JisooLee Revised Proposal Submission

Supporting Self-Experimentation of
Behavior Change Strategies
THESIS PROPOSAL
Jisoo Lee
School of Arts, Media and Engineering
Arizona State University
jisoo.lee@asu.edu
COMMITTEE
Winslow Burleson, Arizona State University (Chair)
Jodi Forlizzi, Carnegie Mellon University
Eric Hekler, Arizona State University
David Tinapple, Arizona State University
Erin Walker, Arizona State University
Abstract
Empowering individuals with tools and support that enable them to explore, invent, and
experiment with behavior change solutions to their uniquely personal needs, throughout
their everyday lives, is critical to fostering robust, personalized, and effective
interventions and novel experiences for human flourishing. I aim to understand how
tools that foster self-experimentation of behavior change strategies for the creation of
user-driven solutions can support fulfillment and increased self-control. I plan to equip
end-users with the capacity to construct sensor-augmented responsive environments by
developing, deploying, and evaluating a toolkit that provides integrated hardware and
software coupled with motivational support pertaining to self-efficacy. I examine if
user’s self-experimentation with the proposed toolkit can promote behavior for a
personally salient home-based behavior (i.e., sitting/TV watching, snacking, or
flossing).
In this thesis proposal, I first describe how my research topic has been formed through
showing limitations of existing research and the value of end-user participation in
achieving solutions that align to individuals’ personal needs. I present my preliminary
work, which led me to identify two primary issues in designing toolkits for user-driven
solutions. Through the preliminary work, I encountered two problems: (1) How to
design construction affordances that foster user’s creation of diverse implementations
but which are still easy to use, and (2) How to guide user’s exploration to realize rich
and meaningful solutions. Focusing on addressing those issues, I propose an approach to
supporting user- and context-driven behavior change self-experimentation and, for the
purposes of this document, I formulate this approach in two parts. First, I will advance a
construction toolkit for end-user implementation of system solutions providing user
interfaces for end-user programing, a commensurate software platform, and ubicomp
smart-home hardware, including sensors and actuators. In particular, to optimize
programming capability for people’s behavior change, I plan to analyze a user-generated
collection of behavior change applications, and extract essential programming logic and
algorithms. Secondly, I propose the design of a web portal that facilitates users ability to
rapidly distill lessons from behavioral science (effective strategies for promoting
behavior change via context), HCI research (design thinking for generating solutions),
and user’s self-tracking (behavioral and contextual information). I present my work plan
on how I will advance those two parts. I first conduct user study to explore the nature of
people’s behavior change strategies in their everyday life, and methods to guide the
user’s exploration to create rich and meaningful solutions. Drawing on lessons on them,
I develop the proposed design support tool with laboratory usability testing. Its
effectiveness in users’ creation of behavior change solutions is evaluated through field
testing, in which effects of systems developed with the proposed technology in users’
behavior change are also examined. Moreover, based on lessons from the previous two
user studies, I develop the construction tool further (expanded programming elements,
and user interface design to accommodate those new features) to optimize usability
affordances and practical usefulness of the construction tool. It is evaluated in a labsetting to examine usability and capability in accommodating users’ solution ideas.
Through this research, I probe into people's exploration of behavior change solutions
with the proposed tools, and gain insights on strategies to support people's self-driven
behavior change processes.
2
1 Introduction
This document proposes the development and evaluation of a toolkit that fosters selfexperimentation of behavior change solutions. In this section, I first present my research
motivation, and preliminary work introducing two primary issues that I will focus on in
the development of the proposed toolkit. Next, I describe research objectives, and
specified questions to be explored in achieving them. I present expected contributions of
this research, and lastly discuss how my research fits the framework of Experiential
Media Systems that the School of Arts, Media + Engineering research advances.
1.1 Motivation
My research interest on behavior change technology has been shaped by the recognition
of the importance of health behavior in daily life, and potential of ubiquitous computing
technology in supporting it. Especially, through reviewing existing work in the HCI
field, I found that little attention has been given to user-driven behavior change,
although its potential is quite apparent considering the diversity of people’s needs and
situations that cannot be easily encompassed by the provision of pre-fabricated
solutions. In addition, a cues-to-action technique, having any event or stimulus that
triggers to perform the targeted behaviors, has been underutilized although it has been
recognized as a particularly effective behavior change strategy, and an opportunistic one
that ubiquitous computing can distinctively contribute to [Fogg, 2002][Intille,
2006][Andrew, 2007]. Such understanding fosters my interest in supporting users to test
context-aware applications employing cues-to-action behavior change techniques, in an
attempt to create better behavioral routines.
Behavior change
Extensive evidence underlines the importance of enhancing people’s behavioral routines
through appropriate self-regulatory processes that improve health and wellness [de
Rider, 2006][Nelson, 2007]. For example, ‘health’ habits are frequently emphasized as a
means to prevent diseases known associated to them [Maes, 2005], such as daily tooth
brushing to maintain oral health, and regular physical activity to reduce risk of
cardiovascular disease, obesity, colon cancer, and so on. In many chronic illnesses,
education for patient’s self-management is considered crucial, which supports the
patient to incorporate recommended behaviors (e.g., monitoring blood glucose for the
diabetes patient) into lifestyles. Yet, even though people have the intention of sustaining
desired behavior, the vast majority report difficulties in consistently performing those
behaviors. For example, people may find it hard to maintain a healthy diet or a pattern of
regular exercise in the face of temptations of modern life [de Rider, 2006].
Recognizing the importance of self-regulation and the significant challenges
individual’s face when employing this skill, substantial exploration has been carried out
to better understand the underlying mechanisms of self-regulation [Karoly, 1993], and
the facilitators and hindrances in establishing and maintaining desired behaviors.
According to social psychological research, people’s failure can be attributed to various
causes, including lack of appropriate goal setting and monitoring, lack of motivational
drives including internal and external sources, and a limited physical and cognitive /
emotional resource [Bandura, 1991][Baumeister, 1994].
Ubiquitous computing for behavior change
As the trend towards creating technology-enriched home environments progresses,
researchers in HCI are increasingly exploring the use of technology to promote behavior
3
change in various domains, such as physical health [Maitland, 2010][Nawyn, 2006],
affective stability [Ståhl, 2009] and energy conservation [Bang, 2007]. Researchers
highlighted several opportunities of ubiquitous computing in supporting people’s better
lifestyles. Intilli recognized the provision of just-in-time information with mobile
computing devices and context-aware computing as a prospective approach in
preventive healthcare, as it can motivate behavior change by providing well-timed and
well-tailored information to users at points of decision, behavior, or consequence [Intilli,
2006]. Pavel, et al. focused on people’s lifestyle adaptation driven by self-awareness,
and asserted a crucial role of technology in allowing people to better understand
themselves [Pavel, 2010] with abundant and reflection-provoking information about
their daily life. There has also been effort to examine if such technology-augmented
interventions have real impacts on people’s behavior change experience. Empirical
evidences from those studies that validate effects of their proposed systems imply the
potential of ubiquitous computing in supporting people’s behavior change [Consolvo,
2008], although Hekler, et al. point out some problematic aspects in those studies such
as confirmation bias [Hekler, 2013].
Limitations of existing work
Although substantial attention has been given to people’s behavior change in the HCI
community as described above, I identified two significant limitations through
reviewing the literature: (1) under-explored use of the cues-to-action technique, and (2)
little attention to user-creation of behavior change solutions.
(1) Under utilization of the cues-to-action technique
The suggestion [Fogg, 2002], just-in-time information [Intilli, 2006] or prompts
[Arroyo, 2005], or cues-to-action [Medynskiy, 2011] behavior change strategy is to
provide the right message at a time when it can be most effective. It is one of seven
types of ‘persuasive strategies’ proposed by Fogg, and one of seven design principles for
persuasion techniques identified by Arroyo, et al. Mendynskiy, et al. also recognized its
potential for use in an interactive application that supports healthy behavior change.
Although some research has explored the just-in-time prompt technique in such ways
[Nawyn, 2006][Arroyo, 2005], little attention has been given to it in spite of its
acknowledged significance, compared with other more frequently employed techniques,
including goal-setting, self-tracking, priming, intrinsic motivation, social influence, etc.
[Klasnaja, 2011]. Considering the crucial role of contextual cues in habit formation
[Wood, 2007], Hekler, et al. identify the application of cues-to-action technique within
ubiquitous computing systems to be a fruitful area of inquiry for developing effective
interventions [Hekler, 2013].
(2) Lack of support for user-driven behavior change solutions
Psychological research on habit formation and HCI work on the cues-to-action
technique for behavior change propose that employment of contextual cues can be
useful to foster habits, however, potential context cues are quite varied and highly
idiosyncratic. For example, Räisänen, et al., who investigated opportune moments to
show warning pictures about the dangers of smoking to people, found that the timing
seems to vary between individuals and it complicates the design of applications
leveraging this strategy to a great extent,[Räisänen, 2008]. However, most ubiquitous
computing research is about proposing pre-fabricated solutions for target behaviors,
with little attention to user’s self-experimentation of alternative strategies. Users may be
able to obtain solutions that better fit their circumstances, by creating their own
interventions. Even though more comprehensive solutions to promote behavior change
4
may be effective [Consolvo, 2008] it is very likely that no one solution will be effective
for everyone.
This methodology, empowering users to create system functionality directly bypassing
the manufacturer or professional developer [Chin, 2010], is commonly referred to as
end-user programming, and is characterized by the use of techniques that allow nontechnical people to create applications [Cypher, 1993]. It is favored by the advantage
that individuals have more intimate knowledge about their activities and environments
than a hired developer, which is essential in building appropriate context-aware
applications. Dey and his colleagues suggested that users need to have the ability to
create and modify applications as those activities and environments change. Without
such ability, context-aware applications acting implicitly often annoy users and fail to
meet their needs [Dey, 2006]. I propose that this end-user programming approach will
be beneficial for people involved in finding and developing solutions that better fit their
behavior change goals, lifestyles, and environments. As well, I expect affective benefits
of user’s involvement in the creation process to obtain behavior change interventions,
including ownership and a sense of attachment toward artifacts created by them. User’s
creative exertion and ownership of outcomes may serve as positive factors that lead to
sustained long-term engagement in using an applications and pursuing a target goal.
Although there has been considerable research on end-user programming tools for
creation of context-aware applications at home environments, little attention has been
given to the provision of toolkits focused on behavior change. Doing so will likely
involve addressing user needs and situations that are quite distinct from the ones
currently approached by existing smart home control and home automation systems
[García-herranz, 2010; Dey, 2006]. Most tools are to support control of appliances or
environmental equipment, and thus the expressiveness of them is primarily examined if
it can encompass application ideas aimed at such usage. Ease of use is typically
considered as the most prominent issue regarding user satisfaction [Chin, 2006].
Shifting a focus from general home environment control to support for behavior change
will require expanded evaluation criteria beyond ease of use, including effectiveness in
fostering behavior change; thus, functionality and research methodology needs to be
advanced to address these requirements.
1.2 Preliminary Work
Review of existing work, discussed in the previous section, shaped my research
intention to support user’s creation of context-aware applications employing the cues-toaction technique for behavior change. My research objectives are based on three
foundations: (1) the impact of context-aware cuing applications on people’s behavior
change, (2) benefits of user creation in obtaining behavior change solutions, and (3)
user’s capacity to creating behavior change interventions. In this section, I present my
preliminary work to examine and established these foundations and discuss primary
findings from it.
Formative user studies
The preliminary work can be divided into: (1) a formative field study with participantconceptualized/ researcher-implemented systems involving seven participants for one to
four weeks of deployment at their homes; and (2) development of a visual programming
tool and a study involving user testing of it, with 36 participants for a 1.5-hour session,
in a lab-based simulated home setting. In the initial study, after being introduced to a
novel smart home technology and scenario, participants created their own application
scenarios, addressing issues that they considered significant in their daily lives through
5
appropriate application of the smart home technology. Systems were then developed and
configured to realize these scenarios. In the second study, the visual programming tool,
GaLLaG Strip [Lee, 2013], was developed and tested with end-users who had no
programming skills or prior programing experience. After a brief introduction, users
defined their applications in a linear fashion, using simple if-then conditions.
Participants were asked to conceive and implement their own applications.
Through these studies, I investigated the following questions: (1) Do context-aware
cuing applications have an impact on people’s behavior change? (2) Is it beneficial for
users to create their own behavior change solutions? (3) Can users produce applications
in a manner that achieves satisfactory outcomes through enjoyable processes?
With respect to the first question, I observed both encouraging results and problematic
aspects with regard to the effect of cues-to-action applications in improving target
behaviors. Some participants showed better performance, but some did not comply with
cues given by systems. Beyond their lack of willpower, the degree of difficulty or
required effort to complete a target activity greatly affected people’s compliance. For
example, a participant who showed distinctive progress in taking vitamin pills compared
to her previous performance, failed in her next objective, in which she was going to
spend an hour on studying a foreign language. Two participants (with the goals of
“learning how to play ‘Blackbird’ on my guitar” and “writing an autobiography”,
respectively) explicitly expressed their increased stress, as the system’s presence exerted
a pressure to engage; these participants expressed a feeling of guilt, when they did not
comply.
With respect to the second question, participants’ diverse objectives (ranging from
sleeping on time to completing an autobiography) and different behavior change tactics
employed (e.g., rewards, punishments, priming, smooth transition) were made possible
by GaLLaG Strip, a tool for end-user scenario-creation. Users demonstrated different
preferences toward behavior change tactics. For example, a participant was willing to
prohibit her Internet use as a punishment, however, another participant did not want to
be reminded about her failure at all.
With respect to the third question, most participants easily understood how to program
applications with the GaLLaG Strip programming tool; were able to create the
applications that they desired, after watching a short instructional video; and were
highly engaged in the experience. However, users differed with respect to their
satisfaction toward their outcomes. While some people generated applications involving
different patterns and interactions, others simply repeated a pattern of a given sample
application. They expressed their discomfort with the sophistication of their outcomes
(e.g. “Though I feel I can make it better, I don’t know what I can do more.” “I know
there are many more things that can be done, but I only made it like this.”). Some people
may need further support to guide their planning in ways that go beyond facilitating
their implementation and to enhance richer explorations, especially during the start-up
phase. Furthermore, such guidance may also be necessary for users to attain outcomes
that are more effective in terms of their behavior change. In these studies, the frustration
with behavioral outcomes may have largely resulted from people’s selection of
inappropriate strategies. For example, a person may have failed in setting an achievable
goal [Baumeister, 1994].
Outstanding issues
Through the studies, I found two salient issues in designing toolkits for fostering user’s
experimentation of alternative behavior change strategies: (1) How to guide user’s
exploration to realize rich and meaningful solutions; and (2) How to design construction
6
affordances that foster user’s creation of diverse implementations but which are still
easy to use. (1) Supporting creativity
In these studies, participants were uncomfortable with their limited ability to explore
and implement a wide range of interventions (e.g., when they engaged in slight or
repetitive re-implementation of a given sample application). To the best of my
knowledge, there is no research that recognizes the need to support user’s ideation
involving toolkits for end-users’ creation of context-ware applications. One reason may
be that existing ubiquitous computing approaches tend to address the most apparent
problems and apply a range of relatively straightforward solutions for them. However, in
moving beyond today’s smart home applications, to the realm of behavior change, we
encounter, so-called, ‘wicked’ problems. That is, the problem of behavior change shares
much of characteristics of wicked problems proposed by Rittel & Webber such as
‘There is no definitive formulation of a wicked problem’, ‘Solutions to wicked problems
are not true-or-false, but good or bad’, and so on [Rittel 1973], which require
stakeholders to exert more resourceful and nuanced approaches to problems.
(2) Balancing between simplicity and expressiveness
Expressiveness, being able to produce a wide range of application types, and ease of use
/ learning are primary concerns in developing end-user programming tools [Garcíaherranz, 2010]. In developing a tool for non-expert users, simplicity is weighed over
functionality. [Resnick, 2005] emphasized the importance of simplicity as one of the key
design principles for creativity support tools, asserting that reducing the number of
features can actually improve the user experience (in the context of their observation of
development of a Programmable LEGO Brick in the mid-1990s). However, for the
realization of effective ubiquitous computing behavior change scenarios it is important
to balance simplicity with expressiveness, in ways that prioritize end-users capacity to
create a wide range of experiences.
1.3 Research Objective and Contributions
In this dissertation, I aim to show the value of users’ self-experimentation of behavior
change solutions in prompting behavior for a personally salient home-based behavior
(i.e., sitting/TV watching, snacking, or flossing). I intend to propose a tool with focus on
the two salient issues that I found through the preliminary work, and evaluate
effectiveness of it in helping users’ behavior change. I will investigate the following
thesis:
People’s use of context-aware applications self-created with appropriate
design support can result in better achievement in resolving target behavioral
issues, compared with no use of it.
I am addressing the following research questions through iterative design processes
involving field deployment of prototypes, and evaluation:
[1] How can a toolkit support users to generate rich and meaningful behavior
change solutions?
[2] What are essential programming elements in user’s creation of applications for
their behavior change?
[3] How can a toolkit holistically support self-experimentation of behavior change
solutions?
[4] What is the value of self-experimentation in people’s behavior change?
7
This research provides several contributions. First, the proposed toolkit can help
people’s pursuit of health behavior, desired results of which can save personal and social
cost. Secondly, the toolkit can serve as an easy-to-use implementation technology for
researchers in behavioral health and related fields. With a tool that is simple but
adaptive to behavior change, technology-augmented interventions can be tested and
adapted relatively easily. Third, I am conducting field-testing to examine the usefulness
of the toolkit in people’s everyday lives and homes. To the best of my knowledge, there
have been no field tests that deployed end-user programming tools for creation of
context-aware applications within users’ homes. This may be partially due to
complicated issues in adapting systems to the dynamic nature of the end-users, who
often live in dynamic heterogeneous environments. Overcoming these challenges with
generalizable strategies will be an additional research contribution.
1.4 Research in Media Arts and Sciences
The proposed work builds on systems and approaches that I am developing within the
Motivational Environments Research Group1, which is directed by Dr. Winslow
Burleson. It aims to enrich personal motivation and self-actualization through ubiquitous
computing by providing the opportunity to integrate virtual and physical domains
holistically. In line with this agenda, I have been focusing on how technology can be
integrated into people’s daily life to support them in learning about themselves and
sustaining their engagement in self-identified activities.
I intend to create new user experiences in pursuing self-directed life improvement and
physical / emotional health and well-being, with novel methods for sensing and
feedback, based on activity-centered, user-centered and participatory design approaches.
My work will explore integrated experiences across the spectrum of information flow:
from sensing physical events, to interpreting this data, to providing persuasive interfaces
and coordinating interactions. The Experiential Media Systems framework2 that the Arts
Media + Engineering program3 works with defines experiential media systems as a
media system with a feedback loop including four components: user(s), sensing,
perception and cognition, and feedback. The following describes how the proposed
work involves those four components:
The User(s): The user is a human who creates and uses systems and derives
benefit from doing so.
Sensing: In many Arts Media + Engineering experiential systems, sensing is
provided through a novel input device such as optical motion capture, a tangible
input device, or the like. In my work, sensors are used to recognize user’s actions,
especially, location and use of objects in home environments.
Perception and Cognition: In experiential media systems, this typically models
the user’s behavior within the system, feeds into a control system, and generates
output. In the proposed work, users build and modify models that represent context
when feedback should be triggered.
Feedback: I intend to integrate two types of feedback modes: audio using situated
display, and textual information via mobile devices. I assume that user-defined or
created content is triggered according to user’s current or past behavior.
1
http://hci.asu.edu
2
Hari Sundaram et al. “Experiential Media Systems”. In: Encyclopedia of Multimedia. Ed. by B. Furtht. New
York, NY: Springer-Verlag, XXVIII. Chap. Experiential Media Systems
3
http://ame.asu.edu
8
The feedback loop is closed when the system’s behavior alters the user’s behavior, and
the user’s behavior alters the system’s behavior. In the proposed work, user’s behavior
change is the ultimate goal. Ideally, user’s ability to adapt systems to their needs and
situations will be fostered. To achieve such research goal and develop systems pertinent
to it, I work with a multi-disciplinary team consisting of researchers from health
psychology, HCI design, software engineering, informatics and decision systems, and
media arts.
2 Background
I have proposed my research objective in the preceding section, with two key foci: (1)
effects of appropriate cueing in behavior change; and (2) benefits of users’ creation in
acquiring behavior change solutions and their willingness to do so. In this section, I first
discuss related research that supports these assumptions. As validation of the cueing
approach, I present a perspective that highlights people’s intuitive processes, and
research on contextual cues in psychology. Secondly, I describe research findings that
suggest prevalence of people’s making things that they need in daily life, and benefits
from it. Next, I review a trajectory of research on end-user development or
programming in HCI, and identify principal issues brought up recently. Lastly, I
introduce two perspectives-- trial-and-error problem-solving and self-regulatory
processes--that I adopt in order to identify and evaluate design goals and requirements
that advance the development of toolkits for end-user self-experimentation.
2.1 Cue-Triggered Behavior
In this section, I present psychological studies that support effectiveness of the cues-toaction technique. Significant research suggests that most actions of people are driven by
intuitive processes, and it may be quite true in habitual behaviors. In particular, most
research on habit formation highlights the influence of contexts that triggers people’s
automatic behavior. This research implies that once a particular behavior is associated
with certain cues, the behavior can be sustained long term without consuming, so called,
‘self-control strength’, which has been found to be ‘quite limited and hence can be
depleted readily’ [Muraven, 2000]. On the other hand, research on ‘implementation
intentions’, which refer to intentions that specify, where, when, and how these goals are
acted upon [Holland, 2006], has shown how intentional goal pursuit can be transformed
into automatic actions. The cues-to-action technique can be understood as enhancing
implementation intentions with explicit cues.
Intuitive processes
As has been discussed by several psychologists, people’s mental energies and processes
can largely be divided between active “rational” vs. more passive and automatic
“intuitive” processes [Haidt, 2001], and the vast majority of people’s daily tasks are
driven and selected by people’s intuitive processes rather than rational processes. Put
differently, people rely on the intuitive processes to function as a sort of “autopilot” for
most of the actions and behaviors they take. This is true even for many decisions that
people may perceive as rational. As Haidt has eloquently pointed out in his work
[Haidt, 2001], many “rational” decisions are often secondary explanations for decisions
that were made by our intuitive processes.
However, many of the proposed behavior change techniques emphasize utilizing
people’s rational processes rather than programming the autopilot [Michie, 2013],
favoring strategies such as goal-setting, self-monitoring, problem-solving, numeric
9
feedback, or education that largely focus on strengthening people’s rational processes to
counteract the autopilot. This strategy, however, would require people to engage in
increasingly longer periods of time of conscious rational thought that consumes some
inner limited resource, called ‘self-control strength’ by Baumeister, et al., and thus
largely cannot last over a sustained period [Baumeister, 2007]. Muraven, et al.
[Muraven, 2000] describe that, “Coping with stress, regulating negative affect, and
resisting temptations require self-control, and after such self-control efforts, subsequent
attempts at self-control are more likely to fail.” Considering such inevitable failures
consequent on the depletion, while it is often important, it is often not sufficient to use
rational self-regulatory skills alone to promote sustained behavior change [Hekler,
2013].
Programming the autopilot
Hekler, et al. highlighted that much of the research on habit formation has started to
place greater emphasis on the impact of context driving our autopilot rather than
attempting to overtake people’s intuitive autopilot via reason [Hekler, 2013].
Specifically, Wood has explored the underlying research and processes whereby
automated behavioral routines or habits occur and has identified the importance of
contextual triggers such as environmental cues (e.g., the refrigerator as a cue for eating),
social cues (e.g., having lunch with friends), previous behaviors (e.g., flossing right after
brushing your teeth), and time of day (e.g., always brushing your teeth at the same time
each evening; [Wood, 2007]). Indeed, the impact of context on behavior was a central
tenet behind Fogg’s Three Tiny Habits System [Fogg, 2012]. In Three Tiny Habits, an
individual is instructed on developing behavior-linked routines (e.g., after you brush
your teeth, floss one tooth). This strategy is to link new behaviors with previously
established behavioral routines. There are a variety of other context cues that might be
useful to foster habits.
Meanwhile, there is a class of studies on ‘implementation intentions’, which attempts to
offer an idea on how goal-directed behaviors can be turned into habitual ones. Motivated
by recognition of the limitations of the reasoned action approach and insufficiency of
only forming goal commitments, it suggests to form ‘if-then’ plans (that is,
‘implementation intentions’; e.g., “if I enter the kitchen after waking up, I will have half
of an apple”). Forming implementation intentions seems to bring benefits over and
above goal intentions (that is, only goal commitments) [Gollwitzer, 2006]. Because
forming an implementation intention implies the selection of a critical further situation,
the mental representation of this situation becomes highly activated, and hence more
accessible. The heightened accessibility of the “if” part of the plan [Parks-Stamm, 2007;
Webb, 2007, 2008] means that people are in a good position to identify and take notice
of the critical situation when they subsequently encounter it [Webb, 2004]. Studies also
reveal that implementation intentions shape a strong association between the specified
opportunity and the specified response [Webb, 2007, 2008], and such links finally make
the initiation of the goal-directed response specified in the if-then plan become
automatic action initiation.
2.2 User Creation
The attempt to integrate users as creators brings some fundamental questions such as:
are they willing to create; are they capable of creating; and what advantages are
expected? I surveyed existing research that examines people’s creative practice in
everyday life and which suggests beneficial aspects by allowing user’s participation.
User creation often results in outcomes useful to them, and even leads to social
10
innovation, taking advantage of their intimate knowledge on their needs and
environments, and persistent work through using systems. Highlights of this literature
survey are presented here.
Everyday creativity
Although people are frequently characterized as consumers in the modern society,
creation of artifacts for everyday use is part of our culture. Buechley, et al. [Buechley,
2009] depict such phenomena as follows:
“People are driven to customize their objects and build things. Passionate makers sew
dresses, build furniture, cook meals and write computer programs. People also spend
copious amounts of time tinkering with the things they own. They decorate their
notebooks, hack their cell phones and fix their cars. Groups often get together to share
these techniques for building, modifying and embellishing artifacts, and vibrant social
communities develop as a result.”[Buechley, 2009]
Wakkary, et al. [Wakkary, 2007] suggest the ongoing presence of designers in the home,
‘everyday designers’, based on his ethnographic study where he observed people’s
spontaneous action of devising things to satisfy their everyday needs. Systems and
routines continually evolve through design-in-use to address individual needs, and testin-use that judge the quality and success of a designed system. Such findings are not
new, Alexander [Alexander, 1964] suggested the presence of unselfconscious design.
“Unselfconscious process is a design process undertaken on a cultural level and over a
long period of time, in which designed items are shaped gradually and continually to fit
the surrounding, ever changing context. Individuals participate in this process in an
unselfconscious way, simply recognizing a failure in the system and reacting in a
corrective way to achieve a well-fitting form.”
Wakkary, et al. found that people do not consciously understand the full complexity of
the system, but have tacit understanding that is clear through use. People appropriate or
create artifacts that can better serve for them than ones developed by professionals,
having tacit knowledge on their daily life that is ever-changing [Wakkary, 2007]. Such
observation corresponds to tacit ‘knowing-in-action’ proposed by Schön in his paradigm
of reflective practice [Schon, 1983]: “When we go about the spontaneous, intuitive
performance of the actions of everyday life, we show ourselves to be knowledgeable in
a special way. Often we cannot say what it is that we know… Our knowing is ordinarily
tacit, implicit in our patterns of action and in our feel for the stuff with which we are
dealing [Schon, 1983, p.49]”
Benefits
Alexander and Wakkary, et al.’s arguments not only suggest people’s involvement in
design for everyday use, but also reveal the existence of ‘bad fit’ [Schon, 1983] between
given systems and their needs. Such mismatch is considered as an inevitable result in the
situation that development of artifacts is separated from actual use of them. Developers
create systems making decisions for users for situational contexts and for tasks that
these designers can only anticipate [Fischer, 2000], however, anticipating all possible
uses in advance is as impossible [Greenbaum, 1991][Nardi, 1993] or at least very costly
[von Hippel, 2002]. Recognition on such drawback has shaped interest in empowering
end-users to build systems that they want [Nardi, 1993]. User’s situated and continuous
creation is considered as a prospective strategy to obtain systems more useful and usable
to individuals.
11
In addition to the benefit from the practical perspective, affective and educational values
are expected, which are well described by the following analogical statements by
Resnick, et al.: “The stereo has many attractions: it is easier than the piano to play, and it
provides immediate access to a wide range of music. But “easy of use” should not be the
only criterion. Playing the piano can be a much richer experience. By learning to play
the piano, you can become a creator, not just a consumer, of music, expressing yourself
musically in increasingly ever-more complex ways. As a result, you can develop a much
deeper relationship with (and deeper understanding of) music.” [Resnick, 1996]
So far significance of end-user participation has been discussed implying the desirability
for people to be designers, however it does not mean that being a consumer is wrong.
Rather, Fischer and Scharff asserted the existence of a continuum of user roles raging
from a passive consumer to a ‘meta-designer’, and the need to provide appropriate
support for each role [Fischer, 2000].
2.3 End-User Programming Tools
In the previous section, limitations of conventional development processes have been
discussed, which separates the design time from the use time [Fischer, 2000], given that
user requirements are diversified and changing, and at times hard to identify precisely
[Lieberman, 2006]. End-User Programming (EUP) is a notion aiming to resolve such
issues by enabling users to develop and program information technology systems. In this
section, I review the emergence and advancement of end-user programming research,
and primary approaches.
End-user development, End-user programming
The idea of EUD/EUP has been widely accepted across fields since the introduction of
the first computers into common workplace use and as EUD/EUP represents the
ultimate level of user participation in design processes [Syrjänen, 2011]. EUD is a
multidisciplinary (software engineering, HCI, CSCW, etc.) research topic, intending to
relieve such issues by empowering end-users to develop and adapt systems themselves
[Lieberman, 2006]. A common view is that end users intimately know their everyday
domain and aspects of their actual knowledge work lack technological and design
knowledge. The research field of EUD has generated different approaches. They can be
divided into two types of end-user activities from a user-centered design perspective: (1)
parameterization or customization, and (2) program creation and modification
[Lieberman, 2006]. The second set seems more desirable, as it allows more flexibility
than the first. However, approaches in the first set are often utilized “for a ‘gentle slope’
of increasing complexity to allow users to easily move up the first to the second set of
activities [Lieberman, 2006]. End-user programming is a class of research approaches
favoring the second type, which intends to enable end users who have not necessarily
been taught how to write code in conventional programming languages to write
programs that instruct computers to conduct their desired objective [Cypher, 1993].
The emergence of the EUD and EUP field is often traced back to the first developments
of the end-user-programmable systems. It was followed by interest in user studies in situ
and in stimulating conflicts between systems design and use, an these have eventually
led to interest in work domains [Syrjänen, 2011]. The pioneering work on end-user
programming and programmable systems began in the early 1970s, and was extended in
the 1980s. There was a notable shift in the direction of research in the 1990s when an
ethnographic stance that emerged at 1980s in IT research became combined with the
topic. It led to several new ideas, such as task-specificity [Nardi, 1993], and support of
user’s entire or larger scope of work [Fischer, 2006][Blum, 1996].
12
Approaches
Task-specificity
The increasing attention to social and dynamic aspects of programming practices led to
emergence of domain driven EUP, which resulted in diversification of EUP solutions
differentiating a variety of application domains and user groups. For example, they
range from professional programming tools that demand special technological and
systems development knowledge, to user friendly solutions based on declarative
specifications and interactive support) [Syrjänen, 2011]. Task-specificity, commonly
credited to Nardi, highlights that end-user programming tools should support a specific
interest of people (that is, application domain), and use formal languages and notations
closely match their interests and domain knowledge: “We have argued that people are
adept at learning formal systems and notations. If that is true, whey then, have so many
trouble learning conventional programming languages? The answer is that it is only
when people have ‘a particular interest in something’, such as knitting or baseball, or…,
that they readily learn the formal languages and notations that describe the elements and
relations of the system of interest [Nardi, 1993].” Similarly, Fischer emphasizes
representations of evolving artifacts that are less abstract and less alienated from
practical use situations, to support human problem-domain interaction [Fischer, 1993].
These task or domain specificity approaches inevitably sacrifice generality for the power
of specialized interactions. A problem is “to determine just how task-specific a language
should be; to some extent this depends on the circumstances under which it will be used.
Careful study of those circumstances is necessary to achieve the right level of taskspecificity [Nardi, 1993].”
Collaborative work
Through investigating users of spreadsheet and CAD, which are two software systems
recognized as success of end-user development in the real world, Nardi observed
programming communities of cooperating users [Nardi, 1993]. Working collectively,
end users can create richer, more sophisticated applications than they could working
independently. Communities of users span a continuum of programming skill; ones who
have a higher level of programming skills help others learn new things. Fischer and
Scharff implied motivation and rewards (e.g., enjoying the feeling of good citizenship to
a community) as another beneficial aspect of community-based collaboration [Fischer,
2000].
Support for a larger scope of work
The scarcity of ideas to try or uncertainty about requirements for a development goal is
considered as one of primary barriers that end users encounter [Cao, 2011][Ko, 2011].
With the concept of ‘domain-oriented design environments’, Fischer emphasized not
only the task- or domain-specific approach discussed above, but also necessity to allow
users to act as designers and be creative going beyond support for construction [Fischer,
2007].
2.4 Self-Experimentation for Behavior Change
On the one hand, self-experimentation can be understood as trial-and-error problem
solving; it has a problem, all aspects of which is difficult to be grasped at once, and thus
attempts to understand it better through testing potential options, and finally reach
acceptable solutions. On the other hand, it should be understood as part of self-regulator
processes to ensure successful results in goal pursuit.
13
Trial-and-error problem-solving
With other aspects such as information gathering, and problem formulating, trial-anderror learning via a process of a conscious experimentation is a prominent feature in
problem-solving [Tomke, 1998; von Hippel, 1988, 1995]. Experimentation begins with
the selection or creation of one or more possible solutions. The solution is then built,
tested against an array of requirements and constraints [Tomke, 1998]. Test outcomes,
including new information that he or she did not (was not able to) know or foresee or
predict in advance, are used to revise and refine the solutions under development, and
generally, progress is made in this way towards an acceptable result.
Self-regulating
Despite a range of views that differ in the various principles of self-regulation they
emphasize and the specific mechanism they propose, generally self-regulatory processes
can be understood “as a dynamic motivational system of setting goals, developing and
enacting strategies to achieve those goals, appraising progress, and revising goals and
strategies accordingly [de Rider, 2006].” Bandura suggested that the major selfregulative mechanism operates through three principal sub-functions: “These include
self-monitoring of one’s behavior, its determinants, and its effects; judgment of one’s
behavior in relation to personal standards and environmental circumstances; and
affective self-reaction. Self-regulation also encompasses the self- efficacy mechanism,
which plays a central role in the exercise of personal agency by its strong impact on
thought, affect, motivation, and action. [Bandura “Social cognitive theory”]”
Zimmerman’s structure of self-regulatory process that divides the whole process into
three cyclical phases provides greater clarity: (1) forethought phase involving goal
setting and strategic planning encouraged with motivational factors, (2) performance
phase involving self-control and self-observation, and (3) self-reflection phase where
self-judgment and self-reaction occur [Zimmerman, 2000].
Such structure and mechanisms of self-regulation may significantly influence
individual’s self-experimentation process. Thus, its effectiveness should be explored in
relation to various self-regulatory problems, goal domains, and potential moderators of
it.
3. Proposed Approach
I intend to develop tools that facilitate user’s self-experimentation of alternative
behavior change strategies, investigating the general hypothesis that individuals will be
empowered to find solutions that better fit oneself through testing various solutions. I
propose to support two phases, which are primary in the trial-and-error problem-solving:
(1) generation of application ideas, and (2) construction of applications.
I develop a construction toolkit that provides hardware and software well-suited
primarily for creating context-aware cues for behavior change, aiming to allow users to
easily construct applications that they desire, with a low learning curve in terms of their
becoming familiar with the toolkit. I especially focus on balancing between simplicity
and expressiveness, which has been identified as one of the issues described in Section
1. With observation in the preliminary user studies, I found the necessity to support
users’ generation of application scenarios. Cao, et al. also highlighted the same issue
[Cao, 2011]; they found lack of mechanisms to nurture problem-solving skills,
creativity, and design thinking, as one of reasons of difficulties that people had in
existing empirical studies of end-user programing tools despite considerable progress in
terms of ease in learning [Gross, 2010][Ko, 2004]. Fischer asserted the necessity to
14
support people’s creativity going beyond simple construction kits to “assist their truly
interesting objects” [Fischer, 1993]. Drawing on existing strategies to support people’s
design processes, I identify specific goals that should be a solution formulation process,
and a variety of approaches proposed in HCI design as an attempt to support designers
and user creativity.
3.1 Toolkit for Construction
The proposed toolkit will allow end users, those with fundamental literacy on computer
use and knowledge on simple logic but having little or no computer hardware and
programming skills, to easily understand functionality of a toolkit and construct sensoraugmented responsive systems, with minimum instruction or external help. More
specifically, users will implement rule and event-based system interactions, by
composing programming elements within an easy to use visual programming interface.
In this section, I describe details on programming elements and user interface, contextawareness, and media types.
Description of application behaviors
Rule-based systems
As presented previously, I aim to foster user’s construction of systems with behavior
change interventions, primarily through employment of the cues-to-action technique.
When users are effective, the systems should produce appropriate actions at appropriate
moments, promoting user’s engagement in the specified behavior. For this purpose, in
the proposed tool, users create rule-based context-aware applications by describing a
'situation', and associated 'action' (sensed via wireless off-the-shelf sensors), with
provided primitives and logic/control constructs. The rule-based approach in contextaware computing has been widely adopted, due to its logical simplicity and
effectiveness in controlling various situations, although it has several limitations.
The use of rules is considered effective in specifying a particular situation with
contextual information and has been demonstrated by a number of researchers using
rules for activity inference [Truong, 2004; Dey, 2006; García-herranz, 2010]. For
example, a rule for detecting meal preparation can be: “IF resident was in the kitchen
AND (resident accessed meals ingredients cabinet AND resident accessed plates or
utensils cabinet) OR resident used an appliance THEN a meal was prepared” [Dalal,
2005] Vurgun et al. compared a rule-based and statistical approach in a practical setting,
intending to develop a system to assist dementia patient’s medication taking (that is,
medication prompting at an appropriate moment. They chose a deterministic rule-based
implementation for two reasons as below [Vurgun, 2007]: (1) it was much simpler for
engineers and ethnographers to agree on the rules than on costs, and to implement the
tens of lines of dispatch code, especially not intending sophisticated system response,
and (2) erroneous behavior caused by its limited accuracy, that is, medication reminders
at wrong moments, did not disturb their target user group, the elders. They showed high
tolerance threshold to such inaccuracy. In the case of my research, I anticipate much
lower disturbance that users may feel. Through creating rules, they can be familiar with
potential limitations and such understandable errors may cause less annoyance. In
addition, target systems to be created with a given end-user tool are to prompt people to
engage in some everyday activities that are less critical with occasional ignorance.
In attempts to provide end-user’s programming tools, rule-based description is favored
due to its correspondence with people’s common description style. For example, Dey
and his colleagues collected a total of 371 application descriptions, and found that every
15
subject described their applications in terms of if-then rules, using the form ‘“if I…” or
“when I…” am in a particular situation, perform this action.’[Dey, 2006] As matching
user’s conceptual model, rule-based description allows ease of user’s representation
with less cognitive load.
The rule-based approach has several limitations, including the demand of creating rules
manually, and high cost in using a very large number of sensors for more precise
activity recognition [Vurgun, 2007]. However, given the effectiveness in terms of
accuracy, cost, use situation tolerable to some errors, and user’s ease in understanding
and expressing, I believe that the rule-based description will be appropriate for my
research and design goals.
Expressiveness
In considering the range of possible applications that will be supported, I focus on
extraction of programming elements optimized for behavior change, balancing between
simplicity and expressiveness. Integrating more elements for the sake of broad
expressiveness often yields complexity against ease in learning and using, and thus
tradeoff between complexity and simplicity is often required [Repenning, 1996]. As a
tool for non-expert users, simplicity is weighed more in spite of limited functionality,
than complexity with wide expressiveness. With increasingly complex artifacts in the
age that everything is going digital, simplicity became a crucial issue in technology and
design. Maeda proclaimed the value of simplicity in technology and design, in contrast
to feature-laden artifacts [Maeda, 2006]. Resnick, et al. [Resnick, 2005] identified the
importance of simplicity as one of design principles for creativity support tools. They
assert that reducing the number of features can actually improve the user experience,
with their observation of development of Programmable LEGO Bricks in the mid1990s. However, it is still important to allow a certain degree of flexibility.
I plan to find a set of programming elements optimized for behavior change
interventions, through analyzing user generated application scenarios. This process is
similar to a method by Dey and his colleagues [Dey, 2006] in their development of enduser programming tools for smart homes. I extract elements from collected scenarios,
based on use frequency and necessity with respect to acknowledged behavior change
principles.
Interaction for programming
I adopt visual programming techniques for ease of learning, and mobility for user’s in
situ making and modifying.
Visual programming
Visual programming systems utilize notations that are primarily visual rather than
linguistic (i.e., encoded in words) [Nardi, 1993]. They are often ‘pictorial’, providing
icons to represent elements and operations, and use abstract visual notations such as
symbols in flowcharts. Visual programming has been favored by a number of end-user
programming tools for smart homes, with some variations: conventional graphical user
interface, use of metaphors (e.g., programming elements represented as jig-saw puzzles
[Humble, 2003]), and adoption of natural language (e.g., ‘magnetic poetry’ style
representation [Truong, 2004]). This technique offers benefits such as: pictures can
convey meaning in a more concise unit of expression than text, and visual languages can
convey more information about structure than one-dimensional text as using two
dimensions [Myers, 1989]. However, there are several concerns. Visual elements rapidly
overflow the bounds of a screen, and additional elements, for example, lines to show
relations, can make visual clutter [Nardi, 1993].
16
Mobility
While most end-user programming tools for context-aware applications utilize desktop
computers as their usage environments, I develop programming interface to run on
mobile devices, intending to facilitate user’s such creation or modification on the spot
and in the moment, when addressing their needs [De Sá, 2009][Seifert, 2011]. The
smartphone-based user interface allows users to roam within a sensor-instrumented
space while programming their applications. This design decision was inspired by my
observation in the preliminary user studies: participants frequently looked around a
target space to remind them of their daily routines and objects that they use for them.
Likewise, the best moment to catch problematic aspects of the present application is
when a user is using the application.
Contextual information and sensing
The proposed toolkit involves time, object, location, and performance history, as
primary information types for context inference, which are frequently suggested in
existing research on context-ware systems [Dey, 2006]. It proposes activity recognition
by detecting objects’ state or initiation/termination of using them with simple, wireless
state-change sensors. A simple switch attached to an object can often provide strong
hints about activity. Prior work has shown the potential of using multiple simple sensors
for activity detection [Tapia, 2004][Dalal, 2005][Vurgan, 2007]. For example, in the
MARC home [Barger, 2002], a set of simple sensors has been placed in a kitchen to
detect meal preparation activities, which include temperature sensors on stove, cabinet
door sensors, and mat sensors. In addition to functional advantages, it can mitigate
user’s concerns regarding their privacy as it does not collect personally identifiable data
(e.g., facial or body images, voice) [Tapia, 2004].
Media events and actuators
The proposed toolkit provides primarily two types of prompting methods: first, audio
contents via location-based displays (i.e., wireless speakers), and secondly, text
messages via mobile devices. The audio contents include machine speech of user-input
text, and play of user-added/selected sound files. Sound display has several advantages
over visual display. It can catch user’s attention immediately regardless of user’s
orientation, and allow user’s involvement in other work better than video display
[Bridger, 2003]. Sound, especially music, is acknowledged effective in inducing
particular emotional quality. Persuasive power of music has been explored much by
several fields of research (psychology and marketing) [Konecni, 2008][Livingstone,
2007]. Meanwhile, people tend to keep mobile devices near them even within their
homes [Oksman, 2003], and thus text messages accompanied by audio signals can
usually reach them effectively.
3.2 Support for Solution Formulation
In my previous user study where participants created GaLLaG application scenarios for
their behavioral issues, I observed a range of limitations that they have in developing
application ideas, such as having difficulty in clarifying goals that they encounter in
their daily life, thinking of sub- or lower-level goals for a specific goal, and being aware
of current situations surrounding their pursuit. They were less resourceful in developing
sophisticated interactions that include various events to better ensure success in
achieving target goals, or give more fun to experience, and using sensors to detect a
specific context.
17
By using a general HCI design process as a reference, issues that users can have in
designing applications can be thoroughly anticipated, part of which have been already
found as described above. An exemplar HCI design process consists of four steps:
Identifying needs, Establishing requirements, Conceptual design, and Concrete design
[Preece, 2002]. If applying it to design of behavior change solutions with GaLLaG
technology, I assume that users need to achieve the goals shown in Table 1 at each step.
Table 1. Goals in design process
General HCI design process
GaLLaG application design
Step 1. Identifying needs
Step 1. Identifying high-level goals, and low-level goals of them and
contextual aspects relevant to pursuit of them
Step 2. Establishing
Step 2. Figuring out contexts and elements to be utilized to promote
requirements
success in keeping target behaviors, being aware of capabilities of the
present technology
Step 3. Conceptual design
Step 3. Generating application scenarios including specific interactions
Step 4. Concrete design
Step 4. Deciding contextual elements for recognition of target
situations, and choosing media types and contents to deliver
This set of goals implies that the solution formulation phase involves various types of
user proficiency, which range from understanding behavior and contexts, to conceiving
interaction ideas that effectively promote target behaviors, and to media literacy and
skills of manipulating objects.
Approaches
A significant amount of research on how creativity and design processes can be
enhanced has been conducted, and especially, with the growing involvement of
computation and information technology in everyday life, much of the work to develop
tools that support design activity and creativity leveraging such technology has been
done [Fischer, 1993] [Shneiderman, 2000][Landay, 2006]. Drawing on existing
approaches proposed in existing research and based on my user study experience where
participants created their own GaLLaG application scenarios with presence of a
researcher, I identified three approaches to explore, as follows.
(1) Design with understanding of past experience
Generally, any design work starts with gathering data on problem domains and
reflecting on them, and it is considered as an essential part that leads to the following
steps [Preece, 2002]. Importance of understanding situations may be also valid in
people’s designing of one’s own solutions for behavior change. A widely used way to
obtain better idea on one’s daily life is self-tracking. The study by Li, et al. [Li, 2011]
reveals how self-tracking are already integrated in people’s striving for behavior change.
On one hand, people in a so-called ‘Maintenance phase’ utilize self-tracking as a means
to help them maintain awareness of their status relative to a goal, that is, progress in
relation to their goals, and adjust the level or direction of their effort or to adjust their
performance strategies to match what the goal requires [Locke, 2002]. It can also bring
motivational impacts for sustained engagement. Although this aspect is most frequently
highlighted regarding behavioral self-regulation, Li, et al. also found the existence of
‘Discovery phase’, where self-tracking is carried to identify low-level aspects related to
one’s pursuit of high-level goals. Unlike in the Maintenance phase, people, who do not
18
know the goal that they are trying to meet and/or they have not identified the factors that
influenced their striving and attainments, try to figure out such aspects, through selftracking. In this research approach I propose that understanding one’s own behavior and
situations with information collected through self-tracking can facilitate identifying
goals and contextual aspects to tackle, and that this can help figuring out detailed plans
to achieve identified targets.
(2) Design with existing behavior change techniques
Incorporating behavior change techniques that have been validated through scientific
study, will be more likely to enable end-users to develop more effective plans.
Utilization of existing theories on behavior change has been considered as a valuable
approach by multiple studies [Consolvo, 2009][Medynskiy, 2011]. For instance, in the
work by Consolvo, et al., the researchers rely on Locke and Latham’s Goal-setting
theory [Locke, 2002] in developing their design. Medynskiy, et al. highlight existence of
an ongoing search for theoretical foundations and design principles obtained based on
them. I envision that users’ consideration on existing behavior change techniques can be
also beneficial in their design development.
(3) Design with examples
Use of examples is a common method for learning. In work of professional designers, it
is considered as a crucial means for better design outcomes. For instance, one of steps in
Fogg’s design process for creating persuasive technologies is finding relevant examples
as an attempt to learn about elements for successful results [Fogg, 2009]. As one of
approaches to nurture end-user programmers’ ideas and help them gradually gain
expertise, Cao, et al. employed ‘examples’, recognizing proved effectiveness of
provision of examples [Cao, 2011].
4. Tool Development
Previously, I presented the two primary issues I am advancing in the development of a
toolkit for users’ self-experimentation of behavior change solutions: (1) support for
users’ ideation to create rich and meaningful solutions, and (2) provision of a simple but
expressive construction tool. I then described the approaches to tackle these. In this
chapter, I present the tool development that has been motivated by approaches.
4.1 Extended Capability in Construction
The initial version of the programming tool only allowed creation of simple if-then rules
[Lee, 2013]. This design decision was supported by existing research highlighting
benefits of simple if-then rules [Dey, 2006][Truong, 2004]. Through the user study with
the initial tool where participants were asked to build applications that they would like
to have for their behavior change, I also found it easy for participants to learn and
utilize, and versatile (that is, can make simple interactions but considered effective in
helping behavioral issues). However, I encountered interactions in the participantgenerated scenarios that cannot be implemented with simple if-then rules. Although
those interactions might be less desirable at the initial phases of using a tool, e.g., during
a test-driving period when users are less skillful, interactions that go beyond simple ifthen rules seem to be desired as user familiarity increases. Thus, I analyzed the
participant-generated scenarios to find out functionality, currently missing but necessary
to be integrated into the construction tool.
19
Key Patterns in User-Generated Rules
Based on frequency of use and significance with respect to fundamentals in behavior
change, I identified the following types of rules for eliciting a system response:
1.
When an action continues for a specific duration
“If I keep brushing my teeth for 2 minutes, an applause sound plays”;
2.
If another action has or has not occurred for specific duration since an action
occurred
“If I have not washed my hands in 10 minutes after coming home, a ‘water’
sound clip plays”;
3.
If an action has or has not occurred between two absolute times
"If I brushed my teeth at three consecutive nights, my favorite songs play when
I open my chocolate box";
4.
When specific duration passes after an action occurs, a system response is
made
“Two minutes later after an entrance door is closed, I hear music from the
bathroom inviting me to washing hands”.
This analysis suggests that inclusion of temporal relationships is quite desirable, if not
essential, to empowering individuals to create their own behavior change applications.
Existing end-user configuration or programming tools for context-aware applications are
discriminated from each other in terms of their involvement of time-related logic for
conditional rules. For instance, ‘Play bits’ [Humble, 2003] does not involve any
temporal logic. While ‘CAMP’ [Truong, 2004] only provides logic to define time
periods (e.g., Dinner can be defined to happen “in the dining room between 7 P.M. and
9 P.M.”, or “beginning at 7 P.M for 2 hours”), ‘iCAP’ [Dey, 2006] allows richer
expression by further including logic for ordering (e.g., “if Tom walks in after Jane
finished her dinner”). I especially conceived the need to embrace the first three patterns.
It involves functionality to track the degree of performance (e.g., duration, frequency),
and respond according to it, that is, enable users to involve the self-monitoring strategy.
Although I focus on the application of the cues-to-action technique, it is likely that other
behavior change strategies, such as self-tracking, may likely be required in the system.
Second, with Pattern 2 and 3, I learned that users would like to be prompted when a
behavior was missed. For systems of these patterns, the construction tool should include
logic for checking whether an activity is performed in a specified period of time.
Through a field study that I am planning (see Chapter 5 for details on the study), I will
examine appropriateness of the capability proposed currently, and collect additional
design cases from users to see if there are any remaining improvements needed.
4.2 Process for Users’ Creation of Behavior Change Solutions
Integrating the three resources (understanding on the past experience, existing behavior
change techniques, and examples) proposed previously, I established a process to guide
users’ creation of behavior change solutions, through iterative prototyping and testing
[Lee, 2014; Lee, 2015]. I developed a low-fidelity prototype (Microsoft PowerPoint
presentation) that administers the process, by providing educational and related
materials, and inviting participants to complete intermediate tasks to acquire a final
outcome.
Among the three resources proposed as resources, use of behavior change techniques
takes a most significant role. Behavior-change techniques are “observable, replicable,
20
and irreducible component of a [behavioral] intervention designed to alter or….regulate
behavior; that is, a technique is proposed to be an ‘active ingredient’ (e.g., feedback,
self-monitoring, and reinforcement).” [Michie, 2013] I conceived that users would be
likely to develop more effective plans by incorporating behavioral principles that
validated through scientific study. Such techniques have informed many of HCI
designers on how tools for behavior change should work, though only some of these
techniques have been the focus of prior investigation.
As a framework, three generic features of a behavior-change plan were chosen: (1) goalsetting, (2) other techniques (e.g., strategies such as self-rewarding for meeting the goal)
that can support a person to meet a goal, and (3) self-monitoring to determine success
and facilitate iteration on the desired goal and techniques. As shown in Figure 1, the
creation phase consisted of three parts. First, users set specific, actionable goals, through
three steps, ‘Choose an issue’ and ‘Generate behavioral goals’, and ‘Set a SMART
goal’, to help users gradually narrow down a scope to be closer to the concept of
‘SMART’ goals. To facilitate actionable goals, an evidence-based goal-setting strategy,
the SMART (Specific, Measurable, Actionable, Realistic, and Timely) goal concept
[Lathm, 2003] was adopted, which involved a reinterpretation of Locke and Latham’s
goal setting theory [Locke, 2002].
Test
Goal-setting: Specific, actionable goals
Choose an
issue
Generate
behavioral
goals
Set a
SMART goal
Creation of application ideas with behavior change techniques and
examples
Learn
about
techniques
Learn
about
Technique
1
Think of
how to
apply the
techniques
Apply
Technique
1
Learn
about
examples
Examples
matching
Technique
1
Think of
application
ideas
Learn
about
Technique
2
Apply
Technique
2
Formulate
a final
application
idea
Selfmonitoring
Set up
selftracking
tool
Examples
matching
Technique
2
Learn
about
Technique
3
Figure 1. Process for users’ creation of behavior change solutions.
In the second part, users learn about fundamentals of behavior change techniques,
generate ideas on how to apply them to their goal pursuit, and think of systems that can
enhance the ideas previously generated with techniques. For example, a user learns
about a technique, ‘Rewarding yourself’, generates an idea that she allows herself to
have a chocolate bar if she met her workout goal during the week, and think of a system
that sends out a text message saying she should have a bar as her goal has been
accomplished. To provide scaffolding on the selection of various behavior change
techniques, a meta-model of behavior has been developed. Specifically, behavior change
techniques were organized into four domains: Opportunity (availability to engage in a
behavior), Triggers (prompts to perform the behavior), Ability (having the required
skills/attributes to perform the behavior), and Motivation (drive to achieve the
behavior). For the development of the meta-model, two existing meta-models were
combined: Fogg’s behavior model [3], and Michie’s COM-B model [Michie, 2011],
which were initially developed to help professionals create interventions. Fogg’s model
proposes ‘Motivation, Ability, and Triggers’, while Michie et al’s COM-B model
emphasized Capability, Opportunity, and Motivation. Ability and Capability were
collapsed, as they are similar constructs. Triggers and Opportunity were considered to
be related but distinct. As an example of this framework in use, a technique such as
“Script critical actions” is an ability technique whereas “Define your inspiration” is a
motivational technique. Currently, a total of twelve behavior change techniques are
selected and classified as shown in Table 2. Initially, users learn an exemplar technique
21
from each of the domains (underlined in Table 2), and then in the later phases they learn
about the model, self-diagnosis the most problematic domain for them (i.e., is this a
trigger, opportunity, motivation, or ability problem?), and explore other techniques of
the problem domain.
Table 2. Behavior change techniques of each of the four domains
Domain
Techniques
Trigger
Define a trigger; Information or inspiration as triggers; Counteracting negative
emotional triggers
Opportunity
Find the opportune/dangerous time and place; Turn off your “auto-pilot”; Make
it the “default” option
Ability
Script critical actions; Shrink the change; Build habit chains
Motivation
Define your inspiration; Ride the wave; Reward yourself.
Following behavior change techniques, system examples are given to help users ideation
of system ideas, each of which is come up with corresponding to a particular principle.
Once a goal and system solution is set, users set up a way to keep track of their
performance with solutions and related aspects. Users test their solutions pursuing goals
in their daily life with self-tracking, and revise them with the outcome from the test, and
additional support (that is, other behavior change techniques, and system examples).
5 Evaluation
To evaluate the effectiveness of the tool in empowering people to self-create behavior
change solutions, I plan a comparative field study. Considering the effect of different
target issues, the study focuses on a user group who desire to improve their sleep quality
by changing current daily routines.
5.1 Hypothesis and Conditions
To examine if use of the proposed tool can lead to better result in behavior change,
compared to no use of it, I compare two conditions. In one, subjects are given sleep
hygiene education and asked to monitor their performance, and in another, subjects are
given full interventions with the tool in addition to the interventions provided for the
first condition. I examine the underlying assumption that sensor-augmented responsive
environments can better help users’ behavior change, compared with non-technical
support. I compare result of a group that uses applications developed with the proposed
tool, to results from a group that only carries out non-technologically supported
behavioral plans. Subjects of the latter group follow the identical process to make
behavioral solutions, as the former group, but their process does not involve system use.
To summarize, the study has the following three conditions:
•
•
•
Condition 1: Default intervention (Monitoring sleep behavior, Learning sleep
hygiene)
Condition 2: Default intervention, and creation/employment of goals and
behavioral solutions
Condition 3: Default intervention, and creation/employment of goals and
technological solutions
22
Through the comparative analysis of these conditions, I examine the following
hypothesis:
•
•
Use of technological solutions based on behavior change techniques can lead
better performance in improving sleep quality, compared with no use
(Condition 3 versus Condition 1).
Use of technological solutions based on behavior change techniques can lead
better performance in improving sleep quality, compared with no involvement
of technological support (Condition 3 versus Condition 2) but having
behavioral plans.
5.2 Procedure
Participants of all conditions will be involved in up to five one-hour in-person sessions,
over seven weeks. The first and second sessions will be one week apart but the last
sessions are two weeks apart (that is, two weeks between the second and third session,
two weeks between the third and fourth session, and two weeks between the fourth and
fifth session). Through the study period, participants of all conditions are asked to wear
an off-the-shelf sleep sensor, ‘Flex’ (www.fitbit.com), to monitor their sleep depth over
night, and observe their sleep behavior by answering four questions – ‘When you went
to bed’, ‘How long it took to fall asleep’, ‘When you woke up’, and ‘How you are
satisfied with your sleep’-- on a daily basis, using a smartphone application,
‘PACO’(www.pacoapp.com).
Participants are given sleep hygiene education with a two-page handout and invited to
find aspects in it that they would like to apply to their daily life. With these interventions
common to all conditions, participants of Condition 2 are asked to set a specific goal
with regard to their sleep-related issues, and make a plan to help their goal achievement,
based on given lessons about goal setting and other behavior change techniques. They
are asked to work toward their goal sticking to their plan, until they come to the
following session to revise their goal and plan with additional materials. Participants of
Condition 3 set a goal and think of ideas on how to apply behavior change techniques to
their goal pursuit, but they come up with ideas of systems based on the ideas generated
with the behavior change techniques. A system for each participant is developed by the
research and installed at their home, and participants are asked to work toward their goal
using their system, until they come to the following session to revise their goal and
system functionality with additional materials.
5.3 Participants
A total of 60 individuals, over 18 years of age, will participate the study (20 participants
per arm). They should show a fairly high degree of intention to improve sleep quality
and/or sleep habits, to ensure their motivation in carrying out study tasks. They should
have no travel and unusual event plans for seven weeks after beginning of the study, for
stable collection of data on their normal daily life. To do self-tracking with given tools,
participants should currently be smart mobile phone users, and be willing to install
required applications into their phones. Participants should have wireless Internet
connection at home, and be willing to allow required devices to be connected into it. As
well they should be willing to install required devices (sensors, wireless speaker
systems, laptop) at their home. I use a survey asking about such circumstances, in
selecting suitable participants from contacted people. Participants selected based on
survey responses are randomized into the conditions.
23
5.4 Measures
To examine effectiveness of each condition in improving sleep quality, I measure sleep
quality with:
•
•
•
‘Pittsburgh Sleep Quality Index (PSQI)’ [Buysse, 1989], a self-rated
questionnaire
data on the depth of sleep collected with the Fitbit sensor, and
data on sleep behavior self-reported with PACO.
The PSQI generate seven ‘component’ scores (subjective sleep quality, sleep latency,
sleep duration, habitual sleep efficiency, sleep disturbances, use of sleeping medication,
and daytime dysfunction), and one global score by summing them. Participants of all
conditions fill out the questionnaire in every session: In session 1, they report on their
behavior for the past month, and in the following sessions, on their behavior for the past
one or two weeks since their last session. The Fitbit device collects data on sleep depth
accelerometers in it, and participants of all conditions are asked to wear the Fitbit device
all day long through the study. Lastly, I analyze participants’ self-report to the questions
using ‘PACO’. In addition, a semi-structured interview is conducted at the end of the
study to examine participants’ subjective assessment on the progress in their sleep
quality and perceived effectiveness of their solutions in goal achievement, if any.
6. Timeline
2015
Mar
Evaluation
Pilot test
Field test
Apr
May
Jun
Jul
X
X
X
X
Aug
Sep
Oct
Nov
X
X
X
X
Data analysis
X
Final delivery
(thesis, oral defense)
References
Alexander, C. (1964). Notes on the Synthesis of Form Vol. 5. Cambridge: Harvard
University Press.
Andrew, A., Borriello, G., & Fogarty, J. (2007). Toward a Systematic Understanding of
Suggestion Tactics in Persuasive Technologies. In Persuasive Technology (pp. 259–
270).
Arroyo, E., Bonanni, L., & Selker, T. (2005). Waterbot : Exploring Feedback and
Persuasive Techniques at the Sink. In Proceedings of the SIGCHI conference on Human
factors in computing systems (pp. 631–639).
Bandura Albert. (1991). Social Cognitive Theory of Self-Regulation. Organizational
Behavior and Human Performance, 50, 248–287.
Bandura, A. (2001). Social cognitive theory: An agentic perspective. Annual review of
psychology, 52(1), 1-26.
Bang, M., Gustafsson, A., & Katzeff, C. (2007). Promoting new patterns in household
energy consumption with pervasive learning games. Persuasive Technology, 55–63.
24
Baumeister, R. F., Heatherton, Todd, F., & Tice, D. M. (1994). Losing Control: How
and Why People Fail at Self-Regulation. Academic Press.
Baumeister, R. F., Vohs, K. D., Tice, D. M., & Baumeister, F. (2007). The Strength
Model of Self-Control. Psychological Science, 16(6), 351–355.
Blum, B. I. (1996). Beyond Programming. To a New Era of Design. New York: Oxford
University Press.
Brooke, J. (1996). SUS-A quick and dirty usability scale. In P. W. Jordan, B. Thomas,
B. A. Weerdmeester, & A. L. McClelland (Eds.), Usability Evaluation in Industry.
Taylor and Francis.
Buechley, L. (2009). DIY for CHI : Methods , Communities , and Values of Reuse and
Customization. In CHI’09 Extended Abstracts on Human Factors in Computing Systems
(pp. 4823–4826). ACM.
Buysse, D. J., Reynolds, C. F., Monk, T. H., Berman, S. R., & Kupfer, D. J. (1989). The
Pittsburgh Sleep Quality Index: a new instrument for psychiatric practice and research.
Psychiatry research, 28(2), 193-213.
Cao, J., Fleming, S. D., & Burnett, M. (2011). An Exploration of Design Opportunities
for “Gardening” End-User Programmers ’ Ideas, In Visual Languages and HumanCentric Computing (VL/HCC), 2011 IEEE Symposium on (pp. 35-42). IEEE.
Chin, J. S., Callaghan, V., & Clarke, G. (2006). An End-User Programming Paradigm
for Pervasive Computing Applications. In Pervasive Services, 2006 ACS/IEEE
International Conference on (pp. 325–328). IEEE.
Chin, J., Callaghan, V., & Clarke, G. (2010). End-user Customisation of Intelligent
Environments. In H. Nakashima, H. Aghajan, & J. C. Augusto (Eds.), Handbook of
Ambient Intelligence and Smart Environments (pp. 371–407). Boston, MA: Springer.
Consolvo, S., McDonald, D. W., & Landay, J. a. (2009). Theory-driven design strategies
for technologies that support behavior change in everyday life. In Proceedings of the
SIGCHI Conference on Human Factors in Computing Systems (pp. 405-414). ACM.
Consolvo, S., Mcdonald, D. W., Toscos, T., Chen, M. Y., Froehlich, J., Harrison, B., …
Landay, J. A. (2008). Activity Sensing in the Wild : A Field Trial of UbiFit Garden. In
Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (pp.
1797–1806).
Cypher, A., Halbert, D. C., Kurlander, D., Lieberman, H., Maulsby, D., Myers, B. A., &
Turransky, A. (1993). Watch what I do: programming by demonstration. Cambridge,
MA: The MIT Press.
Dalal, S., Alwan, M., Seifrafi, R., Kell, S., & Brown, D. (2005). A Rule-Based
Approach to the Analysis of Elders ’ Activity Data : Detection of Health and Possible
Emergency Conditions. In AAAI Fall 2005 Symposium (EMBC), Vol. 3 (pp. 1–8).
de Ridder, D., & de Wit, J. (Eds.). (2006). Self-regulation in health behavior. John
Wiley & Sons.
De Sá, M., & Carriço, L. (2009). A mobile tool for in-situ prototyping. In Proceedings
of the 11th International Conference on Human-Computer Interaction with Mobile
Devices and Services - MobileHCI ’09, 1.
Dey, A., Sohn, T., Streng, S., & Kodama, J. (2006). iCAP: Interactive prototyping of
context-aware applications. In Pervasive Computing (pp. 254–271).
25
Dow, S., Saponas, T., Li, Y., & Landay, J. (2006). External Representations in
Ubiquitous Computing Design and the Implications for Design Tools. In Proceedings of
the 6th conference on Designing Interactive systems (pp. 241-250). ACM.
Fischer, G. (1993). Beyond Human Computer Interaction: Designing Useful and Usable
Computational Environments. In People and Computers Viii: Proceedings of the
HCI ’93 Conference (pp. 17–31). Cambridge, UK: Cambridge University Press.
Fischer, G. (2007). Meta-Design : Expanding Boundaries and Redistributing Control in
Design. In Human-Computer Interaction–INTERACT 2007 (pp. 193–206). Springer
Berlin Heidelberg.
Fischer, G., & Giaccardi, E. (2006). Meta-Design : A Framework for the Future of
End-User Development. End User Development (pp. 427–457). Springer Netherlands.
Fischer, G., & Scharff, E. (2000). Meta-Design — Design for Designers. In Proceedings
of the 3rd conference on Designing interactive systems: processes, practices, methods,
and techniques (pp. 1–10). ACM.
Fogg, B. (2009). Creating Persuasive Technologies : An Eight-Step Design Process. In
Persuasive ’09 (pp. 44:1–44:6).
Fogg, B. (2012). Three Tiny Habits. Retrieved from tinyhabits.com
Fogg, B. J. (2002). Persuasive Technology: Using Computers to Change What We Think
and Do. Morgan Kaufmann.
García-herranz, M., Haya, P., & Alamán, X. (2010). Towards a Ubiquitous End – User
Programming System for Smart Spaces. J. UCS, 16(12), 1633–1649.
Gollwitzer, P. M., & Sheeran, P. (2006). Implementation intentions and goal
achievement: A meta‐analysis of effects and processes. Advances in experimental
social psychology, 38, 69–119.
Greenbaum, J. M., & Kyng, M. (Eds.). (1991). Design at work: Cooperative design of
computer systems. Routledge.
Gross, P. A., Herstand, M. S., Hodges, J. W., & Kelleher, C. L. (2010). A code reuse
interface for non-programmer middle school students. In Proceedings of the 15th
international conference on Intelligent user interfaces (pp. 219-228). ACM.
Haidt, J. (2001). The emotional dog and its rational tail: A social intuitionist approach to
moral judgment. Psychological review, 108, 814–834.
Hekler, E. B., Burleson, W., & Lee, J. (2013). A DIY self-experimentation toolkit for
behavior change. In Published & Presented via the Personal Informatics in the Wild:
Hacking Habits for Health & Happiness at the ACM-CHI Conference.
http://personalinformatics.org/chi2013/hekler.
Hekler, E. B., Klasnja, P., Froehlich, J. E., & Buman, M. P. (2013). Mind the
Theoretical Gap : Interpreting , Using , and Developing Behavioral Theory in HCI
Research. In Proceedings of the SIGCHI Conference on Human Factors in Computing
Systems (pp. 3307–3316).
Humble, J., & Crabtree, A. (2003). “Playing with the Bits” User-configuration of
Ubiquitous Domestic Environments. In UbiComp 2003: Ubiquitous Computing (pp.
256–263).
26
Intille, S. S. (2006). The goal: smart people, not smart homes. In Proceedings of
ICOST2006: The International Conference on Smart Homes and Health Telematics (pp.
3–6).
Karoly, P. (1993). Mechanisms of self-regulation: A systems view. Annual Review of
Psychology, 44, 23–52.
Klasnja, P., Consolvo, S., Pratt, W., Informatics, H., & Seattle, I. L. (2011). How to
Evaluate Technologies for Health Behavior Change in HCI Research. In Proceedings of
the SIGCHI Conference on Human Factors in Computing Systems (pp. 3063–3072).
Ko, A. J., Myers, B. A., & Aung, H. H. (2004). Six learning barriers in end-user
programming systems. In Visual Languages and Human Centric Computing, 2004 IEEE
Symposium on. IEEE.
Konečni, V. J. (2008). Does music induce emotion? A theoretical and methodological
analysis. Psychology of Aesthetics, Creativity, and the Arts, 2(2), 115.
Latham, G. P. (2003). Goal Setting: A Five-Step Approach to Behavior Change.
Organiz Dyn, 32, 309- 318.
Locke, E.A. & Latham, G.P. (2002). Building a Practically Useful Theory of Goal
Setting and Task Motivation: A 35- Year Odyssey. Am Psychol, 57, 9, 705-17.
Lee, J., Garduño, L., Walker, E., & Burleson, W. (2013). A tangible programming tool
for creation of context-aware applications. In Proceedings of the 2013 ACM
international joint conference on Pervasive and ubiquitous computing (pp. 391–400).
ACM.
Lee, J., Walker, E., Burleson, W., and Hekler, E. B. (2014a). Programming tool of
context-aware applications for behavior change. In Proceedings of the 2014 ACM
International Joint Conference on Pervasive and Ubiquitous Computing: Adjunct
Publication (UbiComp '14 Adjunct) (pp. 91-94). ACM.
Lee, J., Walker, E., Burleson, W., and Hekler, E. B. (2014b). Exploring users' creation
of personalized behavioral plans. In Proceedings of the 2014 ACM International Joint
Conference on Pervasive and Ubiquitous Computing: Adjunct Publication (UbiComp
'14 Adjunct) (pp.703-706). ACM.
Lee, J., Walker, E., Burleson, W., Hekler, E.B. (2015). Understanding Users Creation of
Behavior Change Plans with Theory-Based Support. CHI'13 Extended Abstracts on
Human Factors in Computing Systems. ACM.
Li, I. (2011). Understanding My Data , Myself : Supporting Self-Reflection with
Ubicomp Technologies. In Proceedings of the 13th international conference on
Ubiquitous computing (pp. 405-414). ACM.
Lieberman, H., Paternò, F., Klann, M., & Wulf, V. (2006). End-User Development: An
Emerging Paradigm. In H. Lieberman, F. Paternò, M. Klann, & V. Wulf (Eds.), EndUser Development (vol. 9) (pp. 1–8). Springer.
Locke, E. a., & Latham, G. P. (2002). Building a practically useful theory of goal setting
and task motivation: A 35-year odyssey. American Psychologist, 57(9), 705–717.
Maeda, J. (2006). The Laws of Simplicity. The MIT Press.
Maes, S., & Gebhardt, W. (2005). Self-Regulation and Health Behavior: The Health
Behavior Goal Model. In M. Boekaerts, P. R. Pintrich, & M. Zeidner (Eds.), Selfregulation in health behavior (p. 343). Academic Press.
27
Maitland, J., & Chalmers, M. (2010). Self-monitoring, self-awareness, and selfdetermination in cardiac rehabilitation. In Proceedings of the SIGCHI Conference on
Human Factors in Computing Systems, 1213–1222.
Medynskiy, Y., Yarosh, S., & Mynatt, E. (2011). Five strategies for supporting healthy
behavior change. In CHI’11 Extended Abstracts on Human Factors in Computing
Systems (pp. 1333–1338).
Michie, S., Richardson, M., Johnston, M., Abraham, C., Francis, J., Hardeman, W., …
Wood, C. E. (2013). The Behavior Change Technique Taxonomy (v1) of 93
Hierarchically Clustered Techniques: Building an International Consensus for the
Reporting of Behavior Change Interventions. Annals of behavioral medicine : a
publication of the Society of Behavioral Medicine, 46(1), 81–95.
Michie, S., van Stralen, M. M., & West, R. (2011). The Behaviour Change Wheel: A
New Method for Characterising and Designing Behaviour Change Interventions.
Implement Science, 6 (1), 42.
Muraven, M., & Baumeister, R. F. (2000). Self-regulation and depletion of limited
resources: does self-control resemble a muscle? Psychological bulletin, 126(2), 247–59.
Myers, B. A. (1990). Taxonomies of visual programming and program visualization.
Journal of Visual Languages & Computing, 1(1), 97–123.
Nardi, B. A. (1993). A Small Matter of Programming: Perspectives on End User
Programming. Cambridge, MA: MIT Press.
Nawyn, J., Intille, S., & Larson, K. (2006). Embedding behavior modification strategies
into a consumer electronic device: a case study. In UbiComp 2006: Ubiquitous
Computing (pp. 297–314). Springer Berlin Heidelberg.
Nawyn, J., Intille, S., & Larson, K. (2006). Embedding behavior modification strategies
into a consumer electronic device: a case study. UbiComp 2006: Ubiquitous Computing,
297–314.
Nelson, M. E., Rejeski, W. J., Blair, S. N., Duncan, P. W., Judge, J. O., King, A. C., …
Castaneda-Sceppa, C. (2007). Physical activity and public health in older adults:
recommendation from the American College of Sports Medicine and the American
Heart Association. Medicine and science in sports and exercise, 39(8), 1435–45.2
Oksman, V., & Rautiainen, P. (2003). ” Perhaps it is a Body Part”: How the Mobile
Phone Became an Organic Part of the Everyday Lives of Finnish Children and
Teenagers. Machines that become us: The social context of communication technology.
Machines that become us: The social context of communication technology, 293-308.
Parks-Stamm, E. J., Gollwitzer, P. M., & Oettingen, G. (2007). Action control by
implementation intentions: Effective cue detection and efficient response initiation.
Social Cognition, 25(2), 248-266.
Pavel, D., Collaghan, V., & Dey, A. K. (2010). Looking Back in Wonder: How selfmonitoring technologies can help us better understand ourselves. In Proc. 2010 Sixth
International Conference on Intelligent Environments (pp. 289–294).
Preece, J., Rogers, Y., & Sharp, H. (2002). Interaction Design: Beyond HumanComputer Interaction. Crawfordsville: John Wiley and Sons, Inc.
Räisänen, T., Oinas-kukkonen, H., & Pahnila, S. (2008). Finding Kairos in Quitting
Smoking : Smokers ’ Perceptions of Warning Pictures. In Persuasive Technology (pp.
254–257). Springer Berlin Heidelberg.
28
Repenning, A., & Ambach, J. (1996). Tactile programming: A unified manipulation
paradigm supporting program comprehension, composition and sharing. In Proceedings
of the IEEE Symposium on Visual Languages (VL ’96) (pp. 102–109). IEEE Computer
Society Press.
Resnick, M., Myers, B., Nakakoji, K., Shneiderman, B., Pausch, R., Selker, T., &
Eisenberg, M. (2005). Design Principles for Tools to Support Creative Thinking.
Institute for Software Research, Paper816.
Rittel, H. W., & Webber, M. M. (1973). Dilemmas in a general theory of planning.
Policy Sciences, 4(2), 155–169.
Schön, D. A. (1983). The reflective practitioner: how professionals think in action. New
York: Basic Books.
Seifert, J.(2011). Mobidev: a tool for creating apps on mobile phones. Proceedings of
the 13th International Conference on Human Computer Interaction with Mobile Devices
and Services, 109–112. ACM.
Shneiderman, B. (2000). Creating creativity: user interfaces for supporting innovation.
ACM Transactions on Computer-Human Interaction, 7(1), 114–138.
Ståhl, A., Höök, K., Svensson, M., Taylor, A. S., & Combetto, M. (2008). Experiencing
the Affective Diary. Personal and Ubiquitous Computing, 13(5), 365–378.
Stiles, W. B., Reynolds, S., Hardy, G. E., Rees, A., Barkham, M., & Shapiro, D. A.
(1994). Evaluation and description of psychotherapy sessions by clients using the
Session Evaluation Questionnaire and the Session Impacts Scale. Journal of Counseling
Psychology, 41, 175–185.
Syrjänen, A., & Kuutti, K. (2011). From Technology to Domain : The Context of Work
for End-User Development. In iConference ’11 (pp. 244–251). Seattle, Washington.
Tapia, E. M., Intille, S. S., & Larson, K. (2004). Activity Recognition in the Home Using
Simple and Ubiquitous Sensors. Springer Berlin Heidelberg.
Thomke, S., Hippel, E. Von, & Franke, R. (1998). Modes of experimentation : an
innovation process and competitive variable. Research Policy, 27(3), 315-332.
Truong, K., Huang, E., & Abowd, G. D. (2004). CAMP: A Magnetic Poetry Interface
for End-User Programming of Capture Applications for the Home. UbiComp 2004:
Ubiquitous, 143–160.
Von Hippel, E., & Katz, R. (2002). Shifting Innovation to Users via Toolkits.
Management Science, 48(7), 821–833.
Von Hippel, E., & Tyre, M. (1995). How “Learning by Doing” is Done: Problem
Identification in Novel Process Equipment. Research Policy, 24(1), 1–12.
Vurgun, S., & Pavel, M. (2007). A statistical reasoning system for medication
prompting. In UbiComp 2007: Ubiquitous Computing (pp. 1–18). Springer Berlin
Heidelberg.
Wakkary, R. (2009, September). Anything Is a Fridge : The Implications of Everyday
Designers. Interactions, 12–17.
Wakkary, R., & Maestri, L. (2007). The Resourcefulness of Everyday Design. In
Proceedings of the 6th ACM SIGCHI conference on Creativity & cognition (pp. 163–
172). ACM.
29
Wallston, K. A., & Wallston, Barbara Strudler DeVellis, R. (1978). Development of the
multidimensional health locus of control (MHLC) scales. Health Education & Behavior,
6(1), 160–170.
Webb, T. L., & Sheeran, P. (2004). Identifying good opportunities to act:
Implementation intentions and cue discrimination. European Journal of Social
Psychology, 34(4), 407-419.
Webb, T. L., & Sheeran, P. (2007). How do implementation intentions promote goal
attainment? A test of component processes. Journal of Experimental Social Psychology,
43(2), 295-302.
Webb, T. L., & Sheeran, P. (2008). Mechanisms of implementation intention effects:
The role of goal intentions, self‐efficacy, and accessibility of plan components. British
Journal of Social Psychology, 47(3), 373-395.
Wood, W., & Neal, D. T. (2007). A new look at habits and the habit-goal interface.
Psychological review, 114(4), 843–63.
Zimmerman, B. J. (2002). Becoming Learner : Self-Regulated Overview. Theory into
Practice, 41(2), 64–70.
30