See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/289323817

The impact of combining gestalt theories with interface design guidelines in

designing user interfaces

Article · January 2012

CITATION READS

1 534

2 authors:

Juan Gomez Lorne Olfman

Autonomous University of Aguascalientes Claremont Graduate University
43 PUBLICATIONS 81 CITATIONS 203 PUBLICATIONS 4,851 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Juan Gomez on 04 February 2021.

The user has requested enhancement of the downloaded file.

Gómez Reynoso et al. Measuring the Impact of Gestalt Theories in GUI

The Impact of Combining Gestalt Theories with Interface

Design Guidelines in Designing User Interfaces
Juan Manuel Gómez Reynoso Lorne Olfman
Universidad Autónoma de Aguascalientes Claremont GraduateUniversity
[email protected] [email protected]

ABSTRACT
Software systems are one of the most important technologies that are present in every task that humans and computers
perform. Humans perform their tasks by using a computer interface. Graphical user interfaces are the most common
interface that developers rely on to create easy-to-use, easy-to-learn and easy-to-understand software systems so that end
users can improve their performance. However, many times, developers tend to create software using their own
preferences based on their skills and abilities but do not consult theories that would produce better outcomes. We
conducted a study to identity whether software that is developed by using Gestalt theories combined with interface
development guidelines produces better outcomes compared to software developed using developers’ current skills.
Results show that for the present research, participants perceived the system that was developed using such approach had
superior quality compared to another that does not. However, results should be taken cautiously.
Keywords
Graphical user interface, Gestalt Theory, Quality, Information systems
INTRODUCTION
Software is one of the most used technologies. Everyday different applications are created for many purposes such as
management, information processing and information sharing, among others. Many software applications are intended
for the general population, so that end users are unlikely to fully exploit all software capabilities. End users interact with
software through interfaces. Wu (2000) believes that a good interface should be: a) easy for novices to learn, b) efficient
for experts to use, and c) provide the means for users to make the transition from the easy-to-learn but inefficient
methods of novices to the more difficult-to-learn and efficient methods of experts. Lane et al. (2005) argue that graphical
user interfaces (GUIs) meet these criteria fairly well. However, this is only true when GUIs are well-constructed by
taking into account end users’ needs.

Frequently, software developers create applications based solely on their own preferences rather than by having in mind
information regarding end user. In addition, Aberg and Chang (2005) state that “different industries require specialized
functions from one another; thereby it is difficult to apply any specific set of interface design guidelines” (p. 23).
Therefore, it is not an easy task to develop GUIs that meets all people’s expectations.

We believe it is important that developers should be capable of designing improved GUIs that would be more likely to be
accepted by end users. Traditional systems design methodologies do not provide “grounded” guidelines for building
GUIs. That is, developers should rely upon past theories and research, such as those based in Gestalt Theory (Chang et
al., 2002) on the subject to create superior information systems (IS). This study intends to provide some insights into how
existing research can be used for the purpose of enhancing software system interfaces. The research question is as
follows: Is there is a significant difference between a system’s GUI developed using a traditional approach compared to
a system’s GUI developed using design guidelines grounded in Gestalt Theory?

MOTIVATION

Software engineers should deliver high-quality software to users, which complies with all requirements, and, ideally
should be easy-to-use, easy-to-learn, and easy-to-understand (Pressman 2007). Software development is an increasingly
complex task (Bowen and Reeves 2008) that requires the effort of a set of experts. Frequently, users report that IS’
interfaces are not easy-to-use. For example, Ko et al. (2006) report that 26% of users are not satisfied with ISs’ GUIs.
Ahonen and Juntilla (2003) report that usually there is not enough budget or time allocated for the development of GUIs,
which could lead to developers being unable to put enough effort into the project.

Since software is an intangible product, its quality depends on the perspective of each user. Pressman (2007) argues that
users put a high software quality value on the interface, which is the mechanism that allows interaction between people
and the IS. Interaction means the coordination of information exchange between the user and the IS (Bosch and Juristo
2003). It is important to remember that the most visible component of any system is the interface. An ill-designed

Proceedings of the Eighteenth Americas Conference on Information Systems, Seattle, Washington, August 9-12, 2012. 1
Gómez Reynoso et al. Measuring the Impact of Gestalt Theories in GUI

interface prevents users from maximizing performance as well as being unable to use all of an IS’s features. Hence, user
interface (UI) design is a critical step for IS developers.

For interface design, the balance and general organization of the graphical elements can have an important impact on end
users’ perception of an IS (Buitrón de la Torre 2004; Liu and Osvalder 2004). In the field of psychology, Gestalt is a
formal theory for visual perception of objects (Gordon 2004), which explains how humans perceive graphical elements
that are “captured” through human senses. Past research (Chang et al. 2002; Chang and Nesbitt 2005) argues that this
theory is formed by a series of principles or laws that can be used to suggest how visual elements can be represented so
that effective results can be achieved. They identify eleven principles that can be used as the basis of an interface design,
which has a structure that is easy-to-understand and learn. This effect would contribute to the development of high
quality UI.

LITERATURE REVIEW

Software quality (SQ) includes both process and product quality. Process quality is completely transparent for end users.
However, they demand that any software complies with product quality standards. Thus, measuring product quality
becomes an absolute necessity.
Graphical User Interfaces
“GUIs as means of human-computer interaction has greatly contributed to the usability of systems” (Costagliola et al.
2000, p. 581). Recently, much effort has been devoted to create more effective graphical applications so that they better
address end users’ needs (Costagliola et al. 2000). Interface design is part of the human-computer interaction field (Galitz
2007), and is one of the factors that has the biggest impact for organizations (Machiraju 1996). In addition, Interface
design focuses on three main areas (Pressman 2007):
1. Software components
2. Software, other products and external entities
3. End users.

In this area, we can distinguish between hardware and software interfaces. For the present research only the latter are
studied.Even with the existence of powerful and useful tools, the development of GUIs is an expensive and difficult task
(Costagliola et al. 2000). For example, previous literature (Costagliola et al. 2000; Te'eni et al. 2007) reports that EUIs
require most of the implementation time and represent about 48% of the application’s code (Costagliola et al. 2000).
Interface design is the process of specifying how end users will use IS functionality as well as the information produced
(Sommerville 2006). Input and output tasks together satisfy end users’ requirements. Galitz (2007) argues that a well
designed interface is the one that goes unnoticed and allows people to focus on the information and the task that is being
performed and not get distracted by the input/output mechanisms. In addition, Galitz believes that UIs were
revolutionized by the use of graphic elements because displaying information in this format enhances end users’ abilities
for interpretation due to the reduction of the required mental load and data recoding.

A well designed UI requires taking into account several very important aspects. Pressman (2007) believes that knowing
end users’ expectations is one of them because issues such as gender, age and physical conditions, among others, affect
such expectations. For example, users that have visual problems prefer large graphics and text (Sharp, Rogers et al.
2007). In addition, another important factor is end user experience and based on that Galitz (2007) classifies them as
expert, intermediate and novice; experts prefer UIs that can lead to a high performance and allow complex tasks; on the
other hand, novices prefer UIs that guide them through tasks, and have minimal options and more help (Pressman 2007).
The challenge is to design UIs that satisfy expert end users’ needs without introducing too much complexity for the less
experienced (Galitz 2007). However, knowing end users’ needs is not an easy task, especially when they are not willing
to cooperate with developers. Moreover, it is important to do more regarding EUI design because in many cases is
unsatisfactory and organizations pay a high price for unsatisfactory quality (Te'eni et al. 2007). Thus, it is important to
research new areas that could help developers create IS that could be well accepted and used.

An Overview of Gestalt Theory

Chang et al. (2002) explain that Gestalt Theory (GT) is a family of psychological theories that have influenced several
research areas, including visual design issues. In addition, they argue that GT can be used as one of the foundations for
instructional screen design. GT is usually expressed as laws, and it intends to explain how people perceive and recognize
patterns (Chang and Nesbitt 2005). This theory proposes that perception is loaded with memories. Originally, was only
studied in psychology, but the concepts have influenced many research and study areas (Chang and Nesbitt 2005). Some
examples that applied GT are diagram design (Lemon, Allen et al. 2007), language patterns (Flieder 2007) and aesthetics
(Lim, Stolterman et al. 2007). The key GT laws taken from past research (Chang et al. 2002; Chang and Nesbitt 2005)

Proceedings of the Eighteenth Americas Conference on Information Systems, Seattle, Washington, August 9-12, 2012. 2
Gómez Reynoso et al. Measuring the Impact of Gestalt Theories in GUI

are: Balance/Symmetry, Continuation, Closure, Figure-Ground, Focal Point, Isomorphic Correspondence, Proximity,
Similarity, Simplicity, and Unity/Harmony.

Development of the Graphical User Interface Studied Attributes

An interface is the first product or service that a user touches to interact with an IS (Blair-Early and Zender 2008). It is in
charge of connecting humans with computing resources such as: operating systems, applications, and data. However,
individual preferences have an impact on how each person prefers to interact with an IS. “In many cases, the way we
access and use, and even the degree to which we rely on technology, may be vastly different from generation to
generation” (Blair-Early and Zender 2008, p. 85). GUIs help people to increase productivity (Orubeondo and Mitchell
2000). The design of GUIs should include human factors principles (Staggers and Kobus 2000) so that end users’
acceptance might be enhanced. Based on the above, our research studied a series of important aspects for GUI design,
which are described below, and which are related to guidelines specified by Blair-Early and Zender (2008), Chang et al.
(2002), Chang and Nesbitt (2005) and Te’eni et al. (2007).
1. Obvious Starting Point. It is important that end users must know how to start interaction with the content. Pre-
attentive features are proven to “pop-out” and include: size, value, hue, orientation, shape, enclosure, blurriness, and
movement, of which movement is the most basic pre-attentive feature. Pre-attentive features should be applied
because they immediately stand out from their peers. A starting point is needed because every new interface
requires a learning process. A Focal Point is important that a visual presentation has a focal point, which is called
the centre of interest or point of emphasis. This focal point intends to catch the viewer’s attention and persuades the
viewer to follow the visual message further. Cognitively, people learn through finding patterns among details. Thus,
the user must know where to begin the learning process.
2. Clear Reverse. The user must know how to reverse any action, including how to end the session. Therefore, the
reversal should become obvious “on demand” and should be omnipresent and clear but subtle. The
Balance/Symmetry is a psychological sense of balance that is usually achieved when visual ‘weight’ is placed
evenly on both sides of an axis; also, Simplicity is a graphical message must be uncluttered, but if the graphics are
complex and ambiguous the simplification process may lead to unintended conclusions.
3. Consistency. An end user must be able to quickly identify a logical, rational pattern of relationships between user
actions and effects in an interface’s content. Design patterns should have an acceptable level of consistency within
the world the interface develops. Users’ representation is just how they usually simplify the real world. Consistent
patterns and rationally connected to actions and content, users with average cognitive abilities will recognize the
patterns and their corresponding meanings. Also, Figure-Ground allows people to distinguish the foreground from
the background in a visual field. Two different foreground colors make the viewer perceive different things that are
presented in the same illustration; in addition, Unity/Harmony is the congruity among the elements in a particular
design; they look like they belong together. If the related objects do not appear within the same form, the viewer
will consider such objects to be unrelated to the main visual design, which could lead to confusion.
4. Observe Conventions. It is important to identify and respect a user’s familiar interface language of words, phrases,
images and conventions because of Isomorphic Correspondence, which is the fact that each image has a different
meaning to different persons since we interpret their meanings based on our personal experiences. In addition,
existing conventions can be built upon, extended, or even played with as appropriate for user and content
parameters.
5. Feedback. End users should receive feedback as they perform tasks. The feedback should be as immediate as
possible to the action performed in time and space. Immediate feedback is necessary so that users are informed that
their actions are having an effect. Simplicity explains that a graphical message must be uncluttered, but if the
graphics are complex and ambiguous the simplification process may lead to unintended conclusions.
6. Landmarks. Information should be available to users that suggest their location in the conceptual space of the
interface. Some of these should be available at any time. In addition, landmarks build upon end users’ ability to
build a mental model of their experience. Closure is the natural tendency that human minds have to close gaps and
complete unfinished forms, especially on those with which they are familiar. When information is missing, people
focus on what is presented, and they fill in the blanks with a familiar line.
7. Proximity. “A user should not have to traverse great physical, conceptual, or time spaces to perform similar actions
or access related content” (Blair-Early and Zender 2008, p. 101). In addition, there are at least three types of
proximity: a) space, associating content and interface in a consistent or logical evolution of X Y Z space; b) time,
making content available when the user wants it; and c) concept, grouping related items together. Proximity means
that items that are placed near each other appear to be a group. People will mentally organize closer elements into a
coherent object, because they assume that closely spaced elements are related and those that are further apart are
unrelated.
8. Interface is content/aesthetics: A user utilizes an interface to get access to content. In addition, the interface is part
of the content, not just a means to access content and it is important that the design should be aesthetic. Thus, it is
important to design the interface so that interaction is as direct with content as possible and avoids interfaces where

Proceedings of the Eighteenth Americas Conference on Information Systems, Seattle, Washington, August 9-12, 2012. 3
Gómez Reynoso et al. Measuring the Impact of Gestalt Theories in GUI

the content interferes with the user. Designers should make the interface part of the content as much as possible, and
not just an unrelated control. The interface must serve the content, not the other way around. It is important visual
objects must appear complete; they must be Balanced/Symmetric to create a sense of balance usually achieved when
visual ‘weight’ is placed evenly. In addition, content should differentiate from the ground where it is presented
(Figure-Ground).
9. Help. Errors are part of human activity, therefore it is important to design a support source of last resort --
available, but subtle. However, help must be only for the current action, not as a help menu so that users have to
search for what they need. Simplicity means that messages must be uncluttered as well as to draw a viewer’s
attention (Similarity).
The Proximity feature was divided into 3 variables based on its corresponding categories (space, time, and content).
Thus, we studied a total of 11 features (see Table 1).
Gestalt Laws

Balance/ symmetry

correspondence

Unity/harmony
Figure-ground
Continuation

Isomorphic
Focal point

Simplicity
Proximity

Similarity
Closure
GUIs guidelines
Obvious Starting X
Point
Clear reverse X X

Consistency X X
Observe X
conventions
Feedback X

Landmarks X

Proximity in space X

Proximity in time X
Proximity in X
concept
Interface is X X
content/aesthetics
Help X X

Table 1. Studied Attributes

METHODOLOGY
In order to test the effects of guidelines and GT on the effectiveness of the design of end user interfaces two versions of
the same system that performs race identification based on facial recognition were developed. System requirements were
delivered to the developer team by the researchers. The first version was developed using a traditional approach using
current developers’ skills and technical knowledge. The development approach was incremental prototyping. Each
prototype was tested in order to address any performance and technical errors. The final prototype was tested by a panel
of three experts, which made final suggestions to the developing team. The final system was completed once such
suggestions were addressed.

Before developing the second version, developers were taught the principles of GT (Chang et al. 2002; Chang and
Nesbitt 2005) and GUI guidelines (Blair-Early and Zender 2008). After that, developers modified only the GUIs without
changing anything in the algorithms used for racial identification and facial recognition. Again, the final version was
tested by the same panel of three experts. In this case, no suggestions were made.

The following describes how each GUI’s development guidelines were addressed in constructing the system used to test
the research question.
1. Design an Obvious Starting Point. The second version included a pop-up box that provides basic instructions so that
users can see what the main goal of the system is. In addition, it provides a description about how to start working
with the system.
2. Clear Reverse. Navigation buttons were added to the second version so that users could identify how to go back to the
previous action. In addition, the Close window button was eliminated.
3. Consistent Logic. Instructions and labels were redesigned so that descriptions and buttons match the action to be
performed and make them consistent throughout the system.

Proceedings of the Eighteenth Americas Conference on Information Systems, Seattle, Washington, August 9-12, 2012. 4
Gómez Reynoso et al. Measuring the Impact of Gestalt Theories in GUI

4. Observe Conventions. Some icons were changed to have a more commonsense description.
5. Feedback. Messages regarding performed actions as well as dialogues were added and rewritten.
6. Landmarks. Information regarding position of the current action performed was added so that users could understand
how deep they are in the actions.
7. Proximity. GUIs were revised regarding proximity for each point as follows:
a. Consistency for position, size and shape for all interfaces was redesigned.
b. Performance was enhanced so that time expended between actions was reduced.
c. Contents that are related were put as close together as possible.
8. Interface is content. Distracters were eliminated or changed such as animations and colors so that content was
emphasized.
9. Help. Help was redesigned so that users do not have to search for a particular issue; instead, it was designed as
context-related.
Pilot Study
Data was collected through a questionnaire which was created based on previous literature. In order to validate the
measurement instrument, a pilot study was conducted in a 30-minute session. A total of 35 students from a bachelor’s
degree program were invited to participate and 28 did. The instrument consists of 11 questions related to the studied
variables and two questions regarding demographics (age and gender) because the system that was developed for this
study fits this demography. Moreover, since all of the subjects were enrolled in the same semester of a Computer
Systems bachelor degree, it is assumed that they are equally technologically-savvy. Questions related to IS features have
a Likert scale that ranges from 1) Excellent to 7) Extremely Low. Each question was analyzed using dispersion analysis
to determine if answers behaved in a normal-like fashion. Results show that all of them were normal.
Data Collection
A two-group design was used in the present study: control GUI (NonGUIG) and Gestalt GUI (GUIG). All participants
were students of a public university in central Mexico that were in the 9th semester of a bachelor’s degree majoring in
Computer Systems. However, the school requested that students should not be assigned randomly to groups because they
already had formed their own groups. We randomly selected two out of three enrolled groups and then the system
version was assigned randomly to those groups. Demographics for each group are shown in Table 1. Age mean in both
groups is very similar as is gender representation, and is consistent with 7% of the Mexican population (INEGI 2011).
Gender
Group N Age Mean
Male Female
NonGUIG 59 20.9 31 28
GUIG 56 21.5 30 26
Table 1. Demographic Data
Each group received a half-hour training session regarding how to use the system. After that, participants used the system
for up to one and a half hours. They were free to drop out of the study at any time but no one did and everyone stayed the
full time. They were required to explore at least twenty faces that are in the database. After the session finished,
participants were asked to answer the questionnaire regarding the studied GUIs’ attributes.
RESULTS
End user perception of the 11 attributes was measured using a discrete Likert scale from 1 (best) to 7 (worst). A Mann-
Whitney U test was conducted to evaluate the hypothesis that GUIs developed using design guidelines would score
lower, on average, than GUIs developed using a traditional approach. This test was used because answers represented
qualitative differences depending on the evaluators’ perception (ordinal level) rather than an exact value of each variable
(interval level). The results of the tests are as expected and significant, for example, Starting Point (z =-4.536, p<.001)
(see Table 2). Therefore, GUI acceptance by end users is significantly dependent on the development approach used.
Test Statisticsa
Mann-Whitney U Wilcoxon W Z Asymp. Sig. (2-tailed)
Starting Point 925.000 2521.000 -4.536 .000
Clear Reverse 396.000 1992.000 -7.318 .000
Consistent Logic 906.000 2502.000 -4.520 .000
Observe Conventions 540.000 2136.000 -6.501 .000
Feedback 420.500 2016.500 -7.175 .000
Landmarks 437.500 2033.500 -7.101 .000
Proximity in Space 655.500 2251.500 -5.810 .000
Proximity in Time 482.500 2078.500 -6.910 .000
Proximity in Concept 827.500 2423.500 -4.875 .000
Interface is Content 721.500 2317.500 -5.569 .000
Help 548.000 2144.000 -6.480 .000
a. Grouping Variable: Group
Table 2. Results from Mann-Whitney Test

Proceedings of the Eighteenth Americas Conference on Information Systems, Seattle, Washington, August 9-12, 2012. 5
Gómez Reynoso et al. Measuring the Impact of Gestalt Theories in GUI

Green and Salkind (2008) suggest that differences in mean ranks between the two groups can serve as an effect size
index. For example, Clear Reverse has the biggest index (-43.72) and the Gestalt Group had an average rank of 35.57,
while the Control Group had an average rank of 79.29, also, the means difference is 2.32 (see Table 3). We argue that
this was the most important feature in designing GUIs. On the other hand, Design and Obvious Starting Point has the
lowest index (-25.30) and means difference (0.71); the Gestalt Group had an average rank of 45.02, while Control Group
(NonGUIG) had an average rank of 70.32. We argue that this is the least important feature for designing GUIs.

Ranks
Mean Sum of Effect Size Std. Means
Group Mean
Rank Ranks Index Deviation Difference
NonGUIG 70.32 4149.00 2.25 0.958
Starting Point -25.30 0.71
GUIG 45.02 2521.00 1.54 0.503
Clear NonGUIG 79.29 4678.00 3.68 1.925
-43.72 2.32
Reverse GUIG 35.57 1992.00 1.36 0.483
Consistent NonGUIG 70.64 4168.00 2.47 1.466
-25.96 0.97
Logic GUIG 44.68 2502.00 1.50 0.572
Observe NonGUIG 76.85 4534.00 3.53 1.906
-38.71 2.10
Conventions GUIG 38.14 2136.00 1.43 0.535
NonGUIG 78.87 4653.50 3.31 1.417
Feedback -42.86 1.83
GUIG 36.01 2016.50 1.48 0.632
NonGUIG 78.58 4636.50 3.63 1.66
Landmarks -42.27 2.00
GUIG 36.31 2033.50 1.63 0.489
Proximity NonGUIG 74.89 4418.50 3.31 1.664
-34.68 1.61
Space GUIG 40.21 2251.50 1.70 0.57
Proximity NonGUIG 77.82 4591.50 2.92 1.164
-40.70 1.37
Time GUIG 37.12 2078.50 1.55 0.63
Proximity NonGUIG 71.97 4246.50 3.14 2.013
-28.69 1.59
Concept GUIG 43.28 2423.50 1.55 0.537
Interface is NonGUIG 73.77 4352.50 3.29 1.702
-32.39 1.54
Content GUIG 41.38 2317.50 1.75 0.548
NonGUIG 76.71 4526.00 3.36 1.627
Help -38.42 1.74
GUIG 38.29 2144.00 1.62 0.62
Table 3. Ranks for the 11 attributes

DISCUSSION
Based on the results of the analysis and observations of how the subjects used the software, we discuss the apparent
reasons for the differences. Attributes are discussed order by its effect size index.
1. Clear Reverse (-43.72). This difference was likely caused because the first version was not clear on how to navigate
through the system. Adding a clear way for cancelling an action, and going back and forward was extremely helpful
for the users.
2. Feedback (-42.86). Participants require feedback about the tasks they are performed so that they can be certain that
every task has been done accordingly. Adding this ability likely caused the mean difference in this guideline.
3. Landmarks (-42.27). Another source of confusion in the first version was that users were not really sure how deep
they were performing actions in the system and were unaware how many clicks had to be done in order to go to a
desired point in the system. Adding a clear landmark helped users better understand the system’s structure and
navigation.
4. Proximity in time (-40.70). Users typically want to perform their tasks on time. The second version requires less time
for performing actions.
5. Observe Convention (-38.71). Users were confused in using the first version of the system. The second version
addressed eliminated the confusion. We believe that since end users are familiar with particular aspects of standard
GUIs, developers should not try to “re-invent the wheel”. Rather, developers should focus on what end users are
familiar with and just add to existing paradigms to complete new actions in the system.
6. Help (-38.42). The first version has help that can be searched. However, the second version has an improved help
because it is context-related to the current task. Therefore, users did not need to search; they just needed to activate
the help.

Proceedings of the Eighteenth Americas Conference on Information Systems, Seattle, Washington, August 9-12, 2012. 6
Gómez Reynoso et al. Measuring the Impact of Gestalt Theories in GUI

7. Proximity in space (-38.42). The second version of the system has a better object distribution in each GUI so that at
first glance, the users can see everything they need for each action. It is important for developers to manage screen
space as well as understand end users’ memory limitations.
8. Interface is content (-32.39). The second version of the system allows end users to concentrate solely on the task
rather than having additional distracters in the GUI. Therefore, they can focus on what is being done, not on the
“fanzines” of the GUI.
9. Proximity in concept (-28.69). The second version grouped tasks that belong together in a better fashion so that they
were easier to understand. This is an important issue because memory load can be reduced helping end users to
understand, learn and perform tasks in the system.
10. Consistent Logic (-25.96). The second version included redesigned instructions and labels and buttons that match the
actions performed. This consistency was of value to those who used this version.
11. Design an Obvious Starting Point (-25.30). The users did not seem to be concerned about having a starting point
because this is the first GUI that they see once the system is running. Therefore, they know where the system starts.
Figure 1 shows a plot of the mean differences for the eleven studied attributes so that differences can be perceived easily.

Figure 1. Plot for Means Differences

CONCLUSIONS
Our results are a very good indication that by applying past research into new software developments developers are able
to provide improved IS to end users. In the present study, we used Gestalt Theory (Chang et al. 2002; Chang and Nesbitt
2005) and guidelines from past research (Blair-Early and Zender 2008) as means for improving a software system. We
found that by combining this two into a single approach improves outcomes, in this case, GUIs quality perception by end
users. Results show that participants that used the software version constructed using both approaches are perceived as
higher quality compared to those that used the software developed using a traditional approach. Thus, we believe that this
approach not only provides a relevant outcome but also follows a more rigorous developing approach.

Based on our early results we believe that developers and users can benefit by using a development life cycle based on
past research. This approach enhanced outcomes in the present research. Therefore, we highly encourage developers and
researchers to further explore our approach.
Limitations and Future rResearch
Results are encouraging. However, the present study has some limitations that might have an effect on outcomes. The
study was conducted only with young adults. Results might be different if it is replicated with end users that are in
different age groups. We call for a study with more diverse participants. Participants were not assigned randomly to
groups; this could also have an effect on outcomes. It is recommended to conduct an additional study assigning
participants randomly.
We cannot identify the threshold for an effect of practical importance. However, the “starting point” effect size may
indicate a limit. We call for a research that identifies such a threshold.
In addition, both systems were developed by the same team of developers, who, unconsciously, would have improved the
second version of the system. However, while some of the benefits of the second version would have been due to general
improvement caused by a second iteration, the specific changes related to guidelines would not have been inferred from
just revisiting the prototype. It is important to conduct a new study with two groups of developers teaching only one the
GUI development guidelines and Gestalt theory principles as well. Then, both groups would be presented with the same
requirements for the system and the same resources.

Proceedings of the Eighteenth Americas Conference on Information Systems, Seattle, Washington, August 9-12, 2012. 7
 Gómez Reynoso et al. Measuring the Impact of Gestalt Theories in GUI

There might be additional issues that have an effect on outcomes that went unnoticed by researchers.
REFERENCES
1. Aberg, G., and Chang, J. Applying Cognitive Science Research in Graphical User Interface (GUI), (2005), Umea
Institute of Design) 2005, pp 23-28.
2. Ahonen, J.J., and Junttila, T. A Case Study on Quality-Affecting Problems in Software Engineering Projects,
(2003), IEEE International Conference on Software—Science, Technology & Engineering (SwSTE’03)) 2003, pp 1-
9.
3. Blair-Early, A., and Zender, A. User interface design principles for interaction design, (2008), Design Issues (24:1)
2008, pp 85-107.
4. Bosch, J., and Juristo, N. (2003), Designing Software Architectures for Usability, Proceedings of the 25th
International Conference on Software Engineering,2003,
5. IEEE Computer Society Washington, DC, USA, Portland, Oregon, pp. 757-758.
6. Bowen, J., and Reeves, S. Formal models for user interface design, (2008), Innovations in Systems and Software
Engineering (4:2) 2008, pp 125-141.
7. Buitrón de la Torre, M. "Consideraciones para el diseño de interfaces gráficas de usuario en ambientes virtuales
educativos," México.
8. Costagliola, G., Ferrucci, F., Tortora, G., and Vitiello, G. A metric for the size estimation of Object-Oriented
Graphical User Interfaces,(2000), International Journal of Software Engineering and Knowledge Engineering
(10:5) 2000, pp 581-603.
9. Chang, D., Dooley, L., and Tuovinen, J.E. (2002), Gestalt Theory in Visual Screen Design A New Look at an Old
Subject, Proceedings of the 7th World Conference on Computers in Education (WCCE’01): Australian topics,2002,
Copenhagen, Denmark, pp. 5-12.
10. Chang, D., and Nesbitt, K.V. (2005), Developing Gestalt-based Design Guidelines for Multi-sensory Displays,
Proceedings of the MMUI '05 Proceedings of the 2005 NICTA-HCSNet Multimodal User Interaction
Workshop,2005.
11. Galitz, W. (2007), The Essential Guide to User Interface Design: An introduction to GUI design principles and
Techniques Wiley Publishing, Inc, Indianapolis, Indiana.
12. Gordon, I.E. (2004), Theories of visual perception Psychology Press, USA & Canada.
13. Green, S.B., and Salkind, N.J. (2008), Using SPSS for Windows and Macintosh: Analyzing and Understanding
Data (5th Ed.) Pearson Prentice Hall, Upper Saddle River, NJ.
14. INEGI (2011), Censo de Población y Vivienda 2010 (Census of Population and Housing 2010), G.e.I. Instituto
Nacional de Estadística (ed.), INEGI, Aguascalientes, p. 54.
15. Ko, A.J., Myers, B.A., and Horng Chau, D. A Linguistic Analysis of How People Describe Software Problems,
(2006), Visual Languages and Human-Centric Computing (VL-HCC'06)) 2006, pp 1-8.
16. Lane, D., Napier, A., Peres, C., and Sándor, A. Hidden costs of graphical user interfaces: Failure to make the
transition from menus and icon toolbars to keyboard shortcuts, (2005), International Journal of Human-Computer
Interaction (18:2) 2005, pp 133-144.
17. Lemon, K., Allen, E.B., Carver, J.C., and Bradshaw, G.L. An Empirical Study of the effects of Gestalt Principles on
Diagram Understability, (2007), IEEE) 2007, pp 156-165.
18. Lim, Y.-K., Stolterman, E., Jung, H., and Donaldson, J. (2007), Interaction Gestalt and the Design of Aesthetic
Interactions, Proceedings of the Designing Pleasurable Products and Interfaces, 2007, Helsinki, Finland, pp. 239-
254.
19. Liu, Y., and Osvalder, A-L., Usability evaluation of a GUI prototype for a ventilation machine, (2004), Journal of
Clinical Monitoring and Computing (18) 2004, pp 365-372.
20. Machiraju, V. A survey on Research in Graphical User Interfaces, (1996), Utah University) 1996, pp 1-16.
21. Orubeondo, A., and Mitchell, L. What makes a useful user interface?, (2000), Infoworld), October 2000 2000.
22. Pressman, R.S. (2007), Software Engineering: A Practitioner's Approach, 6/e Mc Graw-Hill, New York, N.Y.
23. Sharp, H., Rogers, Y., and Preece, J. (2007), Interaction Design, beyond human - computer interaction Jhon Wiley
& Sons, Ltd, England.
24. Sommerville, I. (2006), Software Engineering. 8/e Addison-Wesley, London, U.K.
25. Staggers, N., and Kobus, D. Comparing response time, errors, and satisfaction between text-based and graphical
user interfaces during nursing order tasks, (2000), Journal of the American Medical Informatics Association (7:2)
2000, pp 164-176.
26. Te'eni, D., Carey, J., and SZhang, P. (2007), Human Computer Interaction: Developing Effective Organizational
Information Systems John Wiley & Sons, Hoboken, NJ.
27. Wu, J. (2000) Accommodating both experts and novices in one interface, http://otal.umd.edu/uuguide/jingwu/,
Department of Computer Science, University of Maryland, Accessed on May-2011

Proceedings of the Eighteenth Americas Conference on Information Systems, Seattle, Washington, August 9-12, 2012. 8

View publication stats