I gave the closing keynote speech at the ACM International Conference on Information Technology for Social Good (GoodIT). Interesting conference I was not aware of, on a topic close to heart. A lot of interesting work with a strong focus on sensing. I hope to see this community grow and strengthen its ties with the Social Sciences and Design.
The Behavior Change Design cards are now available at persuasive.ac.cy/bcdcards. For more information on their creation and evaluation, you may read the full paper at the link below:
- Konstanti, C., Karapanos, E., & Markopoulos, P. (2021). The Behavior Change Design Cards: A Design Support Tool for Theoretically-Grounded Design of Behavior Change Technologies. International Journal of Human–Computer Interaction, 1-17.
Evangelos will be giving a keynote speech at Persuasive Technology 2022 that takes place on March 29-31, 2022. The conference will be virtual this year due to COVID-19. Registration is open and free.
Our interactions article summarised our recent work on nudging: the ’23 ways to nudge’ framework and the Nudge Deck. Read the article here.
The Nudge Deck is now available to download from the following link: http://persuasive.cut.ac.cy/nudgedeck
We’re joining the upcoming panel on “Teaching a Persuasive Technology course” that takes place as part of the Persuasive Technology 2021 conference. We’re looking forward to the much needed discussion!
Panel session: Teaching a Persuasive Technology Course
Moderator: Dena Al-Thani, Hamad Bin Khalifa University, Qatar
- Harri Oinas-Kukkonen, Oulu University, Finland
- Evangelos Karapanos, Cyprus University of Technology, Cyprus
- Lisette van Gemert-Pijnen, University of Twente, the Netherlands
- Raian Ali, Hamad Bin Khalifa University, Qatar
On March 9, I gave a keynote at the closing event of the project ‘Mobile monitoring of movement and joint loading in persons with degenerative hip- and knee problems’, at the University of Hasselt. The talk’s slides can be found here.
My flight was affected by the Ciara storm, but I ended up giving the talk remotely, as part of a two-day workshop on “Considering Health Behavior Change” that took place back in February, in Eindhoven, NL. It went surprisingly well according to comments, and turned out to be an early preview of what was coming. Slides can be found here.
We’ve just came back from a productive sabbatical. Chrysanthi spend two months at the Telefonica research labs in Barcelona, supported by the Encase project, Loukas spend his semester at the Jheronimus Academy of Data Science in Tilburg, NL, and I visited the Interaction Design Lab of the University of Melbourne. We had lots of fun and have some interesting stuff on the pipeline…
One of our very own, Rúben Gouveia, successfully defended his Ph.D. thesis entitled “Tracking in the wild: Exploring the everyday use of physical activity trackers”. Rúben is moving to the University of Twente as an Assistant Professor as of Sept 1st, 2019. Rúben! it has been more than a pleasure!
We had the pleasure of hosting between February and April 2019, Elçin Hancı – Ph.D. candidate at the Human-Technology Interaction group of the Eindhoven University of Technology. Elcin’s working hypothesis, in her PhD studies, is that we treat physical activity trackers as social actors. Her work involves experimental studies of social effects on participants’ perceptions of the tracker, and motivation for physical activity (see her recent publication at Persuasive’19). We had a lot of interesting discussions and worked towards a joint publication. The photo depicts Elçin and the rest of the lab – Chrysanthi, Georgia and Loukas (from right to left) – as happy student volunteers at the Persuasive’19 conference that we hosted here in Limassol. Elçin, it was a pleasure!
We recently published an article summarizing our work and our vision for a methodological paradigm we call Technology-Assisted Reconstruction. You may be asking: what is Technology-Assisted Reconstruction?
Suppose you want to measure how people feel, or what people do throughout the day. In Human-Computer Interaction (HCI), you may be interested to study how a novel communication system affects feelings of connectedness among family members, or how a self-tracking technology affects users’ motivation for, say, physical exercise. How do you go about it?
One approach is to ask people at the end of the day to recall episodes where they interacted with the system and recount how they felt and what they did during those episodes, or to ask them to provide “global reports” on how they system affects their feelings of connectedness or their motivation for physical activity, overall. Such retrospective judgments suffer from recall and social biases.
Another alternative is to employ the Experience Sampling Method (ESM), which consists of prompting the user with short questionnaires, at random, or otherwise-defined times throughout the day. With the widespread adoption of sensor-filled mobile technologies, we can now schedule such prompts rather effectively. For instance, we can ask people to report how motivated they feel, right after they interact with the self-tracking technology, thus being able to study the proximal impact of feedback on individuals’ motivation. ESM minimizes recall and social biases and is considered to be the gold standard of momentary assessment.
One challenge with ESM, is that it imposes a high burden on participants due to the repetitive interruptions. Daniel Kahnneman and colleagues published an article in Science in 2004 on an alternative to ESM, the Day Reconstruction Method (DRM). DRM does not interrupt participants throughout the day but rather asks participants, at the beginning of the next day, to list their daily activities as a continuous series of episodes. This is assumed to minimize recall bias through increasing the number of cues to episodic memory people have access to, and which they use to infer experiential information. DRM has been found to provide a reasonably good approximation to experience sampling data and the method has been well adopted by the HCI community.
With Technology-Assisted Reconstruction we argue that our field has the capacity to contribute towards a next step in the field of momentary assessment, where passively logged data from sensor-based technologies, such as the locations we visit, the social interaction we perform through technology, and what we see, as captured from lifelogging cameras, can be leveraged to support recall. This article attempts to make three contributions:
- It reviews the use of ESM and DRM in HCI literature and discusses the challenges researchers have encountered.
- It presents our vision on Technology-Assisted Reconstruction and illustrates this, through five different methods we have developed over the past ten years.
- It proposes a framework for the development of TAR methods (see table 2), which summarizes three distinct strategies: (a) enhancing the encoding process as the event takes place, (b) guiding the recall process, and (c) providing external memory cues during recall.
|Karapanos, E. (2019): Technology-assisted reconstruction: a new alternative to the experience sampling method, Behaviour & Information Technology, DOI: 10.1080/0144929X.2019.1608303|
|Caraban, A., Karapanos, E., Gonçalves, D. & Campos, P. (2018) 23 Ways to Nudge: A Review of Technology-Mediated Nudging in Human-Computer Interaction, In Proceedings of CHI’19.|
Ruben Gouveia has been selected as a finalist for the Gaetano Borriello Outstanding Student Award at Ubicomp 2018 along with four other students from Cornell University, the University of Washington and Stuttgart University. The award was given to Alex Mariakakis from the University of Washington. More information at: http://ubicomp.org/sc/awards.html
Ruben is finishing his PhD studies at University of Madeira, under the supervision of Dr. Evangelos Karapanos. Ruben’s work has tried to understand how users engage with physical activity trackers and to design appropriate strategies that increase user engagement and lead to behavior change. More information about Ruben’s work can be found at https://rubengouveia.com/
We’re proudly hosting Persuasive’19 in Limassol, from April 9 to 11, 2019. Persuasive Technology (PT) is a vibrant interdisciplinary research community, focusing on the design, development and evaluation of interactive technologies aimed at changing people’s attitudes or behaviors through persuasion and social influence, but not through coercion or deception.
Dr. Tuomas Kari visited our lab during the period 3-7.9.2018. His background is in information systems science, and his main research interest lays in the use of technology in everyday life, especially in the context of health and wellness. During the visit, Tuomas gave a talk on exertion games and gamification, and we talked about our mutual research interests, and explored opportunities for collaboration.
The official kick-off of the H2020 project Co-Inform took place in May’18. Co-Inform is about empowering citizens, journalists and policy-makers with co-created socio-technical solutions, to increase resilience to misinformation and to generate more informed behaviours and policies. The aim of Co-Inform is to co-create this solutions with citizens, journalists and policy-makers for:
- Detecting and combating a variety of misinforming posts and articles on social media;
- Supporting, persuading and nourishing misinformation-resilient behaviour;
- Bridging between the public on social media, external fact-checking journalists, and policy-makers;
- Understanding and predicting which misinforming news and contents are likely to spread across which parts of the network and demographic sectors;
- Infiltrating echo-chambers on social media to expose confirmation-biased networks to different perceptions and corrective information; and
- Providing policy-makers with advanced misinformation analysis to support their policy-making processes and validation.
Our work will explore the design of technological interventions that nudge individuals towards misinformation-resilient behaviours. CUT is further represented by Dr. Elena Kyza (PI) and Dr. Dionysis Panos.
While recent research has emphasized the importance of understanding the lived experience of personal tracking, very little is known about the everyday coordination between tracker use and the surrounding environment. We combine behavioral data from trackers with video recordings from wearable cameras, in an attempt to understand how usage unfolds in daily life and how it is shaped by the context of use. We recorded twelve participants’ daily use of activity trackers, collecting and analyzing 244 incidents where activity trackers were used. Among our findings, tracker use was strongly driven by reflection and learning-in action, contrasting the traditional view that learning is one of deep exploration, following the collection of data on behaviors. We leverage on these insights and propose three directions for the design of activity trackers: facilitating learning through glances, providing normative feedback and facilitating micro-plans.
|Gouveia, R., Karapanos, E., Hassenzahl, M. (2018) Activity Tracking in vivo, In Proceedings of CHI’18.|
We had the pleasure of welcoming two research visitors over the past months.
Mr Olli Korhonen, PhD student at the Interact Research Unit of the University of Oulu, advised by Prof. Minna Isomursu, visited our lab between January and February 2018 (16.1-13.2) as part of a short-term scientific mission in the context of the European Network for the Joint Evaluation of Connected Health Technologies. Olli’s work focuses on the personalisation of e-health services. The scope of Olli’s visit was to inquire into the nature of personalisation in the context of health behavior change technologies, across three domains of activity of the lab – gamification, goal-setting and nudging.
Our own Rúben Gouveia, who is completing his PhD studies at the University of Madeira and Madeira Interactive Technologies Institute, visited our lab between October 2017 and February 2018. During his time here, Rúben designed and conducted a field trial of a novel smartwatch application for the promotion of physical activity. He is currently analyzing the data and planning the write-up of what is expected to be the last study in his thesis. In the meantime, Ruben also got a full paper accepted at the top-tier ACM CHI conference.
We’re involved in the organisation of the Persuasive’18 conference to be held in Waterloo, Canada. Deadline for papers: November 1, 2017. Submit!
PERSUASIVE 2018 (PT-18)
The 13th International Conference on Persuasive Technology
April 16-19, 2018, Waterloo, Canada
Paper submission deadline: November 1, 2017
Administration: Krystina Bednarowski
We will be co-organizing a workshop on New Directions for Self-Quantification for Behavior Change as part of Mobile HCI 2017 in Vienna. Submit your paper and join us!
New Directions in Self-Quantification for Behavior Change – Call for Papers
Workshop at MobileHCI
September 4th, 2017
Elisabeth Kersten – van Dijk, Eindhoven University of Technology
Naomi Jacobs, Eindhoven University of Technology
Heleen Rutjes, Eindhoven University of Technology
Marc Hassenzahl, University of Siegen
Evangelos Karapanos, Cyprus University of Technology
Boris de Ruyter, Philips Research, Radboud University Nijmegen
Wijnand IJsselsteijn, Eindhoven University of Technology
The vision of the Quantified Self proposes that by gathering data about ourselves, we can gain insight into what makes us tick and shed light on avenues for self-improvement. However, examples from practice show that obtaining insight from data, let alone achieving long-term behavior change, is not trivial. In addition, self-tracking may have unintended side effects and hidden assumptions which complicate its effects. In this interactive workshop, we aim to bring together researchers from various fields with an interest in behavior change support systems, to compile an interdisciplinary, critical perspective on the use of self-quantification for behavior change.
Perspectives that are of interest to the workshop include, but are not limited to, the following:
– From data to action: How and to what extent do users understand and believe their data, and does it lead to new, actionable and meaningful insights? How can we help users translate their data into action?
– Unintended effects: What are the side effects of self-quantification and how can we investigate these?
– Individual differences: What individual differences change the effects of self-quantification, both stable (e.g., personality, socio-cultural background, gender) and temporary characteristics (e.g., mood, situational characteristics)?
– Ethical perspectives: how should we incorporate ethical considerations in the design of applications employing self-quantification for behavior change?
– Methodological perspectives: Which methods can be applied to answer the questions above?
** To participate **
Submit a position paper of 2-4 pages in SIGCHI extended abstracts format. Research results are not required; we invite specifically ideas and research in early stages.
Submissions will be reviewed by the organizers based on relevance and expected contribution to the discussion in the workshop.
Please direct queries and paper submissions to email@example.com.
– The deadline for submission of workshop contributions is May 19th, 2017
– Notification will be sent to authors on June 9th, 2017
Please note that at least one author of each accepted position paper must register and attend the workshop. All workshop participants must register for both the workshop and for at least one day of the main conference. Please register before the early registration deadline of June 19th.
We received two awards at the 12th International conference on Persuasive Technology. The best paper award went to Ana’s work on Subly, while Ligia, just on the very start of her PhD, won the best poster award. Well done girls!
With 50% of people spending over 6 h per day surfing the web, web browsers offer a promising platform for the delivery of behavior change interventions. One technique might be subliminal priming of behavioral concepts (e.g., walking). This paper presents Subly, an open-source plugin for Google’s Chrome browser that primes behavioral concepts through slight emphasis on words and phrases as people browse the Internet. Such priming interventions might be employed across several domains, such as breaking sedentary activity, promoting safe use of the Internet among minors, promoting civil discourse and breaking undesirable online habits such excessive use of social media. We present two studies with Subly: one that identifies the threshold of subliminal perception and one that demonstrates the efficacy of Subly in a picture-selection task. We conclude with opportunities and ethical considerations arising from the future use of Subly to achieve behavior change.
|Caraban, A., Karapanos, E.,Teixeira, V., Munson, S., Campos, P. (2017) On the Design of Subly: Instilling Behavior Change During Web Surfing Through Subliminal Priming, In Proceedings of Persuasive’17.|
The use of motivational textual messages is a popular strategy for encouraging behavior change. Social media platforms, such as Instagram, have afforded tailored visual presentations of the text of such messages, that range in style, by exploring color, illustration, photography, or typography. Following the fact that different text aesthetics can invoke different perceptions about what is written, we suggest that perceived beauty coming from the presentation of textual messages can increase the appeal of changing behavior. Beauty should increase extrinsic and intrinsic motivation, as long as beauty elicits positive affect. In intrinsic, beauty should lead to a relatively enduring behavioral intention, whereas in extrinsic it should increase the intention temporarily. We present our initial hypotheses and an experimental study we plan to conduct.
|Duro, L., Karapanos, E.,Romão, T., Campos, P. (2017) Does Beauty Matter in Behavior Change?, In Adjunct Proceedings of Persuasive’17.|
Recent research reveals over 70% of the usage of physical activity trackers to be driven by glances – brief, 5-second sessions where individuals check ongoing activity levels with no further interaction. This raises a question as to how to best design glanceable behavioral feedback. We first set out to explore the design space of glanceable feedback in physical activity trackers, which resulted in 21 unique concepts and 6 design qualities: being abstract, integrating with existing activities, supporting comparisons to targets and norms, being actionable, having the capacity to lead to checking habits and to act as a proxy to further engagement. Second, we prototyped four of the concepts and deployed them in the wild to better understand how different types of glanceable behavioral feedback affect user engagement and physical activity. We found significant differences among the prototypes, all in all, highlighting the surprisingly strong effect glanceable feedback has on individuals’ behaviors.
|Gouveia, R., Pereira, F., Karapanos, E., Munson, S., Hassenzahl, M. (2016) Exploring the Design Space of Glanceable Feedback for Physical Activity Trackers, In Proceedings of Ubicomp’16.|
I intentionally chose a provocative title. Can we really design behaviour? Can technologies have such an influence on individuals’ behaviour that we can talk about “designing behaviour” through technology?
To a large extend, behaviour change technologies nowadays rest on the assumption that knowledge leads to behaviour change, an idea that is deeply rooted in the “quantified self” community. We assume, for instance, that making people aware of how much, when, and where, they walk (or don’t), will lead them to uncover patterns in their behaviours and take actions to increase their physical activity. Our research has revealed that this is not necessarily true, and that we often hold incorrect assumptions about how people use these technologies. For instance, our habito study, revealed that people rarely look back at their past performance data, and that only 30% of the users set their own walking goal.
Rather than thinking of such technologies as ones that enable deep reflection over one’s behavioural patterns, could we have more success if we focus on shaping the choice architecture at moments of decision making? Consider, for instance, the popular road sign stripes that are strategically positioned closer and closer together as we approach a steep curve, making us believe that we are over-speeding. Such a simple intervention has been found to decrease car accidents by 36%.
How could we transfer such approaches to the design of behaviour change technologies? I personally see two approaches. One is to infer when such critical moments of decision making happen – approaching the elevator, browsing through online food delivery options etc. A second is to provide persistent, glanceable feedback that keeps reminding people of their goal. Consider, for instance, that people glance at their smartwatches 80-150 times a day to check the time or incoming notifications. Might we leverage on this to present physical activity feedback at the periphery of their attention?
In our recent paper we explored just that. For the purpose of this blog article, I wanted to emphasize what we found with one of the prototypes we developed, TickTock.
Most activity trackers will tell you how much have you walked so far in the course of the day. This, of course, is routed in the idea that people set daily walking goals, and feedback serves to make them aware of their performance so they can meet their goal.
With TickTock, we tried to change the logic. TickTock will only tell you how active you were over the past hour. As a result, TickTock emphasises keeping a balance in your day, i.e. avoiding prolonged periods of sedentarism, which has been found to be a health risk factor independently of the amount of physical activity one performs over the course of a day.
What we found was rather astonishing, I think. I initially thought there must some error in the data. I asked Ruben to scrutinise the data collection script. He couldn’t find any error.
I turns out, TickTock had a profound impact on individual’s behaviours. We performed a linear regression analysis to predict the time people took till their next walk after checking the watch, based on the feedback they received, namely how active (o-60 mins) they were over the past hour. We found this single variable to account for 76% of the variance in the data! For every additional 10 min of physical activity that the participants saw they performed over the past hour, they would take an extra 9.5 min till their next walk. Participants who saw that they walked 10 or less min over the past hour had a 77% probability of starting a new walk in the next 5 min.
What do those findings suggest? First, that these technologies are far from neutral, they emphasise certain ways of living, and as designers, we should pay close attention to the assumptions embedded in our designs and to align feedback with the goal we want to achieve. Second, they highlight they surprising impact glanceable feedback has on individuals behaviours. The habito study highlighted that the dominant mode of use of physical activity trackers is glancing and it serves to support self-regulation of immediate behaviour. This study further demonstrated the power of glanceable feedback in shaping individual’s behaviours.
|Gouveia, R., Pereira, F., Karapanos, E., Munson, S., Hassenzahl, M. (2016) Exploring the Design Space of Glanceable Feedback for Physical Activity Trackers, In Proceedings of Ubicomp’16.|
In her recently published work, Jordan Etkin asked individuals to count their steps over the course of a day. Not surprisingly, she found an increase in the number of steps taken as compared to the control condition, when individuals weren’t asked to count their steps. What was more interesting, was that step counting led to a reduction in people’s enjoyment of walking. Etkin argued that by invoking the metaphor of “measuring” and highlighting quantitative outcomes, attention is drawn away from the intrinsic joys of an activity towards external rewards (see Deci et al. 1999). Exercise is experienced a little more like work, which in turn may decrease the likelihood of continued engagement in one’s free time (as found in Etkin’s study).
In other words, instead of supporting people with construing exercise as an enjoyable and meaningful activity to guarantee prolonged engagement, activity trackers might establish mechanisms, which guarantee a short-term increase in physical activity at the risk of potentially detrimental long-term effects.
Of’course, this might be a rather narrow vision of these technologies. No doubt, self-quantification is the dominant narrative in marketing, and often in design. But richer narratives may be found if one looks at the ways in which users appropriate such technologies.
To inquire into this, we surveyed recent, memorable experiences people had with activity trackers, and tried to understand these in terms of need fulfilment, using a theoretical and methodological framework proposed by Sheldon et al., 2001. A factor analysis revealed a two-dimensional structure of users’ experience driven by the needs of physical thriving or relatedness. More than just supporting behavioral change, we found trackers to provide multiple psychological benefits, such as enhancing feelings of autonomy as people gained more control about their exercising regime, or experiencing relatedness, when family members purchased a tracker for relatives and joined them in their efforts towards a better, healthier self. Interestingly, we found that while numerical feedback lost its relevance over time, users continued to wear the tracker for a number of reasons, ranging from the perceived future value of data accumulation, to the mere symbolic empowerment that users felt when wearing the tool.
With multiple studies highlighting high abandonment rates of physical activity trackers, a question is raised as to how well they perform for different individuals. One meaningful framework to characterise diversity in populations is Prochanska’s et al. Transtheoretical model, also known as the stages of behaviour change . TTM suggest five stages that people will typically go through when planning and implementing a behaviour change, such as quitting smoking, or increasing physical activity. These are: precontemplation (not ready), contemplation (getting ready), preparation, action and maintenance.
In our Habito study, we found the tracker to work best for people that are in the intermediary stages of behaviour change. Individuals in the contemplation and preparation stages, who have the intention but not yet the means (i.e. motivation, strategies) to change, had an adoption rate of 56% (with adoption being defined as use that extends beyond the first two weeks), whereas individuals in precontemplation, action or maintenance stages had an adoption rate of only 20%.
Yet, these individuals (in the intermediary stages of behavior change) are only about 43% of the population that are likely to purchase an activity tracker, or download an app on their smartphones (based on our sample). So, there is a significant population of users for whom we currently fail to address their needs.
To remediate this situation, we need to ask new questions, such as, how can trackers instill initial motivation for behavior change rather than merely supporting the process of it? Individuals in the precontemplation stage are often unaware of the extent of their inactivity. As a result, initial experiences are marked by dismay as individuals realize their low activity levels. Rather than confronting users with this “truth”, one could ask how trackers could increase individuals’ perceptions of self-efficacy and competence and support them in the gradual increase of physical activity.
A second challenge is detecting the stage of behavior change individuals are in from behavioral cues. In doing so, one should bear into account that transitions across stages are not always unidirectional. Individuals often relapse to prior stages of behavior change. When this occurs, some individuals “feel like failures – embarrassed, ashamed and guilty” . Future work should thus embrace behavior change as a dynamic journey, should seek to understand the experiential side of behavior change, and to design strategies that support individuals across the full spectrum of their journey.
Together with Ruben Gouveia and Marc Hassenzahl, we received an Honorable Mention Award for Best Paper for the paper How Do We Engage With Activity Trackers? A Longitudinal Study of Habito. The Honorable Mention Award is given to a paper that was identified by the Program Committee as being among the top 5% of all submissions to UbiComp 2015. UbiComp is the premier interdisciplinary conference in the field of ubiquitous and pervasive computing, and a top-ranked international conference in computer science.