sherry (2004) point out in their study clarity plays a role here since overly complete
of e-commerce systems. Development of that information leads to information overload
trust depends strongly on people’s perception of and distracts from overall clarity. An excess
the quality of the information. Unfortunately for of irrelevant information, with respect to a
design teams, the level of perceived quality varies person’s current situation, distracts from
among people. Extensive research into information comprehension. Design can also contribute
credibility and perception of information quality to a perception of completeness. The appear-
ance of content such as statistics and quotes

146
How People Approach Information

Table 2. Information quality factors. The criteria on the right hand side use the common definition with
all of them summed up to form the quality factor. (adapted from Stvilia, Mon, & Yi, 2009, p. 1786)

Quality Factor Contributing Criteria

Accuracy Accuracy, credibility, reliability
Completeness Completeness, clarity
Authority Authority
Usefulness Ease of use, objectivity, utility
Accessibility Accessibility, cohesiveness, consistency, volatility

increases people’s perception of credibility quality are factors here. Obviously, high-
(Rains & Donnerstein Karmikel, 2009). quality information would fit this factor;
Authority: Authority reflects the quality of the however, poor, but well-written information
information source. Most guidelines say can have a high surface-level accessibility
that web pages must state who is writing the but lack real information value. In a study of
information. As knowledge level on a topic diabetes sites, Bedell, Agrawal, and Petersen
increases, the quality evaluation depends (2004) found a wide range of accessibility
more strongly on the credibility of the infor- and quality issues which could lead patients
mation source (Stanford et al., 2002). Sites to assign a better accessibility rating to an
with known brands (in health care) were also information-poor site than it deserved.
highly rated for both credibility and visual
design (Robins, Holmes, & Stansbury, 2010).
These studies are more or less in agreement AGE ISSUES
that the general population, not knowing
the credibility of most information sources Age is a significant factor in how people ap-
(thus, not being able to make judgments), proach information, but HII-age interactions are
does assign authority when they have heard much more complex than typically presented in
of the source. most articles. The age range of the people using
Usefulness: Usefulness describes how the in- the information can have a strong influence on
formation relates to people’s immediate the overall information usability and effective-
situation. They tend to perceive information ness of the communication. Age-related differ-
as higher quality when it fits their needs— ences do appear and must be considered, such as
specifically, when it fits their short-term, poorer working memory and reasoning ability
immediate needs rather than longer-term, (Freudenthal, 2001). A problem arises in many
epistemic needs. The more practical the in- design situations where the user characteristics
formation for the current situation, the more are over-generalized. For instance, one signifi-
it will be perceived as high-quality (Jung cant problem is that “older adults” are much too
Lee, Park, & Widdows, 2009). Information complex as a group and too highly differentiated
design factors had much stronger effects on to be considered a single entity. Too many proj-
consumer evaluation than on expert evalu- ects try to consider the “older population” as a
ation (Stanford et al., 2002). single unit, while they would not accept “25 to 40
Accessibility: Accessibility deals with how easily years old” as a sufficiently descriptive audience
information can be obtained from the overall characterization.
content. Overall text coherence and writing

147
 How People Approach Information

Regarding the moderating effects of user age, the Aptitude: Aptitude is about expertise and experi-
findings show clearly, that it is definitively not ence with information situations. Obviously,
sufficient to examine only young users’ technical people who are familiar working with infor-
acceptance and performance and to generalize the mation sets will be able to manipulate them
results to the whole user population. The found more easily than those who are less familiar
age-specific pattern of relationships indicates that (Freudenthal, 2001; Kang & Yoon, 2008).
it is highly important to integrate the older user Attitude: Attitude is about the type of approach
group in studies about user attitudes and behavior taken toward situations. It considers differ-
(Arning & Ziefle, 2007, p. 2922). ences ranging from people who are risk-
takers to those who are risk-averse.
Compared to other factors that affect how
people interact with information, age must be Design teams must take into account all of
considered, but is not the dominate factor. Older these four dimensions when considering age as a
people are often motivated to use computers component of the design. In addition, social and
because they perceive such use as an important cultural issues are just as important, if not more
element for inclusion in contemporary society, and important than just age.
as a method of fighting isolation. The increasing Age is shown to be a factor in a number of
penetration of email and social networking into usability issues with Internet interaction, even in
the older population stems from this motivation. studies that control for experience level. On the
For example, people age 40–59 make up over 50% other hand, it seems that usability problems simply
health information seekers, with 15% over age 60. have a stronger effect on older people than on the
The American Association of Retired Persons young. When design changes were made to fix
(AARP) uses a four-dimension model to describe usability issues, overall performance improved
people’s abilities: in both older and younger groups (Chadwick-
Dias, McNulty & Tullis, 2003). Poor usability is
Age: Physical age is the least important element in a significant factor on many websites failing to
the model. Age by itself has little influence conform to guidelines of the National Institute
on how people interact with information. on Aging (Becker, 2004, 2005). Problems can
For example, in a study of a portable mul- range from fonts and link targets that are too
timedia player, the people’s trial-and-error small to poorly written content. Many of these
interactions and frustration levels were more usability problems affect everyone using the site
strongly influenced by background knowl- and need to be addressed in a global manner, not
edge than by age (Kang & Yoon, 2008). piecemeal. For example, most design guidelines
Ability/disability: Ability is people’s capacity to say to increase the font size for older audiences
perform the physical actions. Ability consid- (NIH, 2002). Although this is valid advice, simply
ers areas such as vision, which degenerates increasing the font size does not begin to address
with age, and motor coordination, which all of the differences design teams need to con-
affects mouse control. People with impaired sider and how changes ripple through the design.
ability will have trouble interacting with Larger fonts make the overall page longer, which
information. A wide range of sources discuss increases the amount of scrolling; yet the same
how to design to overcome these problems, guidelines say to avoid scrolling.
with the baseline coming from American Long-term experience with poor usability gives
Disabilities Act, section 508. rise to the perception that many older adults have
a mistrust of computer systems. However, such

148
How People Approach Information

negative attitudes are more closely correlated with (Birdi & Zapf, 1997). They also make more syntax
computer experience than with age. Marquie, errors, which leads to higher frustration rates.
Jourdan-Boddaert, and Huet (2002) consider the However, the cognitive slowdown does not
low confidence of many older people in their own mean that older readers will always respond more
computer ability and believe it is the main reason slowly; it only affects the information interaction
they have trouble with computers. In general, when the cognitive requirements exceed the task
people who do not believe they are capable of requirements. For example, text size on a web page
performing a task will not make the required effort did not significantly affect performance across age
to succeed and have lower perseverance. Older groups until the font is too small for easy reading
adults show less confidence in their interaction, (Chadwick-Dias, McNulty & Tullis, 2003).
and consequently, tend to rate systems as harder Designs need to work to minimize the overall
to use. Lower confidence also plays out in error cognitive load. Making procedural or navigational
recovery: low-confidence users are less likely to paths easier to remember and making functions
try to solve a problem without outside help. easier to perform can reap major gains in overall
The ease-of-use issue, while important to effec- communication effectiveness. The same task
tive communication, also helps to show how us- performance of Fitt’s law applies, but older adults
ability tests must cover all audience groups. When will tend to suffer cognitive overload at a lower
asked about the ease of use of menu navigation, task level.
younger and older adults used different criteria One area where age differences can be im-
for rating performance. “Younger adults refer to portant is in motor (hand motion) skills (Cerella
their efficiency, i.e. how easy and effort-free they 1985), with major implications for the use of a
experienced the interaction with the device. For mouse (Walker, Millians, & Worden, 1996). Older
older adults, the determining factor in the evalua- people move the mouse slower and with less ac-
tion of the perceived ease of use is effectiveness, curacy. They also have a higher level of perceived
i.e. the success when using a technical device” exertion during click-and-drag tasks, which means
(Arning & Ziefle, 2007, p. 2921). that designs should minimize mouse use if they
are to be friendly to older adults (Chaparro et al.,
Age and Information Interaction 1999). But instead of minimal use of the mouse,
Becker (2004, 2005) found older users performed
Older people do not process information differ- extensive use of movements which require precise
ently, but do show a lower ability to cope with poor mouse pointing ability, such mouseovers. Move-
usability issues (van der Meij & Gellevij, 2002). ment time in most HII situations is often trivial
As people age, their general cognitive processing compared to the cognitive processing time of
speed decreases (Cerella 1985), with most of the reading and interpreting the content.
slow-down attributed to a slowing of how the mind Many usability tests focus on task completion
handles basic operations (Salthouse & Babcock, time. However, in many HII situations, speed of
1991). A direct side effect of this slow-down is task completion is not a significant factor, because
tasks consume a higher percentage of available people are more focused on understanding a set
cognitive resources which can result in cognitive of information and reaching decisions based on it.
overload (Sayago & Blat, 2010). Older people do For simple timed-retrieval tasks, older adults will
not understand poorly designed manuals as well be slower. Likewise, for tasks with high cognitive
as younger people (van Hees 1996). On the other load, such as navigating deep into a hierarchical
hand, older people are more likely to use written structure, pronounced age-dependent time differ-
documentation rather than asking other people ences will be found. Freudenthal (2001) points out

149
 How People Approach Information

that deep menu structures impose a high cognitive needs with the HII required for information
load and are less suited for older adults. As people comprehension.
move deeper into a menu structure, the choice of Each of the roles people play varies, even
the next move is increasingly slowed as memory within a single task, and they can bring different
resources are consumed, making it harder to levels of expertise to each. All of these views affect
mentally handle the complex structures required how people define themselves with respect to the
to interact with the information. This supports novice–expert continuum. When interacting with
claims of a breadth versus depth in menus and computer-based information systems, these types
navigation structures. of differences can be easily confused (Valero &
Older audiences do present challenges for Sanmartin, 1999).
design teams seeking to maximize information
search and comprehension. Because of increased • A new hire can be an expert at a tool, but
vision problems, older people have a harder time not understand the current business rules or
reading a screen. This either forces them to use how to apply them to the job. They know
glasses or to set their monitors at a lower resolu- how to do the task, but now the business
tion, which reduces the amount of text displayed. does it. For example, an accountant knows
A larger font also pushes more material below how to enter account payable records, but
the fold, increasing the difficulty of integrating does not know the coding used.
an entire information set. Still, older people tend • People faced with a new software system
to reject, for as long as possible, measures such may be expert at the task and with the old
as larger font sizes, lower screen resolution, or tool, with many years of experience at the
alternative input devices (Sayago & Blat, 2010). company, but they are a novices with the
new tool. They know what they need to
accomplish, but don’t know how to do it
NOVICE TO EXPERT CONTINUUM with this tool. For example, an accountant
knows how to code the account payable
Design teams often talk about people as ranging charge, but don’t know how to enter it us-
from novice to expert. However, people don’t ing the new software system.
actually think of themselves in such terms. Their • People may have many years of experience
focus is not on a skill level, but on accomplishing with the system, but only know how to per-
a task or performing their job. As such, they think form the tasks required for their own day-
of themselves in terms of: to-day work and have minimal knowledge
of other functions. For instance, they may
• Job operate as an expert with the accounts re-
• Job role ceivable part of an accounting system, but
• What they know about the task have minimal knowledge of the accounts
• What they know about the tools payable or human resources sections. Also,
• Motivational differences they may not have any understanding of
creating reports beyond selecting report
Notice how none of these views explicitly names off a menu.
contains a placement within the novice-expert
continuum. The design team needs to make that Although it is easy to form stereotypes of
translation in order to fit people’s information either novice or expert, in reality, most people
fall somewhere in the middle (Santhanam &

150
How People Approach Information

Figure 6. Movement from novice to expert. People

and quickly as possible” (Hackos & Redish,
start at novice, where they remain for a short
1998, p. 82). Advanced beginners are content
time. They then follow the path toward expert, but
to know a few tasks that allow them to get
will probably stop at some point when they know
their job done. Their use of information, or
what their job requires (adapted from Hackos &
of the system, is incidental and infrequent.
Redish, 1998, p. 88).
They have a poor mental model which often
lacks a basis in knowledge and contains
significant errors.
Competent Performer: Knows a significant num-
ber of the tasks and possesses a sufficient
mental model to support their informational
needs. Can perform more complex tasks and
is sometimes willing to learn more efficient
ways. Santhanam and Wiedenbeck (1993)
found that people in the middle (those who do
not fit as either a novice or an expert), show
expert traits in the areas they consistently
use and novice traits in other areas. Thus,
even within the same information system,
people’s abilities can vary dramatically. For
example, a person may show expert skill in
the graphing part of Microsoft Excel, but
novice skill in page formatting.
Expert: Commands a mental model that is com-
prehensive and consistent. Understands how
to diagnose and correct a wide variety of
problems. Highly motivated to learn more
and to help others. Tends to experiment to
learn more.

Wiedenbeck, 1993). Hackos & Redish (1998) The early audience analysis helps define all the
define the novice-expert continuum as a four-tier groups who may be receiving the information and
structure (Figure 6). Each of the levels contains provides an understanding of the people’s prior
its own HII challenges. knowledge, attitudes, and needs, allowing the
design of appropriate content. Each group’s needs
Novice: A new user. Often a short-lived status. are different and, unless they are understood in
Prior knowledge does not prevent novice detail, providing usable content that addresses all
actions, but can reduce the duration of nov- levels is difficult (Rosenbaum & Walters, 1986).
ice behavior. Novices tend to be very goal- In a review of expertise in HCI, Mayer (1997)
oriented, and want to do, not learn. They concluded, “The results of the expert–novice dif-
transition quickly to advanced beginner. ference studies of computer users are remarkably
Advanced Beginner: Focused “simply and exclu- consistent with corresponding research in other
sively on getting a job done as painlessly problem-solving domains” (p. 792).

151
 How People Approach Information

Novice higher-level structure of the information (Cuevas

et al., 2004). Novices find it almost impossible to
By definition, a novice does not possess knowl- make or justify predictions and inferences since
edge or skill. People may start as a novice, but they lack knowledge. For example, Lowe (2002)
remain so for a very short time before moving found that novices incorrectly ascribe cause-effect
into the advanced beginner stage. Much of the relationships to weather phenomena visible in
technical information design literature tends to typical meteorological animations. Since these
focus on designs for either novice- or expert- subjects lacked a mental model of how weather
level performance. Or, perhaps worse, design develops, they were forced to create one—people
teams seem to assume that a design will only will always impose a mental structure on the in-
be used by one those groups. Unfortunately, an formation—which contained errors. Novices do
over-focus on designing information for novices not look for much information, nor are they able
can degrade comprehension and performance of to interpret it. Instead, they will ask others, and
more knowledgeable people. Thus, most informa- only as they gain knowledge will they look for
tion should not be focused on the novice, but on information themselves.
the intermediate levels where most people fall.
Information designed for a novice often fails to Advanced Beginner
provide the proper level of cohesiveness for ef-
fective comprehension by intermediate and expert People tend to progress quickly from the novice
level people. Surprisingly, more knowledgeable to the advanced beginner stage. Advanced begin-
people work better with less cohesive text (Mc- ners are focused on accomplishing their tasks and
Namara & Kintsch, 1996). possess a very basic understanding of the system.
Carroll and Rosson (1987) point out that Since advanced beginners have an incomplete or
novices are not novices on everything, but are inaccurate mental model, their problem-solving
often experts in the basic area for which they are skills are weak and they are normally unable to
performing the task. Consider these examples: diagnose problems. Information aimed at infre-
quent users should be focused on this stage.
• Secretaries are novices on the new soft- Infrequent use makes many people operate
ware system, but experts in the tasks the at an advanced beginner level, even when they
system is asking them to perform. interact with the system at regular, but intermittent
• A car mechanic using a new automotive intervals. If sufficient time has elapsed between
diagnostic system may be a novice with re- uses of the information, they will have forgotten
spect to using the system (interacting with nuances of interacting with it. An example are
the system to understand the car’s prob- the problems encountered with actions such as
lem) but is an expert at understanding and forms or reports that are only done on a quarterly,
using the information to fix the car. semi-annual, or annual basis. The events occur so
• A master cabinet maker with a new com- infrequently that most people do not remember
puter-controlled planer would show nov- how to complete them efficiently.
ice-level skills operating the planer, but ex- Advanced beginner errors arise from inappro-
pert-level skills at knowing what he wants priate actions because their mental view of how
to produce with it. to approach the situation is incorrect. More than
just being incomplete, it often contains erroneous
A major deficiency in novices is the lack of a concepts and ideas which will be replaced with bet-
mental model that allows them to comprehend the ter concepts as they gain in overall knowledge. As

152
How People Approach Information

Box 15.
their mental model improves, they better interpret
situations and react accordingly (Fu, Salvendy, & Infrequent use turns experts into competent performers
Turley, 2002). As such, domain knowledge is a Technical improvements in the quality of machine parts have
resulted in machines that last longer and need less mainte-
powerful predictor of cognitive performance with nance. As a result, maintenance people have less experience
that domain. Design teams need to explicitly ask with repairing the machine. This causes the maintenance
people, who once had expert-level repair skills, to become
about domain knowledge, rather than allowing
competent performers. Lack of use causes the repair people to
people to self-identify since they typically over- forget many of the low-level interaction details, and to require
rate their knowledge level. that the information be supplied each time. They have become
people who interact with a system at irregular intervals rather
than at the frequency of use assumed for experts (Brinkman et
Competent Performer al., 2001).

Competent performers know how to perform a

significant number of tasks and have a sound current knowledge is sufficient for their needs and
mental model on which to base their information learning more requires expending effort with a
interaction. They are also able to diagnose and low payback. While they are highly motivated to
correct unexpected results of their actions within do their task, paradoxically, they are unwilling to
the limited areas where they work. The biggest learn methods that would enable them to do their
differences between competent performer and task more effectively and efficiently. Once people
expert people are gaps in knowledge, forgetting know a process that works, they lose interest in
information, and commission of errors. Santhanam finding a more efficient way (Krug, 2000).
and Wiedenbeck (1993) found that the knowledge
and performance levels of competent performers Experts
on any specific system are very similar, although
their fields of expertise vary. For example, users of Experts bring a wider repertoire of methods and
a desktop publishing program have similar perfor- techniques for achieving goals than do people
mance although their actual uses of the program who know a single procedure to accomplish
(marketing department versus legal secretary) a task (Johnson, 1992). Experts have a com-
can be very different. This consistent knowledge prehensive and consistent mental model of the
level allows a design team to assume comparable system, allowing them to resolve ambiguities in
mental models of the various audience groups. information and to spot inconsistencies. Studies
Competent performers can perform more com- of differences between people operating at the
plex interaction tasks and are willing to learn more novice and expert level demonstrate that experts
efficient ways. Competent performers have some go beyond merely acquiring knowledge; they
task knowledge and will search for information are able to retrieve and apply knowledge within
about things that they have forgotten (Brinkman different contexts (Bransford & Schwartz, 1999).
et al., 2001). However, they are not interested in “Expert behavior requires an exquisite sensitivity
learning new features that are not directly relevant to context and an ability to know what to commit
to their goals. to” (Weigand, 2006, p. 45). Some of the character-
Grayling (2002) points out that once people istics of expert interaction include efficient time
reach some intermediate point (which can be any allocation, sophisticated problem representation
point within advanced beginner or competent and assessment, and a selection of appropriate
performer), they stop increasing their knowledge; strategies matching task difficulties (Davidson,
they are happy at that level, can accomplish their Deuser, & Sternberg, 1994).
goals, and don’t feel a need to know more. The

153
 How People Approach Information

Unlike the people at other levels, experts are plex, since it much more is at play than simple
highly motivated to learn more tasks and to help assumptions that experts know more. As people
others. Experts also tend to experiment with a move toward the expert level, their fundamental
system to learn more about it; their mental model approach to understanding a situation changes.
allows them to make accurate predictions of the An important aspect of audience and content
system response to these experiments. Competent analysis is determining where on the continuum
performers are happy with knowing just enough audience groups exist and how widely they are
to perform their job, but experts will try to learn spread along it. Many methods in audience and
the entire system even if many parts of it are not task analysis depend on people self-reporting
relevant to their day-to-day needs. their knowledge level. However, this can be
An expert has two types of knowledge: de- problematical. Competent performers users are
clarative and procedural. Declarative knowledge often confused about their own knowledge levels,
is composed of facts and their meanings for thinking they know more than they actually do.
specific contexts. Procedural knowledge consists Grayling (2002) noted that most self-identified
of the strategies and sequences of operations “experts” fared no better than novices in a set of
used in understanding a situation or solving a usability tests. The self-identified “experts” had
problem. Declarative knowledge makes up the sufficient but inefficient methods to support how
expert’s background knowledge (Chao, Salvendy, they normally worked, or they possessed a nar-
& Lightner, 1999). Many information systems row range of expertise and showed novice-level
provide declarative knowledge, and people feel skills once the usability tests moved them beyond
they have learned something. But without the their expertise.
procedural knowledge, they do not know how to An advanced beginner has a surface-level
apply declarative knowledge. Experts understand understanding of how to approach a situation.
both types of knowledge and can combine them An expert is able to perceive the deeper-level
in order to comprehend a situation. structure of the program and considers it from that
view. A very obvious difference in how advanced
Characteristics of People in the beginners and experts approach a situation is
Novice-Expert Continuum how quickly they arrive at a workable solution.
Although the experts arrive at a solution first, there
People do not fall exactly into a category of novice is an interesting difference, as well as a similarity,
or expert. Instead, they will show characteristics of in their approaches.
both, depending on the specific context. However,
their knowledge in one area may transfer to other • Difference. An advanced beginner tends
areas. Table 3 shows a summary of various char- to start solving the problem right away,
acteristics on the novice–expert continuum. Most immediately leaping in and trying to find
people would not be placed in either column, but a solution. An expert, on the other hand,
at some transitional area between them, indicated tends to take time to evaluate the situation
by the large arrow. before interacting with it.
• Similarity. Both use about the same depth
HII Approaches along the Novice- of search and thought processes evaluat-
Expert Continuum ing possible solutions. Research compar-
ing typical chess players to grandmasters
The HII differences between the novices/advanced found that both evaluate about the same
beginners and the experts can be highly com- number of possible moves and look ahead

154
How People Approach Information

Table 3. Characteristics of the novice-expert continuum. The middle area with the arrows combines the
characteristics of Hackos and Redish’s advanced beginner and competent performer as people’s experi-
ence moves them from novice to expert.

Novice characteristics Expert characteristics

Novices jump into the problem and immediately start working Experts spend more time examining a problem and structuring
toward a solution (Sutcliffe & Maiden, 1992). the problem space.
Novices do not try to reformulate the problem (Brand-Gurwel, Experts first try to reformulate the problem into a form they
Wopereis, & Vermetten, 2005).  understand.
Novices focus on the problem at the surface level and do not Experts represent, articulate, and solve problems at a deep
construct any large meaningful patterns (Glaser & Chi, 1988). level and use abstract concepts to describe them. They also
work to construct large-scale patterns within the problem.
Novices tend to describe and explain problems at a basic and Experts are able to use top-down organization in which they
superficial level and use bottom-up organization in which they approach a problem by identifying the final and intermediate
try to find solutions using step-by-step procedures (Anderson, goals before developing a solution plan.
1990; Vu et al., 2000).
Novices focus more attention on finding a solution to a prob- Experts spend more time examining, familiarizing, and trying
lem than understanding the problem. to understand the problem before trying to find a solution.
Novices view the entire problem as a single entity (Doane et Experts show more detailed and conceptual organization than
al., 1990).  novices and decompose problems into more subparts.
Novices rely on the first plan they generate (Jeffries, Turner, Experts tend to consider more alternative solutions to solving
Polson, & Atwood, 1981). problems.
Novices are more likely to request information which cor- Experts make fewer requests for information but they place it
responds to learning more about an overall concept. They deal in the context of the problem.
more with general knowledge than with information specific to
the situation (Kolodner, 1983; Mao & Benbasat, 2001).
Novice errors arise from erroneous concepts and ideas caused Experts make fewer errors, and the errors they do make tend to
by poor mental models. They also have a hard time seeing be errors in execution of a proper action, rather than perform-
that they made an error (Glaser & Chi, 1988; Fu, Salvendy, & ing the wrong action. Experts can rapidly access the situation
Turley, 2002; Vu et al., 2000). and realize an error occurred.
Novices have a poor mental model which may cause them Experts understand which information is relevant and seek it
to reject or ignore relevant information as irrelevant (Duin,
1989).
 out.

Novices consider a problem space as a linear structure. Experts break the problem space into hierarchical structures.

the same 2-3 moves (Chase & Simon, accounting problem—while the experts di-
1973). The difference is that the grand- vided them based on the type of algorithm
masters only evaluate good solutions while which would be used to solve them. From
typical players spend time evaluating poor the programming view, a sorting algorithm
solutions. orders items, whether they are plane arrival
times, widgets, or budget reports. Experts
Two examples which show how the approaches could see this and categorized the prob-
are different: lem based on the deeper structure, not on
the surface level of what is being done—
• New and experienced programmers were something new programmers could not do.
asked to categorize problems. The new • In a web study, when asked to draw a
programmers divided them based on the graphical presentation of a website, novice
external topic—an inventory problem, an web users created labels that were more

155
 How People Approach Information

Box 16.
descriptive and precise than an expert’s.
However, their labels only described the
Characterizing the Tower of Hanoi problem
surface-level issues of what the link con- Many psychology experiments on problem-solving methods
tained. The expert’s labels described the use variations of the Tower of Hanoi problem. Some studies
have presented people with variations of the same problem, but
generic stereotype of the various links and couched in different terms (Kotovsky, Hayes, & Simon, 1985).
what information should be contained be- Rather than realizing they are the same problem, most people
jump right in and try to address and solve each problem from
hind the link. scratch. An expert performer, on the other hand, examines
the deeper-level issues of the problem before trying to solve
Experts think differently about topics in their it, realizes that the problems are the same and uses the same
strategy to solve each one.
area of expertise and build knowledge into large,
meaningful patterns (Glaser & Chi, 1988). To
illustrate, as compared with novices, experts
have better memory for domain-relevant mate-
rial, such as chess positions (Chase & Simon,
1973) or bridge hands (Engle & Bukstel, 1978).
Experiments with chess players, and other related
studies, show that experts have a large mental
The Tower of Hanoi. The object is to move all three disks, one
collection of stereotypical solutions which they at a time, so that they are stacked on tower 3. You can only put
can apply. Experts can almost immediately sort a small disk onto a larger disk, never the other way. Hint: the
opening move is to put the smallest disk on tower 2.
through those solutions and select the best ones.
The advanced beginner has not developed these
solutions and thus has nothing to draw on. In sup-
port of this idea, Antony, Batra, and Santhanam beginner performance. People will sometimes
(2005) found that novice designers did better with need both feedback on a situation’s progress and
the help of a knowledge-based system that gave help interpreting that feedback: it depends on their
them design advice. The computer system essen- knowledge level.
tially was fulfilling the part of the expert’s learned
collection of solutions. Hayes and Akhavi (2008)
found similar results when providing computer- EXAMPLES
based support to student engineers engaged in a
complex design project. Astronomical Information
The comparison between the experts and the
novices indicates some interesting differences Issues of active versus passive approaches to
between the two groups. First, experts spend more information must be addressed by the design
time than novices do on the whole information team. Students and the general public will tend
problem-solving task. Also, experts can accurately to have a very passive approach: they will expect
monitor their performance and progress (Glaser the site to provide them with fully integrated
& Chi, 1988; Vu et al., 2000). Experts are able and synthesized information. They will also not
to mentally track progress toward a solution, and be willing to exert much effort to either reading
they know if the situation is progressing properly. or understanding the text. Poor page design will
Advanced beginners cannot do this. Competent make them quickly leave.
performers are able to do this in a limited area, Gaining an understanding of the astronomical
but if the situation moves outside of their narrow information requires the readers to comprehend
range of expertise, then they revert to advanced the information and to build relationships within

156
How People Approach Information

it. Otherwise, they risk learning a collection of • The information of interest/concern to peo-
disjoint facts (being able to quote the text, but ple differs depending on their management
not understanding it). One issue the design needs style.
to address is how people ignore information they • The researching and handling of problem
don’t understand. Content written at a too complex situations is different from checking that
or technical level will be skipped, which will everything is ok.
cause later information to not be understood. The
information relationships convert the text from a Work with a broad base of managers to deter-
mass of data about astronomy to an interrelated mine what they consider the major goals to be,
and understandable collection of information. A goals that must be better articulated than “ensure
table of the specifications for a star might make my department is running ok.” What is most im-
sense to the knowledgeable amateur, but that’s portant: budget, productivity, employee retention
because she has deep background knowledge to or satisfaction? There will be along collection of
call on to internally form the relationships. Most goals, with each goal only applying to a subset
people, however, lack that background knowl- of managers. However, as they will all be using
edge and would only see a table of meaningless the resulting system, it must support each person.
random numbers. For any situation, there can be a large number
Building information relationships is one place of possible goals. Yet an effective, experienced
were science writing departs from science journal- person quickly eliminates them, normally so
ism. In a newspaper article on astronomy, there quickly that Klein (1999) found they would say
is almost always a focus on the “why study this” they didn’t consider them. Define what makes
and “how does this affect me, as a person on the the goals the business analysts thought were
street,” both of which are not really relationships, important and relevant to the situation and what
but part of the journalistic requirements to build makes other goals irrelevant. Irrelevant goals
interest and draw the reader in. This information that at first glance can appear to be relevant or
should be contained within any astronomy infor- might be relevant to a related situation should be
mation system, but it should not be the focus. Most specifically addressed. What information provides
people with a strong interest in astronomy either clues to distinguishing and differentiating them?
already know this information or they consider Design teams, not readers, should be doing the
it secondary. Instead, they want to see how this work of distinguishing the relevant information.
piece of information connects to other pieces of The number of reports and amount of informa-
astronomical information. They want to build a tion which a business analyst must process often
more complete understanding of the topic being leads to satisficing. They will look at a section
researched. of a report and see the that the values are good
enough. Rather than digging deeper and uncover-
Business Reports ing potential problems which may become critical
in a few reporting periods, they accept them as-is
Think of these basic design problems when and move on to other information. Good design of
supporting report analysis (the list can easily be salient information and placing information into
extended): context can help the analyst judge if the values are
good enough or if they deserve more inspection.
• Each person works in a different way when A business report has a huge amount of infor-
analyzing reports. mation behind it. Without a proper information
needs analysis, the design team can create reports

157
 How People Approach Information

Table 4. Regional sales figurse

East West South North

Year over Year over Year over Year over Year over
Product Total year change Total year change Total year change Total year change Total year change
1/8” adapter 101 11% 34 3% 23 -5% 20 5% 24 8%
1/4” adapter 161 6% 36 7% 46 12% 63 -7% 16 -6%
1/2” adapter 119 31% 27 23% 35 -8% 34 12% 23 4%
3/4” adapter 168 -16% 68 -3% 27 -4% 27 -11% 46 2%
1” adapter 119 0% 34 -5% 23 4% 37 5% 25 -4%
Total 700 32% 199 25% 154 -1% 181 4% 134 4%

with improper salience. Many business analysts Approaching Information

have to extract information such as Table 4 to create
a view that fits their needs. The low level infor- Understanding what information is desired and
mation about each regions total sales dominates audience expectations are essential to meeting
the table. If that is what the required information, their information needs and communication goals.
then it is ok, but often it is higher summary data Information is useful only when it fits within the
that is needed. The last row has the totals, but contextual constraints of the current situation and
that row is lost in all of the other data. This table the reader can understand how it fits. How people
only shows one product category; if the analyst approach information can range from a very ac-
needs to compare multiple product categories, tive to a very passive approach to interpreting it.
then the amount of non-salient information which Some people question all the information and some
is receiving a high salience because of volume people prefer to avoid the information altogether.
increases even more. Design teams need to ensure that the information
Tables can be easy for design teams to construct source conforms to both the situational context
based on queries into the database. But design and the way people approach the information.
teams must ensure the information has been in-
tegrated to fit the reader’s needs and is not simply Effort
a data dump of easy-to-query or easy-to-display
information. When people engage in a task or read informa-
tion, they engage in an effort trade-off between
obtaining the desired information and expending
SUMMARY the least acceptable effort. People stop looking
for information when they feel the information
These chapters considered some of the factors they have adequately addresses a question, with
influencing the ways people approach and in- factors such as prior knowledge and desired level
terpret information. Information interaction and of knowledge influencing when they stop looking.
interpretation will vary dependent on people’s As long as they know some way of accomplishing
goals and information needs. These differing goals a task, they see no reason to expend effort find-
and information needs shape what information is ing a better way, assuming they even know that
actually deemed important, and how hard people a better way might exist. People use strategies
will work to understand it. based on a trade-off between time and the amount

158
How People Approach Information

of effort required to gain accurate and sufficient Information Quality

information and design teams need to account for
this trade-off early in the design. Information must meet a certain level of quality
before people will trust it and people must trust it
Sufficing before they will use it. People judge quality based
on their perception of its accuracy, completeness,
People stop looking for and interpreting informa- authority, usefulness, and accessibility.
tion when they are happy with the answer because
it fits their information needs or understanding of Age
the situation. They only expend enough effort to
obtain an answer that is “good enough,” rather than Age and its associated changes in cognitive abil-
working to obtain an optimal answer. Although the ity change how people interact with information,
selected answer is not always sufficient, people but HII-age interactions are much more complex
have a hard time judging sufficiency and know than simply being age-based. Older people do
how sub-optimal their solution is. not process information differently, but do show
a lower ability to cope with poor usability issues.
Multitasking Design teams must allow for age-related factors
such as poorer working memory and reasoning
Multiple tasks pull people in different directions ability, but must also avoid over-generalizing the
and they can rarely give full attention to interacting audience purely on age.
with information. Multitasking involves mentally
switching between tasks and their related mental Novice to Expert
models in constantly changing sequence. Although
it allows people to work on multiple things at Level of experience and background knowledge
once, they all suffer a performance decrease and change how people interact with information.
comprehension of information is seriously im- People interact with information in many different
pacted. Design commonly ignores multitasking roles, even within a single task, and they bring
and focuses on people accomplishing one task different levels of expertise to each role. All of
at a time. However, multitasking is a part of the these views affect how people define themselves
typical HII and design teams must allow for it. with respect to the novice–expert continuum and
the skill levels they exhibit. It can be easy for
Salience design teams to form stereotypes of the system
users as either novice or expert, in reality, people
Not all information carries the same level of remain at novice for a very short time and most
importance. The most important information people exhibit skill levels that are somewhere in
should have the highest salience. Any information the middle between novice and expert.
system has an overabundance of information that
can quickly lead to cognitive overload. However,
information salience is not “one-size-fits-all.” REFERENCES
The amount of salience it deserves depends on a
wide set of factors. Design teams need to resolve Albers, M. (2004a). Communication of complex
what information people need to understand a information: User goals and information needs for
situation and ensure that information receives dynamic Web information. Mahwah, NJ: Erlbaum.
proper salience.

159
 How People Approach Information

Albers, M. (2004b). Signal to noise ratio of in- Bechky, B. (2003). Sharing meaning across oc-
formation in documentation. Proceedings of 22nd cupational communities: The transformation of
Annual International Conference on Computer understanding on a production floor. Organi-
Documentation (pp. 41–44). New York, NY: ACM. zation Science, 14(3), 312–330. doi:10.1287/
orsc.14.3.312.15162
Albers, M. (2009). Information relationships: The
source of useful and usable content. Proceed- Becker, S. (2004). A study of web usability for
ings of 29th Annual International Conference on older adults seeking online health resources. ACM
Computer Documentation. (pp. 171–178). New Transactions on Computer-Human Interaction,
York, NY: ACM. 11(4), 387–406. doi:10.1145/1035575.1035578
Albers, M. (2010). Usability and information Becker, S. (2005). E-government usability for
relationships: Considering content relationships older adults. Communications of the ACM, 48(2),
when testing complex information. In Albers, 102–104. doi:10.1145/1042091.1042127
M., & Still, B. (Eds.), Usability of complex In-
Bedell, S., Agrawal, A., & Petersen, L. (2004).
formation Systems: Evaluation of user interac-
A systematic critique of diabetes on the World
tion (pp. 109–131). Boca Raton, FL: CRC Press.
Wide Web for patients and their physicians. In-
doi:10.1201/EBK1439828946-9
ternational Journal of Medical Informatics, 73,
Alvarado, C., Jeevan, J., Ackerman, M., & Karger, 687–694. doi:10.1016/j.ijmedinf.2004.04.011
D. (2003). Surviving the information explosion:
Belkin, N. (1980). Anomalous states of knowledge
How people find their electronic information (pp.
as a basic for information retrieval. The Canadian.
1–9). MIT AI Memo AIM-2003-006. Retrieved
Journal of Information Science, 5, 133–143.
from http://www.csail.mit.edu/~teevan/work/
publications/papers/aim03.pdf Benbasat, I., Dexter, A., & Todd, P. (1986a).
An experimental program investigating color-
Anderson, J. R. (1990). Cognitive psychology and
enhanced. and graphical information presentation:
its implications. New York, NY: Freeman.
An integration of findings. Communications of the
Andriole, S., & Adelman, L. (1995). Cognitive ACM, 29(11), 1094–1105. doi:10.1145/7538.7545
System Engineering for User–Computer Interface
Bernhardt, S. (2002). Technical writing for adults.
Design, Prototyping, and Evaluation. Mahwah,
ATTW discussion list Sept 26, 2002.
NJ: Erlbaum.
Bhavnani, S., & John, B. (2000). The strategic
Antony, S., Batra, D., & Santhanam, R. (2005). The
use of complex computer systems. Human-
use of a knowledge-based system in conceptual
Computer Interaction, 15, 107–137. doi:10.1207/
data modeling. Decision Support Systems, 41,
S15327051HCI1523_3
176–188. doi:10.1016/j.dss.2004.05.011
Birdi, K., & Zapf, D. (1997). Age differences
Arning, K., & Ziefle, M. (2007). Understanding
in reactions to errors in computer-based work.
age differences in PDA acceptance and perfor-
Behaviour & Information Technology, 16(6),
mance. Computers in Human Behavior, 23,
309–319. doi:10.1080/014492997119716
2904–2927. doi:10.1016/j.chb.2006.06.005
Baddeley, A. (1986). Working memory. Oxford,
England: Oxford UP.

160
How People Approach Information

Brand-Gurwel, S., Wopereis, I., & Vermetten, Y. Cerella, J. (1985). Information processing rates
(2005). Information problem solving by experts in the elderly. Psychological Bulletin, 98, 67–83.
and novices: Analysis of a complex cognitive skill. doi:10.1037/0033-2909.98.1.67
Computers in Human Behavior, 21, 487–508.
Chadwick-Dias, A., McNulty, M., & Tullis,
doi:10.1016/j.chb.2004.10.005
T. (2003). Web usability and age: How design
Bransford, J. D., & Schwartz, D. L. (1999). Re- changes can improve performance. Proceedings
thinking transfer: A simple proposal with multiple of the 2003 Conference on Universal usability,
implications. In Iran-Nejad, A., & Pearson, P. D. (pp. 30–37). New York, NY: ACM
(Eds.), Review of research in education (Vol. 24,
Chadwick-Dias, A., McNulty, M., & Tullis,
pp. 61–101). Washington, DC: American Educa-
T. (2003). Web usability and age: How design
tional Research Association. doi:10.2307/1167267
changes can improve performance. Proceedings
Brinkman, W., Buil, V., Cullen, R., Gobits, R., of the 2003 Conference on Universal usability,
& Van Nes, F. (2001). Design and evaluation of (pp. 30–37). New York, NY: ACM.
online multimedia maintenance manuals. Behav-
Chao, C., Salvendy, G., & Lightner, N. (1999). De-
iour & Information Technology, 20(1), 47–52.
velopment of a methodology for optimizing elicited
doi:10.1080/01449290010020639
knowledge. Behaviour & Information Technology,
Burgess, P. W., Veitch, E., De Lacy Costello, A., 18(6), 413–430. doi:10.1080/014492999118841
& Shallice, T. (2000). The cognitive and neuro-
Chaparro, A., Bohan, M., Fernandez, J., Choi,
anatomical correlates of human multitasking.
S., & Kattel, B. (1999). The impact of age on
Neuropsychologia, 38, 848–863. doi:10.1016/
computer input device use: Psychophysical and
S0028-3932(99)00134-7
physiological measures. International Journal of
Carroll, J., & Rosson, M. (1987). The paradox Industrial Ergonomics, 24, 503–513. doi:10.1016/
of the active user. In Carroll, J. (Ed.), Interfacing S0169-8141(98)00077-8
thought: Cognitive aspects of human-computer
Chase, W., & Simon, H. (1973). The mind’s eye
interaction (pp. 80–111). Cambridge, MA: MIT
in chess. In Chase, W. (Ed.), Visual information
Press.
processing (pp. 215–281). New York, NY: Aca-
Cary, M., & Carlson, R. (2001). Distributing work- demic Press.
ing memory resources in problem solving. Journal
Cuevas, H., Fiore, S., Bowers, C., & Salas, E.
of Experimental Psychology. Learning, Memory,
(2004). Fostering constructive cognitive and
and Cognition, 27, 836–848. doi:10.1037/0278-
metacognitive activity in computer-based com-
7393.27.3.836
plex task training environments. Computers in
Casaday, G. (1991). Balance. In Karat, J. (Ed.), Human Behavior, 20, 225–241. doi:10.1016/j.
Taking software design seriously: Practical Tech- chb.2003.10.016
niques for HCI design (pp. 45–62). New York,
Cypher, A. (1986). The structure of users’ activi-
NY: Academic Press.
ties. In Norman, D., & Draper, S. (Eds.), User
Casner, S. (1994). Understanding the determi- centered system design: New perspectives on
nants of problem-solving behavior in a complex human-computer interaction (pp. 243–264).
environment. Human Factors, 34(4), 580–596. Mahwah, NJ: Erlbaum.

161
 How People Approach Information

Cyr, D., Head, M., & Larios, H. (2010). Colour Emerson, M., & Miyake, A. (2003). The role of
appeal in website design within and across cul- inner speech in task switching: A dual-task inves-
tures: A multi-method evaluation. International tigation. Journal of Memory and Language, 48,
Journal of Human-Computer Studies, 68(1-2), 148–168. doi:10.1016/S0749-596X(02)00511-9
1–21. doi:10.1016/j.ijhcs.2009.08.005
Endsley, M. (1995). Toward a theory of situation
Czerwinski, M., Horvitz, E., & Wilhite, S. (2004). awareness in dynamic systems. Human Factors,
A diary study of task switching and interrup- 37(1), 32–64. doi:10.1518/001872095779049543
tions. Retrieved December 10, 2008, from http://
Engle, R., & Bukstel, L. (1978). Memory processes
research.microsoft.com/users/ marycz/chi2004d-
among bridge players of differing expertise. The
iarystudyfinal.pdf
American Journal of Psychology, 91, 673–689.
Davidson, J., Deuser, R., & Sternberg, R. (1994). doi:10.2307/1421515
The role of metacognition in problem solving. In
Fennema, M., & Kleinmuntz, D. (1995). Antici-
Metcalfe, J., & Shimamura, A. (Eds.), Metacog-
pation of effort and accuracy in multi-attribute
nition: Knowing about knowing (pp. 207–226).
choice. Organizational Behavior and Human
Cambridge, MA: MIT Press.
Decision Processes, 63(1), 21–32. doi:10.1006/
Doane, S., Pellegrino, J., & Klatzky, R. (1990). obhd.1995.1058
Expertise in a computer operating system:
Freudenthal, D. (2001). The role of age, foreknowl-
Conceptualization and performance. Human-
edge and complexity in learning to operate a com-
Computer Interaction, 5, 267–304. doi:10.1207/
plex device. Behaviour & Information Technology,
s15327051hci0502&3_5
20(1), 23–35. doi:10.1080/01449290010020666
Duggan, G., & Payne, S. (2001). Interleaving
Fu, L., Salvendy, G., & Turley, L. (2002). Effec-
reading and acting while following procedural
tiveness of user testing and heuristic evaluation as
instructions. Journal of Experimental Psychol-
a function of performance classification. Behav-
ogy. Applied, 7(4), 297–307. doi:10.1037/1076-
iour & Information Technology, 21(2), 137–143.
898X.7.4.297
doi:10.1080/02699050110113688
Duin, A. (1989). Factors that influence how read-
Fu, W., & Gray, W. (2004). Resolving the para-
ers learn from test: Guidelines for structuring
dox of the active user: Stable suboptimal perfor-
technical documents. Technical Communication,
mance in interactive tasks. Cognitive Science,
36(2), 97–101.
28, 901–935. doi:10.1207/s15516709cog2806_2
Dye, K. (1988). When is document accurate and
Ganzach, Y., & Schul, Y. (1995). The influence
complete. Professional Communication Confer-
of quantity of information and goal framing
ence, 1988 (pp. 269–272). Seattle, WA. October
on decisions. Acta Psychologica, 89, 23–36.
5–7, 1988.
doi:10.1016/0001-6918(94)00004-Z
Einstein, G., McDaniel, M., Owen, P., & Cote,
Glaser, R., & Chi, M. (1988). Overview. In Chi,
N. (1990). Encoding and recall of texts: The
M. T. H., Glaser, R., & Farr, M. J. (Eds.), The
importance of material appropriate processing.
nature of expertise (pp. xv–xxviii). Hillsdale,
Journal of Memory and Language, 29, 566–581.
NJ: Erlbaum.
doi:10.1016/0749-596X(90)90052-2

162
How People Approach Information

Gloria Mark, G., Gonzalez, V., & Harris, J. (2005). Janiszewski, C. (1998). The influence of display
No task left behind? Examining the nature of characteristics on visual exploratory search be-
fragmented work. Proceedings of the SIGCHI havior. The Journal of Consumer Research, 25,
Conference on Human Factors in Computing 290–301. doi:10.1086/209540
Systems, April 2–7, 2005. Portland, Oregon.
Jeffries, R., Turner, A., Polson, P., & Atwood,
Gorry, G. (2005). As simple as possible, but not M. (1981). The processes involved in designing
simpler. Communications of the ACM, 48(9), software. Hillsdale, NJ: Erlbaum.
119–122. doi:10.1145/1081992.1082026
Jenkins, V., Fallowfield, L., & Saul, J. (2001).
Gray, W., & Fu, W. (2004). Soft constraints in Information needs of patients with cancer: Results
interactive behavior: The case of ignoring perfect from a large study of UK cancer centres. Brit-
knowledge in-the-world for imperfect knowledge ish Journal of Cancer, 84, 48–51. doi:10.1054/
in-the-head. Cognitive Science, 28, 359–382. bjoc.2000.1573
Gray, W., Schoelles, M., & Sims, C. (2005). Johnson, E., Payne, J., & Bettman, J. (1988). Infor-
Adapting to the task environment: Explorations mation displays and preference reversals. Organi-
in expected value. Cognitive Systems Research, zational Behavior and Human Decision Processes,
6, 27–40. doi:10.1016/j.cogsys.2004.09.004 42, 1–21. doi:10.1016/0749-5978(88)90017-9
Grayling, T. (2002). If we build it, will they come? Johnson, J., Andrews, J., & Allard, S. (2001). A
A usability test of two browser-based embedded model for understanding and affecting genetics
help systems. Technical Communication, 49(2), information seeking. Library & Information
193–209. Science Research, 23(4), 335–349. doi:10.1016/
S0740-8188(01)00094-9
Hackos, J., & Redish, J. (1998). User and task
analysis for interface design. New York, NY: Johnson, P. (1992). Human computer interaction:
Wiley. Psychology, task analysis and software engineer-
ing. London, UK: McGraw-Hill.
Hallgren, C. (1997). Using a problem focus to
quickly aid users in trouble. Proceedings of the Jung Lee, Y., Park, J., & Widdows, R. (2009). Ex-
1997 STC Annual Conference. Washington, DC: ploring antecedents of consumer satisfaction and
STC. repeated search behavior on e-health information.
Journal of Health Communication, 14, 160–173.
Hayes, C., & Akhavi, F. (2008). Creating effec-
doi:10.1080/10810730802659830
tive decision aids for complex tasks. Journal of
Usability Studies, 3(4), 152–172. Kang, N., & Yoon, W. (2008). Age- and expe-
rience-related user behavior differences in the
Herbig, P., & Kramer, H. (1992). The phenomenon
use of complicated electronic devices. Interna-
of innovation overload. Technology in Society, 14,
tional Journal of Human-Computer Studies, 66,
441–461. doi:10.1016/0160-791X(92)90038-C
425–437. doi:10.1016/j.ijhcs.2007.12.003
Hsee, C. (1995). Elastic justification: How
tempting but task irrelevant factors influence
decisions. Organizational Behavior and Human
Decision Processes, 62(3), 330–337. doi:10.1006/
obhd.1995.1054

163
 How People Approach Information

Kieras, D., Meyer, D., Ballas, J., & Lauber, E. Kwan, M., & Balasubramanian, P. (2003).
(2000). Modern computational perspectives on KnowledgeScope: Managing knowledge in con-
executive mental processes and cognitive control: text. Decision Support Services, 35, 467–486.
Where to from here? In Monsell, S., & Driver, J. doi:10.1016/S0167-9236(02)00126-4
(Eds.), Control of cognitive processes: Attention
Langholtz, H., Gettys, C., & Foote, B. (1995).
and performance XVIII (pp. 681–712). Cambridge,
Are resource fluctuations anticipated. in resource
MA: MIT Press.
allocation tasks? Organizational Behavior and
Kieras, D., & Polson, P. (1985). An approach to the Human Decision Processes, 64(3), 274–282.
formal analysis of user complexity. International doi:10.1006/obhd.1995.1105
Journal of Man-Machine Studies, 22, 365–394.
Lefevre, J., & Dixon, P. (1986). Do written instruc-
doi:10.1016/S0020-7373(85)80045-6
tions need examples? Cognition and Instruction,
Klein, G. (1988). Do decision biases explain too 3(1), 1–30. doi:10.1207/s1532690xci0301_1
much? Human Factors Society Bulletin, 32(5),
Lowe, R. (2002). Animation and learning: Selec-
1–3.
tive processing of information in dynamic graph-
Klein, G. (1993). A recognition-primed decision ics. Learning and Instruction, 13(2), 157–176.
(RPD) model of rapid decision making. In Klein, doi:10.1016/S0959-4752(02)00018-X
G., Orasanu, J., Calderwood, R., & Zsambok, C.
Mao, J., & Benbasat, I. (2001). The effects of
(Eds.), Decision making in action: Models and
contextualized access to knowledge on judgment.
methods (pp. 138–147). Norwood, NJ: Ablex.
International Journal of Human-Computer Stud-
Klein, G. (1999). Sources of power: How people ies, 55, 787–814. doi:10.1006/ijhc.2001.0507
make decisions. Cambridge, MA: MIT.
Marquie, J., Jourdan-Boddaert, L., & Huet,
Kolodner, J. (1983). Towards an understanding N. (2002). Do older adults underestimate
of the role of experience in the evolution from their actual computer knowledge? Behaviour
novice to expert. International Journal of Man- & Information Technology, 21(4), 273–280.
Machine Studies, 19, 497–518. doi:10.1016/ doi:10.1080/0144929021000020998
S0020-7373(83)80068-6
Mayer, R. (1976). Comprehension as affected by
Kotovsky, K., Hayes, J., & Simon, H. (1985). Why structure of problem representation. Memory &
are some problems hard? Evidence from Tower Cognition, 4, 249–265. doi:10.3758/BF03213171
of Hanoi. Cognitive Psychology, 17, 248–294.
McFarlane, D., & Latorella, K. (2002). The
doi:10.1016/0010-0285(85)90009-X
scope and importance of human interruption in
Krug, S. (2000). Don’t make me think! A common human-computer interaction design. Human-
sense approach to Web usability. Berkeley, CA: Computer Interaction, 17, 1–61. doi:10.1207/
New Riders. S15327051HCI1701_1
Kushleyeva, Y., Salvucci, D., & Lee, F. (2005). McGovern, G. (2006a). Making the customer
Deciding when to switch tasks in time-critical CEO. Retrieved from http://www.gerrymcgovern.
multitasking. Cognitive Systems Research, 6, com/nt/ 2006/nt-2006-06-12-customer-ceo.htm
41–49. doi:10.1016/j.cogsys.2004.09.005

164
How People Approach Information

McKenzie, P. (2002). Communication barriers and Navarro-Prieto, R., Scaife, M., & Rogers, Y.
information-seeking counterstrategies in accounts (1999). Cognitive strategies in Web searching.
of practitioner-patient encounters. Library and Proceeding of the fifth Conference on Human Fac-
Information Science, 24, 31–47. doi:10.1016/ tors and the Web, (pp. 43–56). Gaithersburg, MD.
S0740-8188(01)00103-7
Nielsen, J. (2001). The 3Cs of critical Web use:
McNamara, D., & Kintsch, W. (1996). Learning Collect, compare, choose. Retrieved December
from text: Effects of prior knowledge and text 3, 2008, from http://www.useit.com/ alert-
coherence. Discourse Processes, 22, 247–287. box/20010415.html
doi:10.1080/01638539609544975
Nielsen, J. (2007b). Banner blindness: Old and
Mirel, B. (1988). Cognitive processing, text new findings. Retrieved November 11, 2007, from
linguistics and documentation writing. Journal http://www.useit.com/alertbox/ banner-blindness.
of Technical Writing and Communication, 18(2), html
111–133. doi:10.2190/20JV-5N1E-6LNR-443U
Nilsson, R., & Mayer, R. (2002). The effects of
Mirel, B. (1992). Analyzing audiences for software graphic organizers giving cues to the structure of
manuals: A survey of instructional needs for real a hypertext document on users’ navigation strate-
world tasks. Technical Communication Quarterly, gies and performance. International Journal of
1(1), 15–35. doi:10.1080/10572259209359489 Human-Computer Studies, 57, 1–26. doi:10.1006/
ijhc.2002.1011
Mirel, B. (2003). Interaction design for complex
problem solving: Developing useful and usable Norman, D. (2004). Emotional design. New York,
software. San Francisco, CA: Morgan Kaufmann. NY: Basic Books.
Mirel, B., Feinberg, S., & Allmendinger, L. (1991). Norman, D. A. (2002). The design of everyday
Designing manuals for active learning styles. things. New York, NY: Basic Books.
Technical Communication, 38(1), 75–87.
O’Hara, K., & Payne, S. (1998). The effects of
Monsell, S. (1993). Task switching. Trends in operator implementation cost on planfulness of
Cognitive Sciences, 7(3), 134–140. doi:10.1016/ problem solving and learning. Cognitive Psy-
S1364-6613(03)00028-7 chology, 35, 34–70. doi:10.1006/cogp.1997.0676
Müller, M. (2007). Being aware: Where we think Orasanu, J., & Connolly, T. (1993). The reinven-
the action is. Cognition Technology and Work, tion of decision making. In Klein, G., Orasanu, J.,
9(2), 109–126. doi:10.1007/s10111-006-0047-7 Calderwood, R., & Zsambok, C. (Eds.), Decision
making in action: Models and methods (pp. 3–20).
Nanard, J., & Nanard, M. (1995). Hypertext
Norwood, NJ: Ablex.
design environments and the hypertext design
process. Communications of the ACM, 38(8), Paiva, A. (Ed.). (2000). Affective interactions:
49–56. doi:10.1145/208344.208347 Toward a new generation of computer interfaces.
New York, NY: Springer.
National Institute of Health. (2002). Making
your web site senior friendly. Retrieved January Payne, J., Bettman, J., & Johnson, E. (1988). Adap-
13, 2010, from http://www.nlm.nih.gov/pubs/ tive strategy selection in decision making. Journal
staffpubs/ od/ocpl/agingchecklist.html of Experimental Psychology. Learning, Memory,
and Cognition, 14, 534–552. doi:10.1037/0278-
7393.14.3.534

165
 How People Approach Information

Payne, S. (1991). Display-based action at the Robins, D., Holmes, J., & Stansbury, M. (2010).
user interface. International Journal of Man- Consumer health information on the Web: The
Machine Studies, 35, 275–289. doi:10.1016/ relationship of visual design and perceptions of
S0020-7373(05)80129-4 credibility. Journal of the American Society for In-
formation Science and Technology, 61(1), 13–29.
Payne, S. (2002). Balancing information needs:
Dilemmas in producing patient information leaf- Rosenbaum, S., & Walters, D. (1986). Audience
lets. Health Informatics Journal, 8, 174–179. diversity: A major challenge in computer docu-
doi:10.1177/146045820200800402 mentation. IEEE Transactions on Professional
Communication, 29(4).
Payne, S., Howes, A., & Reader, W. (2001).
Adaptively distributing cognition: A decision Rosson, M. (1983). Patterns of experience in text
making perspective on human-computer interac- editing. Proceedings of the CHI’83 Human Fac-
tion. Behaviour & Information Technology, 20(5), tors in Computing Systems (pp. 171–175). New
339–346. doi:10.1080/01449290110078680 York, NY: ACM.
Penrose, J., & Seiford, L. (1988). Microcomputer Rouet, J. (2003). What was I looking for? The
users’ preferences for software documentation: influence of task specificity and prior knowledge
An analysis. Journal of Technical Writing and on students’ search strategies in hypertext. Inter-
Communication, 18(4), 355–366. acting with Computers, 15, 409–428. doi:10.1016/
S0953-5438(02)00064-4
Pomerantz, J., & Schwaitzberg, S. (1975). Group-
ing by proximity: Selective attention measures. Rubens, P., & Rubens, B. (1988). Usability and
Perception & Psychophysics, 18, 355–361. format design. In Doheny-Farina, S. (Ed.), Effec-
doi:10.3758/BF03211212 tive documentation: What we have learned from
research (pp. 213–234). Cambridge, MA: MIT.
Rains, S., & Karmikel, C. (2009). Health infor-
mation-seeking and perceptions of website cred- Rubinstein, J., Meyer, D., & Evans, J. (2001).
ibility: Examining Web-use orientation, message Executive control of cognitive processes in task
characteristics, and structural features of websites. switching. Journal of Experimental Psychology.
Computers in Human Behavior, 25, 544–553. Human Perception and Performance, 27(4),
doi:10.1016/j.chb.2008.11.005 763–797. doi:10.1037/0096-1523.27.4.763
Raskin, J. (2003/2004). Silicon superstitions. Russell, M. (2005). Using eye-tracking data to
ACM Queue; Tomorrow’s Computing Today, 1(9). understand first impressions of a website. Usability
News, 7(1). Retrieved from http://psychology.
Redish, J. (2007). Expanding usability testing to
wichita.edu/surl/ usabilitynews/71/eye_tracking.
evaluate complex systems. Journal of Usability
html
Studies, 2(3), 102–111.
Salthouse, T., & Babcock, R. (1991). Decom-
Redish, J., & Schell, D. (1989). Writing and
posing adult age differences in working mem-
testing instructions for usability. In Fearing, B.,
ory. Developmental Psychology, 72, 763–776.
& Sparrow, K. (Eds.), Technical writing: Theory
doi:10.1037/0012-1649.27.5.763
and practice (pp. 63–71). New York, NY: Modern
Language Association of America.

166
How People Approach Information

Salvador, T., Barile, S., & Sherry, J. (2004). Ubiq- Slovic, P. (1972). From Shakespeare to Simon:
uitous computing design principles: Supporting Speculations—and some evidence—about man’s
human-human and human-computer transaction. ability to process information. Oregon Research
Proceedings of the SIGCHI Conference on Human Institute Research Bulletin, 12, 1–29.
Factors in Computing Systems (pp. 1497–1500).
Spool, J. (1997). Why on-site searching stinks.
New York, NY: ACM.
Retrieved December 28, 2005, from http://www.
Santhanam, R., & Wiedenbeck, S. (1993). Nei- uie.com/articles/search_stinks/
ther novice nor expert: The discretionary user
Stanford, J., Tauber, E., Fogg, B., & Marable, L.
of software. International Journal of Man-
(2002). Experts vs. online consumers: A compara-
Machine Studies, 38, 201–229. doi:10.1006/
tive credibility study of health and finance Web
imms.1993.1010
sites. Consumer WebWatch Research Report.
Sayago, S., & Blat, J. (2010). Telling the story of Retrieved from http://www.consumerwebwatch.
older people e-mailing: An ethnographical study. org/dynamic/ web-credibility-reports-experts-vs-
International Journal of Human-Computer Stud- online-abstract.cfm
ies, 68, 105–120. doi:10.1016/j.ijhcs.2009.10.004
Storch, N. (1992). Does the user interface make
Scanlon, T. (1996). Making online information us- interruptions disruptive? A study of interface style
able. Retrieved from http://www.uie.com/articles/ and form of interruption. Proceedings of the SIG-
online_information/ CHI Conference on Human Factors in Computing
Systems, (pp. 14–14). New York, NY: ACM.
Schneider, W., & Fisk, A. (1982). Concurrent
automatic and controlled visual search: Can pro- Stvilia, B., Mon, L., & Yi, Y. (2009). A model
cessing occur without resource cost? Journal of for online consumer health information quality.
Experimental Psychology. Learning, Memory, Journal of the American Society for Informa-
and Cognition, 8, 261–278. doi:10.1037/0278- tion Science and Technology, 60(9), 1781–1791.
7393.8.4.261 doi:10.1002/asi.21115
Schwartz, D., & Teleni, D. (2000). Tying knowl- Sutcliffe, A., & Maiden, N. (1992). Analyzing
edge to action with kMail. IEEE Intelligent the novice analyst: Cognitive models in software
Systems and Their Applications, 15, 33–39. engineering. International Journal of Man-Ma-
doi:10.1109/5254.846283 chine Studies, 36, 719–740. doi:10.1016/0020-
7373(92)90038-M
Sheridan, S., Felix, K., Pignone, M., & Lewis,
C. (2004). Information needs of men regarding Szulanski, G. (1996). Exploring internal sticki-
prostate cancer screening and the effect of a brief ness: Impediments to the transfer of best practice
decision aid. Patient Education and Counseling, within the firm. Strategic Management Journal,
54, 345–351. doi:10.1016/j.pec.2003.12.003 17, 27–43.
Shneiderman, B. (1997). Designing the user in- Thuring, M., Hannemann, J., & Haake, J. (1995).
terface: Strategies for effective human-computer Hypermedia and cognition: Designing for com-
interaction. Reading, MA: Addison-Wesley. prehension. Communications of the ACM, 38,
57–66. doi:10.1145/208344.208348
Simon, H. (1979). Models of thought. New Haven,
CT: Yale UP. Tufte, E. (1983). The visual display of quantita-
tive information. Cheshire, CT: Graphics Press.

167
 How People Approach Information

Tyre, M. J., & von Hippel, E. (1997). The situ- Webster, D., & Kruglanski, A. (1994). Individual
ated nature of adaptive learning in organizations. differences in need. for cognitive closure. Jour-
Organization Science, 8(1), 71–83. doi:10.1287/ nal of Personality and Social Psychology, 67(6),
orsc.8.1.71 1049–1672. doi:10.1037/0022-3514.67.6.1049
Valero, P., & Sanmartiân, J. (1999). Methods for Weigand, H. (2006). Two decades of the language-
defining user groups and user adjusted information action perspective. Communications of the ACM,
structures. Behaviour & Information Technology, 49(5), 45–26.
18(4), 245–259. doi:10.1080/014492999119002
Wickens, C. (1992). Engineering psychology and
van der Meij, H., & Gellevij, M. (2002). Effects human performance. New York, NY: HarperCol-
of pictures, age, and experience on learning to use lins.
a computer program. Technical Communication,
Wickens, C., & Carswell, C. (1995). The
49(3), 330–339.
proximity compatibility principle: Its psy-
van Hees, M. (1996). User instructions for the chological foundations and its relevance to
elderly: What the literature tells us. Journal of display design. Human Factors, 37, 473–494.
Technical Writing and Communication, 26(4), doi:10.1518/001872095779049408
521–536.
Wickens, C., & Hollands, J. (2000). Engineer-
Visser, W., & Morals, A. (1991). Concurrent use ing psychology and human performance. Upper
of different expertise elicitation methods applied Saddle River, NJ: Prentice Hall.
to the study of the programming activity. In Ack-
Woods, D. (1984). Visual momentum: A concept
ermann, D., & Tauber, M. J. (Eds.), Mental models
to improve the cognitive coupling of person
and human-computer interactions (pp. 97–114).
and computer. International Journal of Man-
Amsterdam, The Netherlands: North Holland.
Machine Studies, 21, 229–244. doi:10.1016/
Vu, K., Hanley, G., Strybel, T., & Proctor, R. S0020-7373(84)80043-7
(2000). Metacognitive processes in human-com-
Woods, D., Patterson, E., & Roth, E. (2002). Can
puter interaction: Self-assessments of knowledge
we ever escape from data overload? A cognitive
as predictors of computer expertise. International
systems diagnosis. Cognition Technology and
Journal of Human-Computer Interaction, 12(1),
Work, 4, 22–36. doi:10.1007/s101110200002
43–71. doi:10.1207/S15327590IJHC1201_2
Zeng, Q., Kogan, S., Plovnick, R., Crowell, J., La-
Walker, N., Millians, J., & Worden, A. (1996).
croix, E., & Greenes, R. (2004). Positive attitudes
Mouse accelerations and performance of older
and failed queries: An exploration of the conun-
computer users. Proceedings of the Human Fac-
drums of consumer health information retrieval.
tors and Ergonomics Society 40th Annual Meeting,
International Journal of Medical Informatics,
(pp.151–154). Santa Monica, CA: HFES.
73(1), 45–55. doi:10.1016/j.ijmedinf.2003.12.015
Wallsten, T., & Barton, C. (1982). Processing
Zhang, Y., Goonetilleke, R., Plocher, T., & Liang,
probabilistic multidimensional information for
S. (2005). Time-related behaviour in multitask-
decisions. Journal of Experimental Psychology.
ing situations. International Journal of Human-
Learning, Memory, and Cognition, 8, 361–384.
Computer Studies, 62, 425–455. doi:10.1016/j.
doi:10.1037/0278-7393.8.5.361
ijhcs.2005.01.002

168
How People Approach Information

Zimmermann, G., & Vanderheiden, G. (2005). Krug, S. (2000). Don’t make me think! A common
Creating accessible applications with RUP. sense approach to Web usability. Berkeley, CA:
Retrieved April 28, 2006, from http://www.128. New Riders.
ibm.com/developerworks/rational/ library/jul05/
Mirel, B. (2003). Interaction design for complex
zimmerman/index.html#notes
problem solving: Developing useful and usable
software. San Francisco, CA: Morgan Kaufmann.
Norman, D. (2004). Emotional design. New York,
ADDITIONAL READING
NY: Basic Books.
Albers, M. (2004). Communication of complex Redish, J. (2007). Expanding usability testing to
information: User goals and information needs for evaluate complex systems. Journal of Usability
dynamic Web information. Mahwah, NJ: Erlbaum. Studies, 2(3), 102–111.
Hackos, J., & Redish, J. (1998). User and task Wickens, C., & Hollands, J. (2000). Engineer-
analysis for interface design. New York, NY: ing psychology and human performance. Upper
Wiley. Saddle River, NJ: Prentice Hall.

169