Skip to content

{ Category Archives } research

iOS 8: quick thoughts on Health and HealthKit

Since Apple unveiled the Health and HealthKit features as part of iOS 8 yesterday, I’ve a few people ask my thoughts. For those working with personal informatics for health and wellness, I think there’s a lot of reason to be excited but some parts of the announcement that make me skeptical about the impact of this application on its own.

One aspect of the announcement that discourages me a bit is the expectation that a dashboard of calories burned, sleep, physical activity, cholesterol, etc is a “really accurate answer” to “how’s your health?”, and that it adds up to a “clear and current overview of your health.” It’s not — at least not for the typical self-tracker. Though scales aren’t exactly new technology, people misinterpret or read too much into small changes in measurements. Even for many enthusiastic quantified selfers, this data is difficult to interpret and turn into something actionable.

Yesterday’s announcement gave scant mention to this challenge of data interpretation. To the extent that it was discussed, the focus was on sharing data with health providers, with a strong emphasis on Apple’s partnerships with the Mayo Clinic and electronic health record juggernaut EPIC.

For some conditions, such as diabetes, health providers have already developed a strong practice of using the sort of data that Apple proposes people store in HealthKit. Remote monitoring also has the potential to reduce re-admittance rates for patients following heart failure. In these cases and others where medical teams already have experience using patient collected data to improve care, HealthKit and similar tools have the potential to immediately improve the patient experience and reduce user burdens, while also potentially reducing the costs and barriers to integrating data from new devices into care.

Yet for some of the other most prevalent chronic conditions, such as weight management, there is not yet good practice around integrating the data that self-trackers collect into medical care. Weight, sleep, diet, and physical activity are some of the most commonly tracked health outcomes and behaviors. In my research, colleagues and I have talked with both patients and providers who want to, and some who try to, use this data to provide better care, but face many non-technical barriers to doing so. Providers describe feeling pressed for time, doubting the reliability or completeness of the data, feeling overwhelmed by the quantity, or lacking the expertise to suggest specific lifestyle changes based on what they see. Patients describe bringing data to providers but only being more frustrated that the health providers are unable to use it.

For these potential uses, HealthKit or other data sharing platforms seem unlikely to improve care in the short term. What they will do, however, is reduce some of the technical barriers to building systems that help researchers, designers, and health providers learn how patient-collected data can best be used in practice, and to experiment with variations on them. As a researcher working on these questions, this is an exciting development.

Individual trackers, their support networks, and the applications they use also will continue to have an important role in making sense of health data. Self-tracking tools collect more types of data, with greater precision and frequency, than ever before. Your phone can now tell you how many steps you took each minute, but unless it helps you figure out how to get what you want out of that data, this added data is just added burden. Many people I’ve talked with have given up on wearing complicated fitness trackers because they get no value from being able to know their heart rate, step count, and assorted other readings minute-by-minute.

What we need, then, is applications that can help people make sense of their data, through self-experimentation or exploration from which they can draw actionable inferences. Frank Bentley and colleagues have done some great on this with Health Mashups and colleagues and I have been working to design more actionable summaries of data available in lifelog applications such as Moves. Jawbone’s UP Coffee app is a great commercial example of giving people more actionable recommendations.

For these, HealthKit is again exciting. It means that application designers can draw on more data streams without requiring users to manually enter it, or without iOS developers having to implement support for importing from the myriad of health trackers out there. For the end user, it means that one can switch which interpretation application you use or tracking technology you use without having to start over from a blank slate of data (changing from HeathKit to another data integration platform, however, may be another story). So, here too, HealthKit has the potential to enable a lot of innovation, even if the app itself isn’t going to help anyone be healthier.

So, HealthKit is exciting — but most users are still a long way from getting a lot value from it. The best aspect of HealthKit may be that it puts reduces barriers to aggregating data about health factors and outcomes, and that it does so in a way that appears to enable people to retain reasonable control and ownership of their data.

Pervasive, Persuasive Health Challenges: Individual Differences

The role of individual differences in persuasive technology is an exciting research area, one in which I believe our community is making — or about to make — quite a bit of progress. Since it’s getting a lot of attention, I had not included it my review of challenges.

On reflection, though, I believe that our progress here will shortly present us with a substantial new challenge: what do we do when an individual’s preferences are different from what actually works for that individual?

From all sorts of research, we know that people are not very good at predicting what features they actually want over the long term or what will work for them. It’s why Halko and Kientz’s work on personality differences and individuals’ receptiveness to different proposed persuasive systems is great, but the actual efficacy needs to be tested in the field. It’s why people who viewed prototypes of our GoalPost app loved trophies and medals in theory, but found them unmotivating — or even demotivating — in actual use.

This probably isn’t a problem when designers use detected or reported personality differences to make small adjustments to messages within an application. It does, however, present a dilemma if people choose applications with bundles of features that are appealing to them but that do not actually help them achieve their goals, or that help somewhat but not as nearly well as a different bundle of features would. Rosa Ariaga brought this issue to mind at Pervasive Health by asking “why don’t we just create a completely customizable app, and let people choose the features that will motivate them?” My reaction was that, oh, gosh, it seems like many would pick the wrong features and, rather than adapting, just get more discouraged.

Thus, it seems like there are additional questions that should be part of the agenda for people working on individual differences and persuasive systems:

  • What personality (and other) attributes predict different individual preferences for adopting persuasive systems, and what attributes predict individual differences in efficacy of persuasive systems?
  • When do people pick systems that are well matched for their actual needs, and when do they pick systems that are poorly matched?

  • What can, or for that matter, should designers of systems do when people are inclined to pick systems that are not actually helpful for them while neglecting systems that would help?

Pervasive, Persuasive Health Challenges: Designing for Cessation of Use of the Intervention

A second area that has received too little attention is whether we, as designers, intend for people to stop using everyday health and wellness systems, and if so, what the optimal process for that is. In my own work (e.g., [1, 2]), I have focused largely on systems that people might use indefinitely, potentially for the rest of their lives. In doing so, I have focused on making applications that are simple and fast to use, so that people would have an easier time starting and continuing to use them. Given common issues and challenges with adoption and initial adherence, as well as reduced use after the novelty effect wears off, it is no wonder that this particular challenge has thus far received little attention. More cynically, another barrier to this issue receiving much attention is the competing interest of the individual and commercial application/system providers: an individual may prefer to some day no longer need an application, but it is potentially much more lucrative for companies to have a customer for life.

It is, nevertheless, important. First, there may be times when designing systems to support temporary use may actually help some of the initial adoption and adherence problems: people might be willing to put up with a tedious process or a somewhat intrusive device if an application promises to teach them new skills and then be gone from their lives. Second, if we consider what it is like to live with persuasive systems, how many of us would want people to have lives that are carefully regulated and nudged by a myriad of systems, until the day we die [3]? And finally, might some persuasive health systems create an effect of learned helplessness in which applications, assuming the role of determining and recommending the most appropriate choices, actually reduce individuals’ competency to make these decisions in the absence of that support?

Anecdotally, many researchers have described high recidivism rates after the conclusion of an intervention, when the fitness sensor or diary, or the calorie counting tool, is no longer available to the former subjects (this has been observed with other types of interventions as well [4]). Why are these applications not helping individuals to develop good, robust fitness habits or competencies for health eating and at least keeping approximate track of calories? Would a study actually find worse post-intervention health habits among some participants?

To help imagine what we might build if we had a better understanding of how to create temporary health and wellness interventions, consider Schwanda et al’s study of the Wii Fit [5]. Some stopped using the system when it no longer fit into their household arrangement or routine, others when they had unlocked all of the content and its activities became boring or repetitive, and others stopped using it because they switched to another, often more serious, fitness routine. From a fitness perspective, the first two reasons might be considered failures: the system was not robust to changes in life priorities or in living space, or it suffered a novelty effect. The third, though, is a fitness success (though possibly not a success for Nintendo, if the hope is that they would go on to buy the latest/greatest gaming product): participants “graduated” to other activities that potentially were more fulfilling or had still better health and wellness effects. Imagine if the design of the system had helped more users to graduate to these other activities before they became bored with it or before it no longer fit into their daily lives.

Returning to the examples of exercise and calorie diaries, what changes might make them better at instilling healthy habits? In the case of a pedometer application, could it start hiding activity data until participants guessed how many steps that had taking that day? Would such an interface change help people learn to better be aware of their activity level without a device’s constant feedback? What if, after some period of use, users of calorie counters started not getting feedback on the calories they had consumed per food until they end of the day? Would such activities support development of individuals’ health competencies better than tools that offer both ubiquitous sensing and feedback? How would such changes affect the locus of control and sense of self-efficacy of applications’ users?

These are some rough ideas – the medical community, perhaps because of a focus on controlling costs and/or lower ability to integrate the interventions they design into daily life, has more history of evaluating interventions for the post-intervention efficacy (e.g., [6], [7]). Other communities have deeper understanding of what it takes to develop habit (e.g., [8], [9]) or to promote development. What does the HCI community stand to learn from these studies, and to what extent should or community conduct them as well?


  1. Munson SA, Consolvo S. 2012. Exploring Goal-setting, Rewards, Self-monitoring, and Sharing to Motivate Physical Activity, Pervasive Health 2012. [pdf]
  2. Munson SA, Lauterbach D, Newman MW, Resnick P. 2010. Happier Together: Integrating a Wellness Application Into a Social Network Site, Persuasive 2010. [pdf]
  3. Purpura S, Schwanda V, Williams K, Stubler W, Sengers P. 2011. Fit4Life: The Design of a Persuasive Technology Promoting Healthy Behavior and Ideal Weight, Proceedings of CHI 2011. [pdf]
  4. Harland J, White M, Drinkwater C, Chinn D, Farr L, Howel D. 1999. The Newcastle exercise project: a randomised controlled trial of methods to promote physical activity in primary care, BMJ 319: 829-832. [pubmed]
  5. Schwanda V, Ibara S, Reynolds L, Cosley D. 2011. Side effects and ‘gateway’ tools: advocating a broader look at evaluating persuasive systems, Proceedings of CHI 2011. [pdf]
  6. Bock BC, Marcus BH, Pinto BM, Forsyth LH. 2001. Maintenance of physical activity following an individualized motivationally tailored intervention, Annals of Behavioral Medicine 23(2): 79-87. [pubmed]
  7. Moore SM, Charvat JM, Gordon NH, Roberts BL, Pashkow F, Ribisl P, Rocco M. 2006. Effects of a CHANGE intervention to increase exercise maintenance following cardiac events, Annals of Behavioral Medicine 31(1): 53-62. [pubmed]
  8. Rothman AJ, Sheeran P, Wood W. 2009. Reflective and Automatic Processes in the Initiation and Maintenance of Dietary Change, Annals of Behavioral Medicine 38(S1): S4-S17. [pdf]
  9. Verplanken B. 2010. Beyond Frequency: Habit as Mental Construct,
    British Journal of Social Psychology 45(3): 639-656.

Pervasive, Persuasive Health Challenges: One-Time Behaviors

Our field has made great strides in addressing recurring, day-to-day behaviors and challenges: exercising more, regular medication adherence, applications for mood tracking and improvement, smoking-cessation, and managing diet. The same might generally be said for persuasive technology, where the focus has often been on starting and then maintaining behaviors on a regular basis, such as in helping people make day to day greener living choices through eco-feedback technology.

Are the lessons we have learned up to or appropriate for the challenge of motivating or promoting one-time, infrequent, or rare behaviors? Is a focus on reflection, regular monitoring, and objective feedback going to teach us lessons that help us make the best use (or non-use [1]) of technology to promote behaviors such as health screenings or immunization? Indeed, with affordances such as ubiquitous, context-aware objective monitoring and the ability to deliver rich, tailored feedback at the right time and place, mobile computing may much more to offer for everyday behavior change and maintenance.

The answer may be mixed; many of the lessons and affordances may apply. Mobile and context aware systems can still help us deliver tailored messaging, at the right time and right place (kairos) [2]. Various forms of monitoring may identify people who would most benefit from a screening or from a vaccination. Knowledge of social networks and social messaging can help messages carry greater weight with the recipients.

But these problems may present unique challenges for which we, as a research and professional community, have developed less expertise. What are the right engagement points for one-time messaging, when people are not installing applications and interacting with them on a day-to-day basis?

Just as the public health community prefers some health behavior change models for day-day behavior change (e.g., Theory of Reasoned Action [3] & Theory of Planned Behavior [4]) and others for screening or other infrequent behaviors (e.g., the Health Belief Model [5]), the pervasive heath and persuasive technology communities may benefit from developing a different set of guidelines and best practices for this different category of behaviors.

This difference is recognized in models and frameworks such as the Fogg Behavior Grid, which recognizes trying to do a new or familiar behavior one time as a behavior change challenges. The recommended strategies, however, seem represent assumptions that all behavior change of this type is hindered by the same set of barriers. For one-time, new behaviors (“green dot behaviors“), the guide argues:

the main challenge that we face while triggering a Green Dot behavior is a lack of ability. Since Dot behaviors occur only once, the subject must have enough knowledge to successfully complete the action on the first attempt. Otherwise, frustration, and quitting, may occur.”

before moving on to note that motivation and triggers also matter. And for one time, familiar behaviors (“blue dot behaviors“), the recommendation is:

Blue Dot Behaviors are among the easiest to achieve. That’s because the person, by definition, is already familiar with the behavior. They know how to perform it (such as exercise, plant a tree, buy a book). In addition, they already have a sense of the costs and benefits for the behavior… With Blue Dot Behaviors, people do not require reassurance (enhancing motivation) or step-by-step instructions (increasing ability). Instead, the challenge is on timing: One must find a way to deliver a Trigger at a moment when the person is already Motivated and Able. This timing issue is well known: ‘Timing is everything.’

These recommendations and guidelines strike me as overly simplistic. It seems incorrect to assume that someone exercise necessarily sees it as beneficial or is able to exercise properly. Someone might be very able to start a new behavior – a doctor might be recommending a brief screening that is fully covered by an individual’s insurance, but if the individual feels there may be discomfort associated or not understand or believe in the benefits, he or she may still opt out. If these suggestions accurately represent the sum of what we know about persuasive technology for getting people to do one-time behaviors, we have considerably more work to do.

Consider, for example, the challenge of adult immunization. Timing certainly is a barrier, as might be some aspects of ability (having adequate medical insurance or finances to cover it, or knowledge of how to get it for free, for example). But at least some studies find that these are not the most common barriers, with common barriers including misconceptions about vaccines’ costs and benefits — including the belief that because they are healthy, vaccination is unnecessary, or that vaccination has common and negative side effects [6]. Even if the person has received vaccinations before, they may have misconceptions that leave them unmotivated. Or they may have once been able but had their circumstances change — such as by losing access to insurance or a shift in their social network to one that disapproves of vaccinations, or a doctor that is less inclined to remind patients about them.

A framework, then, that errantly, or over-generally, assumes and emphasizes certain barriers and not others may miss more effective opportunities for intervention, interventions that only work with people for whom it has accurately described the barriers. For the vaccination challenge, focusing on changing social norms, and making pro-vaccination norms visible, may be more effective in some communities.

There are also questions about how to deliver technical interventions for one-off activities (or if/when technical interventions are even well-suited). When the challenge is a trigger, getting a patient to install a reminder application that will trigger at an appropriate time (when the seasonal flu shot is available, for example) and context (when in a pharmacy that accepts their insurance). Even then, an individual might not see the benefits to keeping a single-purpose application around and delete it, or may witch phones, in the meantime, making the reminder less effective. Would bundling many one-time behavioral interventions into a single application, perhaps with day-to-day interventions as well, work? For vaccinations, an application to manage a patient’s interactions with a caregiver (including scheduling, billing, suggested vaccinations and screenings, access to health records, and so on), might be optimal. Text4Baby bundles many one-time health tips into a stream of health advice that is timed with expectant and new mothers’ needs; are there other such opportunities?

Furthermore, for health conditions that are more stigmatizing, some traditional techniques to increase motivation may be problematic. Despite the effectiveness of seeing celebrities or friends pursing a health behavior (e.g., the “Katie Couric effect” for colonoscopies [7]), social messages about who in your network has received a screening or vaccine may sometimes disclose more than is appropriate. I applaud efforts, such as Hansen and Johnson’s work on “veiled viral marketing” [8], to develop social triggers that work but are also appropriate for sensitive health behaviors. In their test of veiled viral marketing, individuals could send a social message to someone in their network recommending that they learn about the HPV vaccine – and the recipient would learn that a friend recommended this content, but not which friend, thus preserving the saliency of a social message while still affording some privacy to the sender.

Social proof, on Facebook, that others, including your friends, voted.

Social proof, on Facebook, that others, including your friends, voted.

For other situations, technology — such as social network data, precise knowledge about communities and attitudes, and electronic health records — might be better used to tailor messages that are delivered through various media, rather than delivering specific triggers. A Facebook app indicating “I was vaccinated,” with numbers and friends (possibly just a count in the case of stigmatizing conditions) – much like the experiment conducted during the US 2008 Presidential Election and 2010 midterm elections (right) – might add social proof and pressure, while messaging that highlights people in one’s network who you could be protecting by getting vaccinated might increase perceived benefits or feelings of responsibility.

It is also possible that some techniques will be better suited for one-time use than ongoing, day-to-day use. Social comparison data has been shown to be effective in yielding higher contributions to public radio [9], reducing energy use (particularly when combined with injunctive norms [10]), and increasing ratings in an online movie community [11]. I would speculate, though, that in at least some long-term, discretionary use applications, some individuals would prefer to avoid sites that regularly present them with aversive comparisons.

From a paper for the Wellness Interventions and HCI workshop at Pervasive Health 2012.


  1. Baumer EPS & Silberman MS. 2011. When the Implication is Not to Design (Technology), Proceedings of CHI 2011. [pdf]
  2. Fogg BJ. 2002. Persuasive Technology: Using Computers to Change What We Think and Do. Morgan Kaufmann. [Amazon]
  3. Fishbein M. and Ajzen I. 1975. Belief, attitude, intention, and behavior: An introduction to theory and research. Reading, MA: Addison-Wesley. [online]
  4. Ajzen I. 1991. The theory of planned behavior, Organizational Behavior and Human Decision Processes 50(2): 179–211. [pdf]
  5. Rosenstock IM. 1966. Why people use health services, Milbank Memorial Fund Quarterly 44 (3): 94–127. [pdf]
  6. Johnson DR, Nichol KL, Lipczynski K. 2008. Barriers to Adult Immunization, American Journal of Medicine 121(7B): 528-535. [pdf]
  7. Cram P, Fenrick AM, Inadomi J, Cowen ME, Vijan S. 2003. The impact of a celebrity promotional campaign on the use of colon cancer screening: the Katie Couric effect, Archives of Internal Medicine 163(13):1601-5. [pubmed]
  8. Hansen D and Johnson C. 2012. Veiled Viral Marketing: Disseminating Information on Stigmatized Illnesses via Social Networking Sites, Proceedings of the 2nd ACM SIGHIT International Health Informatics Symposium. [slideshare, acm]
  9. Sheng J, Croson R. 2009. A Field Experiment in Charitable Contribution: The Impact of Social Information on the Voluntary Provision of Public Goods, The Economic Journal 119(540): 1422-1439. [pdf]
  10. Schultz PW, Nolan JM, Cialdini RB, Goldstein NJ, Griskevicius V. 2007. The Constructive, Destructive, and Reconstructive Power of Social Norms, Psychological Science 18(5): 429-434. [pdf]
  11. Chen Y, Harper F Maxwell, Konstan J, Li, Sheri Xin. 2010. Social Comparisons and Contributions to Online Communities: A Field Experiment on MovieLens. American Economic Review 100(4): 1358-98. [pdf]

Pervasive, Persuasive Health: Some Challenges

From a paper for the Wellness Interventions and HCI workshop at Pervasive Health 2012.

As a community, or perhaps more accurately, as communities, health and persuasive technology researchers have made considerable progress on understanding the opportunities, challenges, and some best practices for designing technology to support health and wellness. There is an incredibly rich stream of current and past research, as well as commercially available applications to support a variety of health behaviors.

I think that there remain some under-researched challenges, and I question whether our existing knowledge and research directions can sufficiently address these challenges. If not, what else we should be including in our research discussions and plans. In particular, are doing enough to study one-time interventions and the process for tapering, weaning, or graduating people off of the interventions we build and deploy?

Over the next few days, I’ll be posting my thoughts on the challenges of one-time use and designing for tapered use. I’d love your feedback and your thoughts on other areas that are under-explored or studied in the persuasive health system communities.

Reflection and Persuasion in Personal Informatics

After a variety of conversations, I’ve expanded on my earlier thoughts on reflection, mindfulness, persuasion, and coercion in systems for this year’s Personal Informatics in Practice workshop at CHI. The expanded article and a blog post introducing it are over on the Quantified Self blog.

3GT is back

Just a quick note that 3GT, our Facebook app based on the positive psychology exercise Three Good Things, is back online after some time at the spa. The new version is based on some of what I learned from the last release. Some of the social and feedback lessons still need to be applied, so look for that in new features over the next couple of weeks.

You can access it as a standalone app or on Facebook. I’m just doing friends and family testing and some refinements on it for now, so feedback is welcome.

Health behavior models in the age of mobile interventions: are our theories up to the task?

I recently read “Health behavior models in the age of mobile interventions: are our theories up to the task?,” by Riley, Rivera, Atienza, Nilsen, Allison, and Mermelstein, which appeared in March’s Translational Behavioral Medicine. The authors review papers, up to 2010 and as indexed by MedLine, that assess mobile health interventions and report at least some clinical outcome.

For each paper, the authors note the theories (if any) that were cited as guiding the development of the intervention. Most common were Theory of Planned Behavior, the Health Belief Model, Self-Determination Theory, Social Cognitive Theory, Self-Regulation Theory, the Transtheoretical Model, and cognitive-behavioral theory. The authors find that these theories are somewhat lacking for their ability to to guide the use of the affordances of mobile technology. In particular, theories based on linear progression tend not to take full advantage of mobile devices’ abilities for context awareness (through sensing and prompted input), increased variation in timing and content of interventions, and real-time awareness of outcomes.

While the health behavior theories may not be up to the task on their own, there are theories from a variety of disciplines that can guide, and are guiding, intervention designers. The authors suggest looking to control theory to address some of the shortcomings they identify, and offer a couple of interesting examples. The HCI and social computing community, which knows interactive systems and technology mediated social interaction, also has quite a bit to offer. Though HCI often borrows or adds to theories from other disciplines — including from the health community — it has developed quite a bit of expertise about how to implement these in systems, and developed its own applicable frameworks, such as the Reader to Leader Framework.

The way the authors found the literature they reviewed — a search of Medline — highlights a bit of a communication gap between these disciplines. No papers from HCI conferences or journals were included in this review, despite work such as UbiFit, MAHI, and Fish’n’Steps being excellent examples of the type of studies the authors reviewed. The authors are not alone in excluding HCI papers from review articles on this topic (e.g., [1], [2]). I agree, to a large extent, with Klasnja et al that deep understanding of users’ experiences with behavior change strategies, as implemented in a system, is one of the primary contributions HCI make can to this health behavior change work. Stopping at such findings, however, may be insufficient for the results to have an impact outside of the HCI community, much as eco-feedback work in HCI has not always reached beyond the HCI community.

Collaborations between the HCI and health communities, such as UCSD’s Center for Wireless & Population Health Systems, the iMed group at UW, or Northwestern’s newly announced Center for Behavioral Intervention Technologies, as well as boundary spanning venues such as WISH, Health 2.0, or the Designing for Healthy Living special issue of IJMI, are helping to close some of the gap between communities. We should, however, keep asking whether this will be enough.

We should also keep asking what might be on a research agenda that combines the strengths of HCI and health behavior change communities (or at least, I should, since I have a dissertation to write), with an eye toward advancing both theory and practice. Some thoughts:

  • What are appropriate ways to elicit social support or social pressure from others on social network sites or other technology mediated spaces?
  • When does anonymity support people trying to make healthy behavior changes, such as by helping them feel comfortable sharing, and when it is a barrier to making those changes, such as by reducing accountability?
  • Are some types of nudges that work once or in non-discretionary use problematic over the long term, such as because people start avoiding the system and aversive feedback?
  • How can we most effectively use leaderboards to support behavior change and maintenance? With data from many people, there are many more options for how to construct leaderboards than in smaller groups. For a given behavior or set of people, which comparison direction(s) and dimensions of similarity are most effective?
  • How applicable are health behavior change theories to other behavior change domains, such as eco-feedback? Where do the existing theories apply, where do they break? We already see things like the health belief model being applied to computer security practices.
  • As features used to nudge or support decision making become more widely adopted, do some feature remain effective or become like infrastructure while others become less effective?
  • What are the privacy implications of integrating sensing, storage, and sharing of of data? Is HIPAA policy even up to this task?
  • When do behavior nudges become coercive? What are good practices for ethics of these interventions? While many counseling strategies promote autonomy and intrinsic motivation as both a matter of efficacy and ethics, does this balance shift as devices that allow constant and long term monitoring (and thus control) become available? For example, some people can already save money on their insurance by agreeing to wear a pedometer and meeting certain step goals.

These are by no means completely open questions. Researchers and practitioners in both communities have been tackling many of them and making progress. It is also by no means a complete list.

CHI Highlights: General

Okay, the third and final (?) set of CHI Highlights, consisting of brief notes on some other papers and panels that caught my attention. My notes here will, overall, be briefer than in the other posts.

More papers

  • We’re in It Together: Interpersonal Management of Disclosure in Social Network Services
    Airi Lampinen, Vilma Lehtinen, Asko Lehmuskallio, Saraki Tamminen

    I possibly should have linked to this one in my post about social networks for health, as my work in that area is why this paper caught my attention. Through a qualitative study, the authors explore how people manage their privacy and disclosures on social network sites.

    People tend to apply their own expectations about what they’d like posted about themselves to what they post about others, but sometimes negotiate and ask others to take posts down, and this can lead to either new implicit or explicit rules about what gets posted in the future. They also sometimes stay out of conversations when they know that they are not as close to a the original poster as the other participants (even if they have the same “status” on the social network site). Even offline behavior is affected: people might make sure that embarrassing photos can’t be taken so that they cannot be posted.

    To regulate boundaries, some people use different services targeted at different audiences. While many participants believed that it would be useful to create friend lists within a service and to target updates to those lists, many had not done so (quite similar to my findings with health sharing: people say they want the feature and that it is useful, but just aren’t using it. I’d love to see Facebook data on what percent of people are actively using lists.) People also commonly worded posts so that those “in the know” would get more information than others, even if all saw the same post.

    Once aversive content had been posted, however, it was sometimes better for participants to ty to repurpose it to be funny or a joke, rather than to delete it. Deletions say “this was important,” while adding smilies can decrease its impact and say “oh, that wasn’t serious.”

  • Utility of human-computer interactions: toward a science of preference measurement
    Michael Toomim, Travis Kriplean, Claus Pörtner, James Landay

    Many short-duration user studies rely on self-report data of satisfaction with an interface or tool, even though we know that self-report data is often quite problematic. To measure the relative utility of design alternatives, the authors place them on Mechanical Turk and measure how many tasks people complete on each alternative under differing pay conditions. A design that gets more work for the same or less pay implies more utility. Because of things like the small pay effect and its ability to crowd out intrinsic rewards, I’m curious about whether this approach will work better for systems meant for work rather than for fun, as well as just how far it can go – but I really do like the direction of measuring what people actually do rather than just what they say.

Perhaps it is my love of maps or my senior capstone project at Olin, but I have a love for location based services work, even if I’m not currently doing it.

  • Opportunities exist: continuous discovery of places to perform activities
    David Dearman, Timothy Sohn, Khai N. Truong

  • In the best families: tracking and relationships
    Clara Mancini, Yvonne Rogers, Keerthi Thomas, Adam N. Joinson, Blaine A. Price, Arosha K. Bandara, Lukasz Jedrzejczyk, Bashar Nuseibeh

    I’m wondering more and more if there’s an appropriate social distance for location trackers: with people are already very close, it is smothering, while with people who are too far, is it creepy? Thinking about preferences for Latitude, I wouldn’t want my family or socially distance acquaintances on there, but I do want friends who I don’t see often enough on there.

The session on Affective Computing had lots of good stuff.

  • Identifying emotional states using keystroke dynamics
    Clayton Epp, Michael Lippold, Regan L. Mandryk

    Fairly reliable classifier for emotions, including confidence, hesitance, nervousness, relaxation, sadness, and tiredness, based on analysis of typing rhythms on a standard keyboard. One thing I like about this paper is it opens up a variety of systems ideas, ranging from fairly simple to quite sophisticated. I’m also curious if this can be extended to touch screens, which seems like a much more difficult environment.

  • Affective computational priming and creativity
    Sheena Lewis, Mira Dontcheva, Elizabeth Gerber

    In a Mechanical Turk based experiment, showing people a picture that induced positive affect increased the quality of ideas generated — measured by originality and creativity — in a creativity task. Negative priming reduced comparison compared to positive or neutral priming. I’m very curious to see if this result is sustainable over time, with the same image or with different images, or in group settings (particularly considering the next paper in this list!)

  • Upset now?: emotion contagion in distributed group
    Jamie Guillory, Jason Spiegel, Molly Drislane, Benjamin Weiss, Walter Donner, Jeffrey Hancock

    An experiment on emotional contagion. Introducing negative emotion led others to be more negative, but it also improved the group’s performance.

  • Emotion regulation for frustrating driving contexts
    Helen Harris, Clifford Nass

We’ve seen a lot of research on priming in interfaces lately, most often in lab or mturk based studies. I think it’ll start to get very interesting when we start testing to see if that also works in long-term field deployments or when people are using a system at their own discretion for their own needs, something that has been harder to do in many past studies of priming.

I didn’t make it to these next few presentations, but having previously seen Steven talk about this work, it’s definitely a worth a pointer. The titles nicely capture the main points.

The same is true for Moira’s work:

Navel-Gazy Stuff

  • RepliCHI: issues of replication in the HCI community
    Max L. Wilson, Wendy Mackay, Dan Russell, Ed Chi, Michael Bernstein, Harold Thimbleby

    Discussion about the balance between reproducing other studies in different contexts or prior to making “incremental” advances vs. a focus on novelty and innovation. Nice summary here. I think the panelists and audience were generally leading toward increasing the use of replication + extension in the training of HCI PhD students. I think this would be beneficial, in that it can encourage students to learn how to write papers that are reproducible, introduce basic research methods by doing, and may often lead to some surprising and interesting results. There was some discussion of whether there should be a repli.chi track alongside an alt.chi track. I’m a lot less enthusiastic about that – if there’s a research contribution, the main technical program should probably be sufficient, and if not, why is it there? I do understand that there’s an argument to be made that it’s worth doing as an incentive, but I don’t think that is a sufficient reason. Less addressed by the panel was that a lot of the HCI research isn’t of a style that lends itself to replication, though Dan Russell pointed out that some studies must also be taken on faith since don’t all have our own Google or LHC.

  • The trouble with social computing systems research
    Michael Bernstein, Mark S. Ackerman, Ed Chi, Robert C. Miller

    alt.chi entry into the debate perceived issues with underrepresentation of systems work in CHI submissions and with how CHI reviewers treat systems work. As someone who doesn’t do “real” systems work — the systems I build are usually intended as research probes rather than contributions themselves) — I’ve been reluctant to say much on this issue for fear that I would talk more than I know. That said, I can’t completely resist. While I agree that there are often issues with how systems work is presented and reviewed, I’m not completely sympathetic to the argument in this paper.

    Part of my skepticism is that I’ve yet to be shown an example of a good systems paper that was rejected. This is not to say that these do not exist; the authors of the paper are speaking from experience and do great work. The majority of systems rejections I have seen are from reviewing, and the decisions have mostly seemed reasonable. Most common are papers that make a modest (or even very nice) systems contribution, tack on a poorly executed evaluation, and then make claims based on the evaluation that it just doesn’t support. I believe at least one rejection would have been accepted had the authors just left out the evaluation altogether, and I think a bad evaluation and unsupported claims should doom a paper unless they are excised (which maybe possible with the new CSCW review process).

    I was a little bit frustrated because Michael’s presentation seemed to gloss over the authors’ responsibilities to explain the merits of their work to the broader audience of the conference and to discuss biases introduced by snowball samples. The last point is better addressed in the paper, but I still feel that the paper still deemphasizes authors’ responsibility in favor of reviewers’ responsibility.

    The format for this presentation was also too constrained to have a particularly good discussion (something that was unfortunately true in most sessions with the new CHI time limits). The longer discussion about systems research in the CSCW and CHI communities that followed one of the CSCW Horizons sessions this year was more constructive and more balanced, perhaps because the discussion was anchored at least partially on the systems that had just been presented.

  • When the Implication Is Not to Design (Technology)
    Eric P.S. Baumer, M. Six Silberman

    Note on how HCI can improve the way we conduct our work, particularly the view that there are problems and technical solutions to solve them. The authors argue that it may be better think of these as conditions and interventions. Some arguments they make for practice are: Value the implication not to design technology (i.e., that in some situations computing technology may be inappropriate), explicate unpursued avenues (explain alternative interventions and why they were not pursued), technological extravention (are there times when technology should be removed?), more than negative results (why and in what context did the system fail, ad what does that failure mean), and to not to stop building – just to be more reflective on why that building is occurring.

CHI Highlights: Diversity, Discussion, and Information Consumption

For my second post on highlights from CHI this year, I’m focusing on papers related to opinion diversity and discourse quality.

  • Normative influences on thoughtful online participation
    Abhay Sukumaran, Stephanie Vezich, Melanie McHugh, Clifford Nass

    Two lab experiments on whether it is possible to foster more thoughtful commenting and participation by participants in online discussion forums by priming thoughtful norms. The first tested the effects of the behavior of other participants in the forum. The dependent variables were comment length, time taken to write the comments, and number of issue-relevant thoughts. Not surprisingly, being exposed to other thoughtful comments led people to make more thoughtful comments themselves. One of the audience members asked the question about whether this would break down with just one negative or less thoughtful comment (such as how merely one piece of litter seems to break down antilittering norms).

    The second study tested effects of visual, textual, and interaction design features on the same dependent variables. The manipulations included a more subdued vs. more playful visual design, differing CAPTCHAs (words positively correlated with thoughtfulness in the thoughtful condition and words negatively correlated with thoughtfulness in the unthoughtful condition), and different labels for the comment box. The design intended to provoke thoughtfulness did correspond to more thoughtful comments, suggesting that it is possible, at least in the lab, to design sites to prompt more thoughtful comments. For this second study in particular, I’m curious if these measures only work in the short term or if they would work in the long term and about the effects of each of the specific design features.

  • I will do it, but I don’t like it: User Reactions to Preference-Inconsistent Recommendations
    Christina Schwind, Jürgen Buder, Friedrich W. Hesse

    This paper actually appeared a in health session, but I found that it spoke much more to the issues my colleagues and I are confronting in the BALANCE project. The authors begin with the observation that most recommender systems are intended to produce content that their users will like, but that this can be problematic. In the health and wellness domain, people sometimes need to hear information that might disagree with their perspective or currently held beliefs, and so it can be valuable to recommend disagreeable information. In this Mechanical Turk-based study, subjects were equally likely to follow preference-consistent and preference-inconsistent recommendations. Following preference-inconsistent recommendations did reduce confirmation bias, but people were happier to see preference-consistent recommendations. This raises the important question: subjects may have followed the recommendation the first time, but now that they know this system gives recommendations they might not like, will they follow the recommendations less often in the future, or switch to another system altogether?

  • ConsiderIt: improving structured public deliberation
    Travis Kriplean, Jonathan T. Morgan, Deen Freelon, Alan Borning, Lance Bennett

    I really like the work Travis is doing with Reflect and ConsiderIt (which powers the Living Voters Guide) to promote more thoughtful listening and discussion online, so I was happy to see this WiP and am looking forward to seeing more!

  • Computing political preference among twitter followers
    Jennifer Golbeck, Derek L. Hansen

    This work uses follow data for Congresspeople and others on Twitter to assess the “political preference” (see comment below) of Twitter users and the media sources they follow. This approach and the information it yields has some similarities to Daniel Zhou’s upcoming ICWSM paper and, to a lesser extent, Souneil Park’s paper from CSCW this year.

    One critique: despite ample selective exposure research, I’m not quite comfortable with this paper’s assumption that political preference maps so neatly to political information preference, partly because I think this may be an interesting research question: do people who lean slightly one way or the other prefer information sources that may be more biased than they are? (or something along those lines)

In addition to these papers, Ethan Zuckerman’s closing plenary, Desperately Seeking Serendipity, touched on the topics of serendipity and homophily extensively. Zuckerman starts by suggesting the reason that people like to move to cities – even at times when cities were really quite horrible places – is, yes, for more opportunities and choices, but also “to encounter the people you couldn’t encounter in your rural, disconnected lifestyle… to become a cosmopolitan, a citizen of the world.” He goes on, “if you wanted to encounter a set of ideas that were radically different than your own, your best bet in an era before telecommunications was to move to a city.” There reasons to question this idea of cities as a “serendipity engine,” though: even people in urban environments have extremely predictable routines and just don’t go all that many places. Encounters with diverse others may not as common as idealized.

He then shifts gears to discuss what people encounter online. He walks through the argument that the idea of a Freshman Fishwrap or Daily Me is possibly quite harmful as it allows people to filter to the news that they want. Adding in social filters or getting news through our friends can make this worse. While Sunstein is concerned about this leading to polarization within the US, Zuckerman is more concerned that it leads people to see only news about where they are and less news about other places or from outside perspectives. This trend might lead people to miss important stories.

I tend to agree with the argument that surfacing coincidences or manufacturing serendipity is an incredibly powerful capability of current technology. Many of the approaches that the design community has taken to achieve this are probably not the kind of serendipity Zuckerman is looking for. I love Dopplr’s notifications that friends are also in town, but the time I spend with them or being shown around by them is time that I’m less likely to have a chance encounter with someone local or a traveler from elsewhere. The ability to filter reviews by friends may make for more accurate recommendations, but I’m also less likely to end up somewhere a bit different. Even serendipity has been repurposed to support homophily

Now, it might be that the definition of serendipity that some of the design community hasn’t quite been right. As Zuckerman notes, serendipity usually means “happy accident” now – it’s become a synonym for coincidence – and that the sagacity part of the definition has been lost. Zuckerman returns to the city metaphor, arguing for a pedestrian-level view. Rather than building tools for only efficiency and convenience, build tools and spaces that maximize the chances to interact and mix. Don’t make filters hidden. Make favorites of other communities visible, not just the user’s friends. Zuckerman elegantly compares this last feature to the traces in a city: one does not see traces left just by one’s friends, no, but traces left by other users of the space, and this gives people a chance to wander from the path they were already on. One might also overlay a game on a city, to encourage people to explore more deeply or venture to new areas.

While I like these ideas, I’m a little concerned that they will lead to somewhat superficial exposure to the other. People see different others on YouTube, Flickr, or in the news, and yes, some stop and reflect, others leave comments that make fun of them, and many others just move on to the next one. A location-based game might get people to go to new places, but are they thinking about what it is like to be there, or are they thinking about the points they are earning? This superficiality is something I worry about in my own work to expose people to more diverse political news – they may see it, but are they really considering the perspective or gawking at the other side’s insanity? Serendipity may be necessary, but I question whether it is sufficient. We also need empathy: technology that can help people listen and see others’ perspectives and situations. Maybe empathy is part of the lost idea of sagacity that Zuckerman discusses — a sort of emotional sagacity — but whatever it is, I need to better know how to design for it.

For SI and UM students who really engage with this discussion and the interweaving of cities, technology, and flows, I strongly, strongly recommend ARCH 531 (Networked Cities).