1. Introduction

Scientists have found that physical inactivity (PI) is related to poor health conditions such as obesity, heart disease, stroke, specific kinds of diabetes and the incidence of at least three types of cancer[1]. PI accounts for 1 in 10 deaths each year and was among the leading global risks for mortality in 2009[2],[3]. In a highly publicized study in 2012, the health risks of PI were found to be similar to smoking[4]. Fitness trackers have been presented as one of the solutions to physical inactivity. In 2015, one in five American adults owned at least one wearable1. 78.1 million wearable devices were sold last year which is 171.6% more than the year before2. Companies that manufacture and support fitness trackers collect a lot of data about their usage — what features are used and what aren’t, how long they are used for and can also make inferences about what kinds of people use fitness trackers. In this sense, they receive direct feedback from users about the product. Consumers themselves provide feedback to other potential consumers — through Amazon reviews. The aim of this project is to analyse patterns in language and sentiments around different fitness trackers through self-reported user reviews and ratings on Amazon.com. The objective is to understand what features people think are important and whether these products inculcate a self-reflective practise of behaviour change. The secondary objective of the project is to explore the possibilities of a quantitative and qualitative approach to analysing large amounts of voluntarily and publicly disclosed information as a form of user research. The nature of this kind of data is very different from that created through traditional one-on-one interviews or focus groups.

2. Method

2.1 Selecting the fitness trackers

A general search on Amazon for ‘fitness trackers’ gives 55,264 results across 34 departments. The products are across 11 brands. Based on Amazon’s own Bestsellers’ List, the following products were selected. Fitbit in particular had multiple products in the Bestsellers’ List but an attempt was made to balance other brands to get more comprehensive feedback. These were the Fitbit Alta, Garmin vívofit, Fitbit Flex, Polar A300 Fitness Tracker and Activity Monitor, Scosche Rhythm+ Heart Rate Monitor Armband, 3DTriSport Walking 3D Pedometer, stardrift Upgraded Non-Bluetooth Pedometer. The armband had a different form and the last two were not marketed as fitness trackers, the features were comparable.

2.2 Creating a Master Features list

Creating a features list was more difficult that it seemed because of unique marketing phrases that the manufacturers used. Each feature had to be really understood before the product was evaluated.

2.3 Defining key words

People are not likely to use the same words for the features and for word signalling behavior change.

2.3.1 Words related to the features

Steps: step, stride, footsteps, walk
Calories: calorie
Time: clock, watch
Sleep tracking: sleep, sleeping, slept
Silent Alarm: vibration alarm, vibrating alarm, haptic alarm
Waterproof: water proof, rainproof, water-resistant
Detection of activity switching: activity switching, SmartTrack, automatically detects activity, automatically senses, switches activity
Reminders to move: reminders to move, idle alert, reminds you to move
Multiple sports: multiple sports
Cardio fitness level: cardio fitness level
Personalized daily goals: personalized daily goals, daily goal, goal
Activity time
Backlight display: backlight, display, screen
LED display: LED, display, screen
Call notification: phone, call, calls, calling, call notification, call notifies, calling notifies
Text notification: phone, text, texts, messages, message, message notification
Timer
Calendar: calendar, appointments
Heart rate monitor: heartbeat, heart rate, pulse, pulse rate
Alarm

2.3 Defining key words

People are not likely to use the same words for the features and for word signalling behavior change.

2.3.1 Words related to the features

Steps: step, stride, footsteps, walk
Calories: calorie
Time: clock, watch
Sleep tracking: sleep, sleeping, slept
Silent Alarm: vibration alarm, vibrating alarm, haptic alarm
Waterproof: water proof, rainproof, water-resistant
Detection of activity switching: activity switching, SmartTrack, automatically detects activity, automatically senses, switches activity
Reminders to move: reminders to move, idle alert, reminds you to move
Multiple sports: multiple sports
Cardio fitness level: cardio fitness level
Personalized daily goals: personalized daily goals, daily goal, goal
Activity time
Backlight display: backlight, display, screen
LED display: LED, display, screen
Call notification: phone, call, calls, calling, call notification, call notifies, calling notifies
Text notification: phone, text, texts, messages, message, message notification
Timer
Calendar: calendar, appointments
Heart rate monitor: heartbeat, heart rate, pulse, pulse rate
Alarm

2.3.2 Words related to behavior change

State of mind
intent — intent, intention, intended, intending, mean, meant, meaning, hope, hoped, hoping
health — health, healthy, fit, fitness, strength, strong, stronger, weight
habit — habit, routine, practice, practise
lifestyle — lifestyle, regime
buy — buy, bought, purchased, got
because — cos, because, since, therefore, hence
motivate — motivate, motivates, motivated, motivating, motivational
aware — aware, awareness, notice, noticed, conscious, consciousness

Change
time — since, earlier, now, used to
amount — more, less, increase, decrease, increased, decreased, increasing, decreasing, same
change — change, changing, changed, turned, transformed, transformation, improve, improved, improving, worse, worsen, worsening
notice — notice, noticed, noticing

Use
goal— goal, aim, target, aimed, targetted, objective
activity — yoga, tai-chi, meditation, aerobic, weight, weights, lift, lifting, lunge
behavior — behavior, behaviour, behaviors, behaviours

Relationship
encourage — encourage, encouragement, encouraging, encouraged, encouragements
help — help, helps, helped, helping

Emotions — happy, sad, angry, guilty, unhappy

2.4 Creating SQl databases

In order to pass multiple queries to the collected data, multiple interlinked sql databases were created that tabulated data across multiple factors. For example, each line extracted was tabulated against the product the overall review was about, the number of stars it got, the positive, negative, neutral and compound scores of the line and the behaviour attribute or feature attribute the line was about. The python code was written to populate these databases

2.5 Python code

The code consisted of several function definitions –

1. Finding the number of pages per product
2. Getting different products’ reviews
3. Getting all reviews per star rating
4. Splitting the reviews into sentences
5. Doing sentiment analysis
6. Reading the database table for features and product
7. Reading the database table for features and line
8. Populating the database tables

2.6 Scanning the data

The last part of the project involved manually scanning the sentences. Without programming, it would have taken forever to scan through all of the 33,602 reviews. The code contained filters in the form of key words and blinders by segmenting the reviews into individual sentences. This made it much easier and faster to scan through the content.

3. Results and discussion

The qualitative and ‘human’ aspect of scanning the data revealed some interesting aspects of how people write reviews and what they choose to communicate to other people.

1. The difference between positive and negative reviews

People tend to write more about change and motivation when the overall review is positive. When they rate the product negatively, most of the complaints are about the inaccuracy of the features (rather than the helpfulness), the physical product design of the tracker or strap, customer service and the usability of the accompanying app. This overall feedback would be great for fitness tracker manufacturers to delve into and resolve. One possible reason for the difference in the kinds of things spoken about in positive and negative reviews is that since people like the product, it has a sort of ‘halo effect’ where other features are also considered good. Also, since people view the product favourably, the y write more, and because they write more, they reflect more about it and can get to the point in the review where they talk about deeper aspects like behaviour, motivation and the effectiveness of the features rather than just the superficial feedback about the product design etc.

2. Difference in the time periods mentioned in the reviews of fitness trackers and pedometers

Pedometers have batteries that cannot be charged while most fitness trackers have to be charged almost daily. The reviews of pedometers contained keywords like ‘year’ and ‘months’ compared to ‘hours’, ‘days’, ‘weeks’. Though it can’t be extrapolated with complete accuracy that people use pedometers longer than they use fitness trackers, there is data that supports this. Studies have shown that most people stop using fitness trackers after 6 months[5]. This also is consistent with user interviews conducted prior to this project.

3. Kind of language used

Language used in the reviews was much more hyperbolic than that used in a conversational interview or short answer survey. This may have to do with the perceived anonymity that makes people feel more comfortable expressing themselves, the fact that they are writing the review out of an altruistic motive to help other people make a decision. On the other hand, it could be argued that because people are writing in order to convince other people, they need to be more extreme in their use of language to communicate their points. Nonetheless, the reviews are more open-ended and less subject to Interviewers’ Bias where people try to answer questions based on what they think the interviewer wants to hear.

4. Emotion scale

There were a wide variety of emotions that reviewers directly alluded to. A quick glance revealed that the most common was ‘very happy’, followed by ‘happy’, ‘sad’, ‘not happy’ and ‘angry’. This indicates that people don’t think of their feelings on a simple scale from ‘happy’ to ‘unhappy’ that the ratings indicate. ‘Sad’ and ‘angry’ reflect more nuances of their emotions.

5. Supplement to recovery

Quite a few reviews were by people who use fitness trackers to recover from illnesses or treatments like chemotherapy or knee replacement surgery. One reviewer wrote — “I’ve been in therapy for a back injury for almost a year, so it’s been painful to walk for anymore than about 40 steps at a time, so you know how the weight can creep up when you’re inactive, making it all the worse.” This indicates a significant user group that wants to use fitness trackers to recover from illness. A lot of older people also tended to use it, or atleast wrote about their experience.

6. People buy it for others too

People wrote about how they bought it for someone else and it could be inferred that this is a practical sign of care. People also bought them in pairs or for their entire family. For example, someone wrote “We are a Fitbitting family and are getting TONS of enjoyment out of these.” And “My husband and I are part of the Fitbit Revolution.”

7. Competition

It was surprising to see that competition was perceived as positive motivating factor by almost everyone who mentioned it in their review. It seemed like the fact that generated competition was itself a motivation to buy it. For example, someone wrote, “So much so that I bought one for my boyfriend (a farmer) and we have daily ‘competitions’ as to who garnered more steps.”

8. Display

In terms of features, the most surprising finding was that people really liked the immediate validation that the screen displays provided. They liked seeing a detailed graphical display and wanted to see see exact number rather than dots indicating overall progress. This may be because numbers provide more granular information and act as ‘micro-motivators’ since they increase faster and provide immediate feedback about activity than four lights that indicate levels of activity.

9. Features that were wished for

This project was a very useful way of finding out what people like, don’t like and would like to have in their fitness trackers. Though somewhat predictable, this finding is still valuable because it was provided without any prompting and is reported directly from users. Some of the quotes about different features are included below.

It would be nice to be able to increase the minimum steps.

Useless on a bike.

I also miss the ever-changing goals that the Vivosmart provided (changing based on how much you walked the day before).

I noticed inputting exercises on it would be a pain in the rear since it doesn’t let you actually enter the workout you’re doing, it just let me enter things like ‘weights’.

The only thing that I would love to see within this app is a place to make notes for each day, so I can have a record all in one place as to what was happening on the days I didn’t reach my goals.

Having the app actually gives me too many options (such as checking calories and connecting to social media), but I found it invaluable to keeping me on track with my daily goals.

4. Conclusion

Based on the findings of the research, people do find fitness trackers helpful in combatting physical inactivity. Studying Amazon reviews doesn’t provide details on usage over time. Amazon as a platform is meant to be a marketplace that uses user reviews as a form of social proof to sell products and as such, is not the place to collect the most neutral feedback. Certain mechanisms (like reviewing products to receive benefits) and user flows (people rate the product before they write the review) impact how people write reviews. Though reviewers aren’t biased by interviewers’ expectations, the content is still biased in terms of how and why people write it. How researchers themselves define and select the key words, and how the last step of scanning the data are also prone to human biases. These can’t be avoided but it’s important to be aware of them.

There are definite benefits of using publicly available data — the vast quantity and variety of it, it could be assumed that people across demographics write reviews and is therefore more representative of the larger population than a small university-based research can have access too. Like with a lot of quantitative data, analysing these reviews answers a lot of the ‘What?’ questions but because of the limitation of the one-way communication, it’s hard to infer answers to the ‘Why?’ questions.

Manual user research is typically done to find new and unique information and has diminishing returns after a certain number of interviews have been conducted. On the other hand, computational user research provides validation of existing hypotheses. For example — more people buy it for other people than themselves. This makes the hypotheses framing aspects of such research critical. It’s also important to build upon these hypothesis and change course based on the findings. For example, if more people buy fitness tracker for others than themselves, researchers could ask if the reviewers also receive fitness trackers as gifts, whom they buy it for — their spouse, parents, siblings or children, and what is the fitness tracker that is most likely to be bought as a gift. To answer the ‘Why?’ questions for any of these, researchers would need to manually read the reviews and draw inferences.

Sentiment analysis posed certain limitations. For example, “I’ve used it on my runs, lifting, hikes” is assigned a neutral score. In reality, in the context of features, the sentence has positive connotations. At the same time, looking at the sentiment scores in conjunction with the lines with key words provided key information about the difference between positive and negative reviews, the difference in the time periods mentioned in the reviews of fitness trackers and pedometers, kind of language used, the scale of emotions expressed, whom people buy it for and how they are used as an aid to recovery, ideas around existing features and what people would like to have.

[1] Cecchini, Michele, Franco Sassi, Jeremy A. Lauer, Yong Y. Lee, Veronica Guajardo-Barron, and Daniel Chisholm. “Tackling of Unhealthy Diets, Physical Inactivity, and Obesity: Health Effects and Cost-effectiveness.” The Lancet 376, no. 9754 (2010): 1775–784. doi:10.1016/s0140–6736(10)61514–0.

[2] “Global Health Risks: Mortality and Burden of Disease … — WHO.” Accessed October 25, 2016. http://www.who.int/healthinfo/global_burden_disease/GlobalHealthRisks_report_full.pdf.

[3] King, Alexandra. “Public Health: Health Risks of Physical Inactivity Similar to Smoking.” Nature Reviews Cardiology 9, no. 9 (2012): 492. doi:10.1038/nrcardio.2012.115.

[4] Ibid

[5] http://www.forbes.com/sites/mattpowell/2015/01/12/sneakernomics-wearable-technology-and-sports-retail/#66c3449114e5

Inactivity

Risks of inactivity

Physical inactivity (PI) accounts for 1 in 10 deaths each year and was among the leading global risks for mortality in 2009[1],[2]. Physical inactivity leads to diseases such as obesity, heart disease, stroke, specific kinds of diabetes and the incidence of at least three types of cancer[3]. In spite of overwhelming proof, not enough is being done to address it. One in five adults across the world is considered physically inactive[4]. PI doesn’t just affect these individuals and has cascading effects on society as a whole. It impacts medical care costs, education, employment, income, earnings and wages, and crime[5]. PI is not limited to just certain sections of the population, though some may be more at risk than others. PI is difficult to address since the risks arise from not doing anything rather than doing the wrong thing. The key lies in motivating people to make small changes in their lifestyle and at the same time, making sure they have a holistic understanding of their health. The perception of risk is what underlies motivation.

A study conducted to measure the accuracy of health risk perceptions among obese individuals and those with normal weights found that they significantly underestimate their risks of developing arthritis or rheumatism, and hypertension within the following five years[6]. Another study found that though people did perceive an increased risk of mortality due to weight gain, they significantly underestimated the risks[7]. One possible reason for this ‘false optimism’ is the lack of actionable information from healthcare providers. Men, those who are not highly educated and those not suffering from diabetes mellitus are less likely to receive any advice from their doctors[8]. This leads to a vicious cycle in which people who need the information the most, don’t receive any and may consequentially adopt a fatalistic attitude. Even when the importance of physical activity is communicated, the means objective is emphasized rather than the fundamental objectives that give a broader perspective of health. Another explanation is that historically ‘activity’ as been framed as something positive, but ‘inactivity’ isn’t framed negatively. If PI isn’t framed as something that has a negative effect, it doesn’t have an effect on motivation. The affect heuristic and availability heuristic also mean that people perceive low risks with events that don’t create strong negative feelings and that don’t show results immediately[9] [10]. Because of the comparatively long term consequences of inactivity, the diseases are not associated with ‘dread’ which also impacts people’s willingness to take precautionary measures[11]. People also suffer from an optimism bias that leads them to believe that they are less at risk compared to others.

There are several approaches to combat the misperceptions of the risks of inactivity. One possible strategy is to create policies that make doctors accountable for providing all patients with the information needed to take care of their own health. Reframing physical activity as ‘preventative healthcare’ would also create a sense of responsibility among patients themselves. Campaigns that highlight how PI is a ‘silent killer’ accompanied with simple, low-commitment activities that people could incorporate through their day would increase awareness and motivation. Fitness and activity trackers are one such way of increasing people’s awareness and motivation to achieve daily goals that have a cumulative effect over a period of time, that bring a sense of the long term benefits into the present.

Fitness trackers

How they work to combat physical inactivity

Fitness trackers belong to a network of internet-enabled devices known as ‘the internet of things’ (IoT). The internet of things consists of devices that can independently communicate with each other without human intervention. Fitness trackers are hugely popular. In 2015, one in five American adults owned at least one wearable[12]. 78.1 million wearable devices were sold last year which is 171.6% more than the year before[13]. There are several possible reasons for its popularity. Fitness trackers are presented as a solution that encourages low commitment physical activity by continuously tracking movements. These trackers address some of the misperceptions of the risks of physical inactivity presented above. The information provided is directly and immediately actionable. Small activities like walking around the house, doing chores or sleeping are tracked to create a sense of achievement. The fundamental objectives of leading a longer and healthier life is made to seem achievable through everyday means objectives of, for example, walking 10,000 steps a day. The data is aggregated over time and visualized to make abstract ‘invisible’ information concrete. When worn, fitness trackers constantly remind people to take breaks from long periods of inactivity during the day. This frames inactivity beyond a certain limit as something to be avoided and creates a negative affect around it. Because these negative feelings are created around it and the results are visualized in real time, the consequences of inactivity are brought into short term consciousness and have a “dread” factor. Fitness trackers leverage a few cognitive biases but they also have certain limitations.

Risks of fitness trackers

Despite how compelling fitness trackers may seem as a way to combat physical inactivity, they pose significant security and privacy risks. More than 70% of cyber security and IT professionals believe that IoT manufacturers aren’t doing enough to maintain data security and that the standards need to be updated[14]. 84% of them think that the manufacturers don’t make consumers aware of the kinds of information the devices can collect[15].

The major security risks that experts have identified are around the following themes.

1. Identity theft by gathering personally identifiable information
2. Profiling to target or discriminate against people based on personal information or health activity
3. Locating people and stalking based on inferred patterns in location data
4. Embarrassment and extortion based on activities or capturing photos or videos
5. Corporate use and misuse of employees’ and customers’ data[16]

Wearable devices are constantly collecting personal data, often without the person realizing that data is being generated and shared constantly. Companies that make fitness trackers can collect and reveal users’ data without their explicit consent. For example, in 2011, Fitbit accidently made details of 200 users’ sexual activity public[17]. At a larger scale, malware can have catastrophic impacts on internet traffic. In October 2016, a distributed denial of service (DDoS) attack blocked traffic to major websites by taking over IoT devices.[18]

A study found that 86% of internet users had taken measures to protect themselves from unwanted surveillance while they were using the internet[19]. According to a study of 1,782 Internet users, the most common risks associated with owning a wearable device were videos or photos of them unclothed or that were otherwise embarrassing[20]. Since fitness trackers can’t collect this kind of data, people don’t associate them with high levels of risk. According to the same report, the least upsetting risks were exercise patterns, moods and emotions, heart rate and gender. These relate to the experts’ concerns about profiling, corporate misuse and drawing inferences based on patterns of activities. One possible explanation for the mismatch between public and expert perception is that people are more concerned about personal short term risks and find it difficult to imagine technologically advanced or complicated security threats at the network level. People tend to view each of their devices as stand-alone tools but when there is a network of ‘internet of things’, single data points from many individual devices or across a period of time can reveal a more complete picture of the customer than the customer herself realizes. There is also a lack of transparency between fitness tracker manufactures and the people who use fitness trackers. Most people do not know or understand what kind of data is collected, how often it’s tracked and shared across networks and how companies manage security and sharing of personal data with third parties. Since they don’t know the risks at each of these, security risks aren’t part of the purchase decision — cost, features, styling, brand, battery life and durability are more important factors for people to decide which, if any, fitness tracker to buy and use.

Wearables pose considerable security and privacy risks and there’s no easy resolution. Companies that create fitness trackers and collect the data sometimes don’t know what data they’re going to collect or what the patterns of big data will reveal. Their business model may be centered around ‘data innovation’. In these cases, it’s hard to explicitly state the privacy risks. Other than updating the software to include better data encryptions methods, tracking prevention and providing privacy from third party analytics, companies need to clearly communicate the risks within the privacy policy. Rather than terminating services to those consumers who don’t provide consent, a flexible product-service model can include features limited to those that have lower risks based on consumers; risk tolerance. Hardware features, like a physical switch for the Bluetooth to reduce continuous exposure to risks, should be incorporated too.

Weighing the privacy costs of fitness tracker against the health costs of being physically inactive

People should weigh the risks associated with physical inactivity and the (currently opaque) security risks of each device before buying one. Information about neither of these two topics is easily available to those who doesn’t know and don’t know that they don’t know. The decision making process is also difficult since these factors are not easily comparable, negotiable or even explicit. Previous research suggests possible approaches that people may take in making the trade-off. They relate to the means versus fundamental objectives, the cultural cognition theory and the effects of numerical and graphical display on professed risk-taking behavior.

Privacy and security risks arise as a by-product of digitally networked infrastructures. It can be considered to be a means through which fitness trackers operate. Fitness trackers are a means objective to resolve the (relatively more) fundamental objective to being more physically active. Being physically active is itself a means objective towards increased mortality. In this case, since physical activity is more ‘fundamental’ than fitness trackers, people are more likely to take it seriously than the risk of a by-product of a means objective. At the same time, context plays an important role. The cultural cognition theory posits that ‘individuals, as a result of a complex of psychological mechanisms, tend to form perceptions of societal risks that cohere with values characteristic of groups with which they identify’[21]. If a person is part of a group that values privacy over all else, and they are aware of the technological processes behind fitness trackers, the group (and the individual) are unlikely to use fitness trackers. If the group is more concerned about leading a healthy lifestyle or prides itself in being early adopters of technology, the individual is more likely to be enthusiastic about using fitness trackers even if they are aware of the risks.

According to Stone et al, “frequency information displayed as stick figures increased participants’ professed risk aversive behavior more than did frequency information given in numerical form”[22]. The risk taking behavior in this case would be ‘not being active’ and ‘using a fitness tracker’. Privacy risks are most often presented in textual or numerical form. Even when presented by popular media, they often take the form of articles, and are not really visualized. On the other hand, fitness trackers present data about physical activity as compelling and easy to understand visuals. This would lead people to perceive the risks of inactivity differently from that of privacy because of the form in which the information is presented. It is in the interest of the fitness tracking manufacturers to convert the data into graphical visuals so that users stay engaged but since privacy risks are almost overlooked in the broader interest of technological innovation, few people would be concerned about presenting it effectively.

According to the Precautionary Principle, “to justify regulation, a certainty of harm should not be required; a risk, even a low one, may well be enough.”[23] As Sunstein argues, applied in a strong sense, this can create a paralyzing situation in which no course of action can be taken. In this context, it would imply that because fitness trackers have significant risks, they shouldn’t be used. Even though the security risks may be more devastating than the risk of physical inactivity, the solution cannot be to put a blanket ban on all IoT devices. Instead, applying the precautionary principle in a weak sense, the privacy risks need to be regulated so that the cumulative costs (malicious DDoS attacks for example) aren’t higher than the apparent benefit (people losing weight or maintaining their health). These risks are difficult to balance and the key is in providing information that enables people to make their own subjective evaluations of the risks and make their own decisions. One way to break out of the sticky situation of having to negotiate these two large risks is to try to solve for the fundamental objective through other means.

Other ways to communicate the risks of physical inactivity and encourage behavior change

Often, a technological solution seems to be the best and most efficient but as shown in this paper, it has limitations — often these have more dangerous consequences than the problem they originally intended to solve. Technological solutions can also be unintentionally paternalistic and normative by oversimplifying human behavior and their judgement of risks. Knowing common biases helps get a glimpse into how people perceive certain risks and how that guides their behavior but in reality, human action consists of complex and constantly changing interactions and states of mind. Trying to tightly control behavior can backfire. Technological solutions often assume that people are inherently lazy and want to reduce any kind of cognitive load, and that they are deficient in their abilities hence make irrational decisions[24]. In reality, people weigh the costs and benefits of multiple non-homogeneous factors, along with the risks associated with physical inactivity and the risks of fitness trackers. That can makes the decision making process much harder since these factors are not easily negotiable or even explicit.

Fitness trackers create self-awareness by making ‘invisible’ activity visible through quantification. But doctors play an important role by providing earlier warnings, even before people think about using fitness trackers. This coupled with other communication that people face in their day to day routines would reinforce their knowledge of the risks of inactivity. For example, a study found that reading nutritional labels does have a positive impact among the obese[25] . Self-learning can be encouraged through reminders to stay active but for some people, constant reminders can be demotivating and accentuate the lack of control their lack of control over their health and schedule. Fitness trackers also create a reward system that generates extrinsic motivation at the cost of more powerful intrinsic motivation. Self-contained technological solutions seem like an easy fix compared to larger scale policy or infrastructural changes but those can prove be more valuable and cost-effective in the long run. A study found that the amount of activity youth got was related to the walkability, the density of cul-de-sacs and park space in their neighborhoods[26]. Providing supportive infrastructures leads to the establishment of healthy intrinsic practices rather than just trying to restrict ‘unhealthy’ habits through increased taxation or restricting availability (which may actually increase desirability). Communicating risks should not be restricted to the single touchpoint of a doctor’s clinic or a fitness tracker but needs to be a holistic and continuous approach over time that takes people’s own preferences into account and gives them the information to make their own decisions.

Conclusion

Physical inactivity has significant and widespread effects and the risks are often misperceived by people. This is because of doctors’ biases, the emphasis on achievable means objectives and how activity and inactivity have been framed historically. The affect and availability heuristics, an optimism bias and the lack of “dread risk” associated with PI has an impact on how seriously people view the risks. Involving doctors early on and encouraging them to have the right conversation with all their patients who need their help and reframing physical inactivity as preventative healthcare would be a good start. Communicating the fundamental objective of a healthy and long life, reframing physical inactivity negatively rather than just physical activity positively and making the long term consequences visible in the present are crucial to motivate people and help them make better decisions. Fitness trackers are a possible solution but they have security risks. These are also misperceived by the public because of the technical knowledge needed to anticipate and understand the risks of the larger picture, the lack of information shared by fitness tracker manufacturers. As a result, the security risks aren’t a large part of the purchase decision. Solutions to this problem include regulation and two-way communication and explicit consent of the users that trust companies with their data and the companies that monetize personal data. There are several non-technological solutions to the basic problem of physical inactivity. Adding daily reminders in the everyday environments of people, improving physical infrastructure and facilitating richer conversations in doctors’ clinics are less didactic and more effective approaches. Ultimately, technologists, designers and policymakers can help people’s decision-making process by creating environments for self-awareness, self-learning, intrinsic motivation and creating conditions for them to achieve their goals.

[1] “Global Health Risks: Mortality and Burden of Disease … — WHO.” Accessed October 25, 2016. http://www.who.int/healthinfo/global_burden_disease/GlobalHealthRisks_report_full.pdf.

[2] King, Alexandra. “Public Health: Health Risks of Physical Inactivity Similar to Smoking.” Nature Reviews Cardiology 9, no. 9 (2012): 492. doi:10.1038/nrcardio.2012.115.

[3] Cecchini, Michele, Franco Sassi, Jeremy A. Lauer, Yong Y. Lee, Veronica Guajardo-Barron, and Daniel Chisholm. “Tackling of Unhealthy Diets, Physical Inactivity, and Obesity: Health Effects and Cost-effectiveness.” The Lancet 376, no. 9754 (2010): 1775–784. doi:10.1016/s0140–6736(10)61514–0.

[5] Dumith, Samuel C., Pedro C. Hallal, Rodrigo S. Reis, and Harold W. Kohl. “Worldwide Prevalence of Physical Inactivity and Its Association with Human Development Index in 76 Countries.” Preventive Medicine 53, no. 1–2 (2011): 24–28. doi:10.1016/j.ypmed.2011.02.017.

[6] Cawley, John, and Christopher Ruhm. “The Economics of Risky Health Behaviors.” 2011. doi:10.3386/w17081

[7] Winter, Joachim, and Amelie Wuppermann. “Do They Know What Is At Risk? Health Risk Perception Among The Obese.” Health Economics 23, no. 5 (2013): 564–85. doi:10.1002/hec.2933.

[8] Galuska, Deborah A. “Are Health Care Professionals Advising Obese Patients to Lose Weight?” Jama 282, no. 16 (1999): 1576. doi:10.1001/jama.282.16.1576.

[9] Kahneman, Daniel, Paul Slovic, and Amos Tversky. Judgment under Uncertainty: Heuristics and Biases. Cambridge: Cambridge University Press, 1982.

[10] Finucane, Melissa L., Ali Alhakami, Paul Slovic, and Stephen M. Johnson. “The Affect Heuristic in Judgments of Risks and Benefits.” Journal of Behavioral Decision Making 13, no. 1 (2000): 1–17. doi:10.1002/(sici)1099–0771(200001/03)13:13.0.co;2-s.

[11] Slovic, P. “Perception of Risk.” Science 236, no. 4799 (1987): 280–85. doi:10.1126/science.3563507.

[12] Comstock, By Jonah. “PwC: 1 in 5 Americans Owns a Wearable, 1 in 10 Wears Them Daily.” MobiHealthNews. 2014. Accessed September 28, 2016. http://www.mobihealthnews.com/37543/pwc-1-in-5-americans-owns-a-wearable-1-in-10-wears-them-daily.

[13] “The Worldwide Wearables Market Leaps 126.9% in the Fourth Quarter and 171.6% in 2015, According to IDC.” Www.idc.com. Accessed September 28, 2016. http://www.idc.com/getdoc.jsp?containerId=prUS41037416.

[14] @threatintel. “How Safe Is Your Quantified Self? Tracking, Monitoring, and Wearable Tech.” Symantec Security Response. Accessed September 28, 2016. https://www.symantec.com/content/dam/symantec/docs/white-papers/how-safe-is-your-quantified-self-en.pdf

[15] Ibid.

[16] Ibid.

[17] @ChrisMatyszczyk, By Chris Matyszczyk. “TMI? Some Fitbit Users’ Sex Stats on Google Search.” CNET. 2011. Accessed September 28, 2016. https://www.cnet.com/news/tmi-some-fitbit-users-sex-stats-on-google-search/.

[18] http://www.usatoday.com/story/tech/2016/10/21/cyber-attack-takes-down-east-coast-netflix-spotify-twitter/92507806/

[19] Rainie, L., Kiesler, S., Kang, R., & Madden, M. (2013). Anonymity, privacy, and security online Pew Internet & American Life Project. Retrieved from http://search.proquest.com/docview/1497406302?accountid=9902

[20] Lee, L., Lee, Joong H., Wagner, D., Egelman, S. “Risk Perceptions for Wearable Devices” arXiv:1504.05694 [cs.CY]

[21] Kahan, Dan M., Ellen Peters, Maggie Wittlin, Paul Slovic, Lisa Larrimore Ouellette, Donald Braman, and Gregory Mandel. “The Polarizing Impact of Science Literacy and Numeracy on Perceived Climate Change Risks.” Nature Climate Change 2, no. 10 (2012): 732–35. doi:10.1038/nclimate1547.

[22] Stone, Eric R., J. Frank Yates, and Andrew M. Parker. “Effects of Numerical and Graphical Displays on Professed Risk-taking Behavior.” Journal of Experimental Psychology: Applied 3, no. 4 (1997): 243–56. doi:10.1037//1076–898x.3.4.243.

[23] Sunstein, and Cass R. “The Paralyzing Principle: Does the Precautionary Principle Point Us in Any Helpful Direction? (Risk).” Regulation, December 22, 2002.

[24] Life would be pretty dull without risk: voluntary risk0taking and its pleasures By: Lupton, Tulloch

[25] Loureiro, Maria L., Steven T. Yen, and Rodolfo M. Nayga Jr. “The Effects of Nutritional Labels on Obesity.” Agricultural Economics 43, no. 3 (2012): 333–42. doi:10.1111/j.1574–0862.2012.00586.x.

[26] Laxer, Rachel E., and Ian Janssen. “The Proportion of Youths’ Physical Inactivity Attributable to Neighbourhood Built Environment Features.” International Journal of Health Geographics Int J Health Geogr 12, no. 1 (2013): 31. doi:10.1186/1476–072x-12–31.

The research referenced here focuses on obesity as a consequence of inactivity but the implications could apply to other diseases caused due to inactivity as well. A study conducted to measure the accuracy of health risk perceptions among obese individuals found that they significantly underestimate their risks of developing arthritis or rheumatism and hypertension within the following five years while overestimating relatively minor risks of getting a heart attack or stroke[7]. The same study conducted with individuals with normal weights showed similar patterns of underestimation and overestimation. Another study looked at the relationship between weight and perceived health risk. They found that though people did perceive an increased risk of mortality due to weight gain, they significantly underestimated the risks. This was referred to as ‘false optimism’[8]. Though people are aware of the relation between PI and morbidity as well as mortality, there is a mismatch. Some possible causes for this false optimism are presented below.

Information is shared unequally
Beside the lack of concrete information in the public media, even doctors often don’t advise their obese patients. A study found that less than half of obese adults received advice from their physician on losing weight[9]. Healthcare providers are an important source of information because those who had been told to lose weight were almost 3 times as likely to try losing weight as those who had not. Those suffering from diabetes mellitus, women, and people who were highly educated did receive more advice from their doctors. This may be because the doctors advise weight loss only when the patient is suffering from a serious medical condition, if they feel that the patient can make lifestyle changes and if they have increased interactions with the patient. This creates a vicious cycle where people who don’t fall into the above three categories are left out of the system of self-knowledge, information and advice. Awareness of their own biases would encourage doctors to dispense the optimum amount of information and advice to all of their patients.

Compelling forms of communication
Means objective are those that are important to people only because of its ‘implications for the degree to which another (more fundamental) objective can be achieved.’[10] Information provided to patients is often in the form of means objectives — things they need to do in the short term. The fundamental objectives are harder to quantify and measure but if patients are provided information about fundamental objectives, it would lead to a better perception of the risks of inactivity. For example, instead of just telling people that they need to walk 10,000 steps a day, communication should be framed around fundamental objectives of morbidity and mortality- that the goal is not just to walk 10,000 steps a day but to increase their quality of life and longevity.

Framing the communication for action
People believe that physical activity has health benefits. This is true but frames exercise as a way to achieve better health. This implies that not exercising is equated with the state of one’s health remaining at status quo. This framing results in people exercising less regularly. Recent research has revealed that it’s not just that exercising is good for health but that inactivity is detrimental to health. This is especially true for those suffering from preexisting health conditions. There have been some efforts to change the frame of exercise and inactivity. ‘Exercise is medicine’ is an Australian initiative to provide resources to physical care providers to get them to increase their patients’ physical activity.[11] The reframing of activity as ‘care’ in gyms and running groups to ‘preventative care’ or ‘physical rehabilitation’ advocated by medical practitioners would lead to an increase in people’s physical activity.

Framing the communication for higher risk perception

Affects are general feelings that people experience and also specific qualities associated with a particular stimulus. According to the affect heuristic, people evaluate the risks and benefits of an action by accessing a pool of positive and negative feelings they have about the action.[12] When the conversation about PI is framed around exercise as enjoyment, the feelings generated are likely to be positive. This will lead to people perceiving the risks to be low and the benefits high. Instead, if the conversation is framed around the illnesses caused by inactivity, the resulting negative affect will reflect as high risk and low benefit. In this case, a perception of high risk and negative affect will motivate behaviour change.

Making the long term effects concrete in the present
According to the availability heuristic, ‘people assess […] the probability of an event by the ease with which instances or occurrences can be brought to mind’[13]. People’s judgement is influenced by what hazards receive public attention and are consequently more readily ‘available’ through memory or imagination. Thus the frequencies of dramatic events such as homicide, or multiple-death catastrophes, which the media focuses on more, would be overestimated, while the impact of ‘quiet killers’ will be underestimated[14]. Inactivity is not ‘catastrophic’ and is one such quiet killer that isn’t composed of several distinct and lethal events but builds up to several significant health issue over the course of a lifetime

Communicating catastrophic qualities

Following Slovic’s scale of unknown and dread risks, people would perceive PI as a high unknown risk since the effect is delayed, it’s not observable in the short term, and it’s relatively unknown to those exposed[15]. But people would perceive it as a low dread risk because they feel can be controlled by individuals’ decisions, is voluntary and that it’s not a catastrophic event. In reality, they lack adequate information and a big picture, long term view. PI can be controlled but it’s not that easy to control. It’s voluntary but there are systemic barriers (like access to infrastructure, time and money) that determine individuals’ choices implicitly. It has evolved to be a global catastrophic problem over a short period of time. The consequences are fatal and the high risks to future generations that can’t be easily reduced. The current perception of low dread is misplaced and affects why people incorrectly perceive the risks of PI.

The adverse effects of prolonged physical inactivity are much worse than previously thought. It impacts the quality of life and mortality of a significant part of the global population. Furthermore, PI causes cascading socio-economic problems. Despite overwhelming evidence against physical inactivity, people don’t perceive the risks accurately. This is true for those who suffer from health conditions related to inactivity and those who don’t currently but may be at risk in the future. We need to analyze why these perceptions are formed so that we can design preventative measures. One possible reason for the mistaken perception is the lack of actionable information from healthcare providers. When the information is communicated, the means objective is emphasized rather than the fundamental objectives. Another explanation is that historically ‘activity’ as been framed as something positive, but ‘inactivity’ isn’t framed negatively. This adversely affects motivation. The affect heuristic and availability heuristic mean that people perceive low risks with events that don’t create strong negative feelings and that don’t show results immediately. Because of the comparatively long term consequences of inactivity, it’s not associated with ‘dread’ which also people’s willingness to take action.

Creating policies that make doctors more accountable for treating their patients equally and providing them with the information to take care of their own health could be a solution. Re-framing physical activity as ‘preventative healthcare’ would also create a sense of responsibility among patients themselves. Campaigns that highlight how PI is a ‘silent killer’ accompanied with simple, low-commitment activities that people could incorporate through their day would increase awareness and motivation. There may be a need for government interventions or regulations around information disclosure at workplaces where people are sedentary for long amounts of time. Infrastructure plays a role in practices around activity too. A study found that the amount of activity youth got was related to the walkability, the density of cul-de-sacs and park space in their neighborhoods[16]. Motivating people is difficult; it’s much easier to get patients to comply with taking medicines once a day than build a practise of activity. Changing people’s perception of risk is an important part of intrinsic motivation. Fighting against the global epidemic of physical inactivity is not just the responsibility of individual patients, doctors or fitness coaches but of urban planners and developers too.

[1] Cecchini, Michele, Franco Sassi, Jeremy A. Lauer, Yong Y. Lee, Veronica Guajardo-Barron, and Daniel Chisholm. “Tackling of Unhealthy Diets, Physical Inactivity, and Obesity: Health Effects and Cost-effectiveness.” The Lancet 376, no. 9754 (2010): 1775–784. doi:10.1016/s0140–6736(10)61514–0.

[2] “Global Health Risks: Mortality and Burden of Disease … — WHO.” Accessed October 25, 2016. http://www.who.int/healthinfo/global_burden_disease/GlobalHealthRisks_report_full.pdf.

[3] King, Alexandra. “Public Health: Health Risks of Physical Inactivity Similar to Smoking.” Nature Reviews Cardiology 9, no. 9 (2012): 492. doi:10.1038/nrcardio.2012.115.

[4] Ibid

[5] Dumith, Samuel C., Pedro C. Hallal, Rodrigo S. Reis, and Harold W. Kohl. “Worldwide Prevalence of Physical Inactivity and Its Association with Human Development Index in 76 Countries.” Preventive Medicine 53, no. 1–2 (2011): 24–28. doi:10.1016/j.ypmed.2011.02.017.

[6] Cawley, John, and Christopher Ruhm. “The Economics of Risky Health Behaviors.” 2011. doi:10.3386/w17081

[7] Winter, Joachim, and Amelie Wuppermann. “Do They Know What Is At Risk? Health Risk Perception Among The Obese.” Health Economics 23, no. 5 (2013): 564–85. doi:10.1002/hec.2933.

[8] Falba, Tracy A., and Susan H. Busch. “Survival Expectations of the Obese: Is Excess Mortality Reflected in Perceptions?” Obesity Research 13, no. 4 (2005): 754–61. doi:10.1038/oby.2005.85.

[9] Galuska, Deborah A. “Are Health Care Professionals Advising Obese Patients to Lose Weight?” Jama 282, no. 16 (1999): 1576. doi:10.1001/jama.282.16.1576.

[10] Huynh, Candice H., and Jay Simon. “Using Means Objectives to Present Risk Information.” Decision Analysis 13, no. 2 (2016): 117–27. doi:10.1287/deca.2015.0328.

[11] Coombes, J. S., J. Law, B. Lancashire, and R. G. Fassett. “”Exercise Is Medicine”: Curbing the Burden of Chronic Disease and Physical Inactivity.” Asia-Pacific Journal of Public Health 27, no. 2 (2013). doi:10.1177/1010539513481492.

[12] Finucane, Melissa L., Ali Alhakami, Paul Slovic, and Stephen M. Johnson. “The Affect Heuristic in Judgments of Risks and Benefits.” Journal of Behavioral Decision Making 13, no. 1 (2000): 1–17. doi:10.1002/(sici)1099–0771(200001/03)13:13.0.co;2-s.

[13] Kahneman, Daniel, Paul Slovic, and Amos Tversky. Judgment under Uncertainty: Heuristics and Biases. Cambridge: Cambridge University Press, 1982.

[14] Lichtenstein, Sarah, and Et Al. “Judged Frequency of Lethal Events.” Journal of Experimental Psychology: Human Learning & Memory 4, no. 6 (1978): 551–78. doi:10.1037//0278–7393.4.6.551.

[15] Slovic, P. “Perception of Risk.” Science 236, no. 4799 (1987): 280–85. doi:10.1126/science.3563507.

[16] Laxer, Rachel E., and Ian Janssen. “The Proportion of Youths’ Physical Inactivity Attributable to Neighbourhood Built Environment Features.” International Journal of Health Geographics Int J Health Geogr 12, no. 1 (2013): 31. doi:10.1186/1476–072x-12–31.

Experts have established that wearables still pose significant security risks in terms of technology(3). The major themes are -

1. Identity theft by gathering personally identifiable information

2. Profiling to target or discriminate against people based on personal information or health activity

3. Locating people and stalking based on inferred patterns in location data

4. Embarrassment and extortion based on activities or capturing photos or videos

5. Corporate use and misuse of employees’ and customers’ data(4)

Wearable devices are a particular kind of products that belong to a larger network called ‘the internet of things’ or IoT. More than 70% of cyber security and IT professionals believe that IoT manufacturers aren’t doing enough to maintain data security and that the standards need to be updated. 84% of them think that the manufacturers don’t make consumers aware of the kinds of information the devices can collect(50. Wearable devices are constantly collecting personal data, often without the person realizing that data is being generated and shared constantly. Companies that make fitness trackers can collect and reveal users’ data without their explicit consent. In 2011, Fitbit accidently made details of 200 users’ sexual activity public(6).

A study found that 86% of internet users had taken measures to protect themselves from unwanted surveillance while they were using the internet(7). Personal privacy on the internet is a concern for the average citizen and they take considerable precautions to protect themselves(8). But how does people’s perception of risks on the internet translate to perceived risks of newer technologies like wearable devices? According to a study of 1,782 Internet users, the most common risks associated with owning a wearable device were also ‘privacy’ (25.32%) and ‘being unaware’ (15.40%) (9). ‘Privacy’ referred to privacy and security concerns similar to the experts’ findings above. ‘Being unaware’ included ‘being unaware of what the device is collecting, doing or which information it is using’. These concerns broadly matched those that the experts had, despite the gap in technical knowledge and specifics of the how the technology works.

‘Loss of privacy’ is a very broad statement. On closer analysis of the study, the risks that people found most upsetting were videos or photos of them unclothed or that were otherwise embarrassing. Experts think of this as a possible risk too. The second kind of risk was related to personal financial information like bank account details, social security number and credit card information. This relates to the first point raised by security experts — risk of identity theft. What is interesting is that the least upsetting risks were exercise patterns, moods and emotions, heart rate and gender. These clearly relate to the experts’ concerns about profiling, corporate misuse and drawing inferences based on patterns of activities.

In the study, participants’ risk perception of 72 risks were measured in relation to each other. Participants might have responded differently if they were asked to rate the risks individually or in smaller sets. The study doesn’t investigate the reason for people’s perception that some things are riskier than others. One possible explanation is that people are more concerned about the short term risks and find it difficult to imagine security threats that are more technologically advanced or complicated. People tend to view each of their devices as stand-alone tools but when there is an ‘internet of things’ single data points from many individual devices or across a period of time can reveal a more complete picture of the customer than the customer herself realizes.

Security risks aren’t likely to be a key factor of a purchase decision — features, styling, brand, battery life and durability are more important for people. They may not know or understand aspects like what kind of data is collected, how often it’s tracked, how companies manage security and sharing of personal data with third parties. Furthermore, the actual risk probabilities are unknown since they are more or less opaque in our technological landscape. Further studies need to be done in this particular area of technology to understand how exactly risks are being perceived at a granular level in a dynamic context and how the benefits are weighed against it.

The authors admit that 83% of their participants didn’t own a wearable. Though this distribution is consistent with the national figures, it can be argued that people who do own wearables have a different relationship with them and negotiate the risks differently over time. People who invest in buying fitness trackers do so to get healthier. They have a vested interest in making it work for them and this can cause them to overlook privacy risks. Once they’ve established a habit with the fitness trackers, they could start viewing it as a social actor(10) causing them to further discount the nature of the ecosystem of technologies that makes their data vulnerable to attack. Neither does the study probe into understanding if the owners of wearables take steps to safeguard their privacy while using them.

Though wearables pose considerable security and privacy risks. It’s not an easy resolution since companies that create fitness trackers and collect the data sometimes don’t know what data they’re going to collect or what the patterns of big data will reveal. Their business model may be centered on ‘data innovation’. In these cases, it would be hard to explicitly state the privacy risks. Collecting personal information, locations, habits and activities may not reveal much in isolation but when analysed in conjunction and with other data sets, could be surprisingly informative. Companies might sell this data to third parties who use it in combination with other datasets to make credit, insurance and employment decisions. This raises a larger policy-level question about ownership of data — does it belong to the consumer who creates it, the company that collects it, those who aggregate and analyses it or those who invest resources to store it?

Other than updating the software to include better data encryption methods, tracking prevention and providing privacy from third party analytics(11), companies need to clearly communicate the risks within the privacy policy. This communication shouldn’t just be in the form of statistics or probabilities, since those have their limitations(12). When subsequent changes are made that affect the risks, consumers should be asked for explicit consent. Rather than terminating services to those consumers who don’t provide consent, features could be limited to those that have lower risks based on the consumer’s risk tolerance. Hardware features, like a physical switch for the Bluetooth to reduce continuous exposure to risks, could be incorporated.

There are significant challenges and so far the issues have remained in the domain of technologists and cyber-security experts. As designers and policy makers it’s imperative to understand how people’s relationship with a personal device informs their perception of the actual risks, not so that we can take advantage of it but so that we can communicate risks better and design to reduce actual and perceived risks.

Bibliography

1. Comstock, By Jonah. “PwC: 1 in 5 Americans Owns a Wearable, 1 in 10 Wears Them Daily.” MobiHealthNews. 2014. Accessed September 28, 2016. http://www.mobihealthnews.com/37543/pwc-1-in-5-americans-owns-a-wearable-1-in-10-wears-them-daily.

2. “The Worldwide Wearables Market Leaps 126.9% in the Fourth Quarter and 171.6% in 2015, According to IDC.” Www.idc.com. Accessed September 28, 2016. http://www.idc.com/getdoc.jsp?containerId=prUS41037416.

3. Goyal, Rohit, Nicola Dragoni, and Angelo Spognardi. “Mind the Tracker You Wear.” Proceedings of the 31st Annual ACM Symposium on Applied Computing — SAC ’16, 2016. doi:10.1145/2851613.2851685.

4. @threatintel. “How Safe Is Your Quantified Self? Tracking, Monitoring, and Wearable Tech.” Symantec Security Response. Accessed September 28, 2016. https://www.symantec.com/content/dam/symantec/docs/white-papers/how-safe-is-your-quantified-self-en.pdf

5. “Press Release.” ISACA Survey: Wide Gap Between Consumers’ and IT Professionals’ Perceptions on Internet of Things Security. Accessed September 28, 2016. http://www.isaca.org/About-ISACA/Press-room/News-Releases/2015/Pages/ISACA-Survey-Wide-Gap-Between-Consumers-and-IT-Professionals-Perceptions-on-Internet-of-Things-Security.aspx.

6. @ChrisMatyszczyk, By Chris Matyszczyk. “TMI? Some Fitbit Users’ Sex Stats on Google Search.” CNET. 2011. Accessed September 28, 2016. https://www.cnet.com/news/tmi-some-fitbit-users-sex-stats-on-google-search/.

7. Rainie, L., Kiesler, S., Kang, R., & Madden, M. (2013). Anonymity, privacy, and security online Pew Internet & American Life Project. Retrieved from http://search.proquest.com/docview/1497406302?accountid=9902

8. “Business Week/Harris Poll: A Growing Threat.” Bloomberg.com. Accessed September 28, 2016. http://www.bloomberg.com/news/articles/2000-03-20/business-week-harris-poll-a-growing-threat.

9. Lee, L., Lee, Joong H., Wagner, D., Egelman, S. “Risk Perceptions for Wearable Devices” arXiv:1504.05694 [cs.CY]

10. Fogg, B. (2003). Computers as persuasive social actors. Persuasive Technology, 89–120. doi:10.1016/b978–155860643–2/50007-x

11. Gigerenzer, G., Hertwig, R., Broek, E. V., Fasolo, B., & Katsikopoulos, K. V. (2005). “A 30% Chance of Rain Tomorrow”: How Does the Public Understand Probabilistic Weather Forecasts? Risk Analysis, 25(3), 623–629. doi:10.1111/j.1539–6924.2005.00608.x

12. Wells, G. L. (1992). Naked statistical evidence of liability: Is subjective probability enough? Journal of Personality and Social Psychology, 62(5), 739–752. doi:10.1037//0022–3514.62.5.739

Once it was all up, everyone added their names to the post it they thought was most interesting.

Team formation was followed by a presentation by Kakee

Analysing Social Practice theory from the lenses of image, stuff, skills and the relationships between them leads to a complex network of intersection connections. Laying this over a Multi-Level Perspective mao over time would surface ‘patterns of practices’. Practices used as units of analysis could be used to redefine practices as units of design — ‘the dance of design and practices’

Each group then moved to creating their own MLPs for their target areas.

The final maps followed the same grid with the same elements on the X and Y axes but the questions that emerged as a result of the exercise were different group to group.

March 4, Monday: Theories of Change: Social Practice Theory and Case Study Assignment was originally published in Transition Design on Medium, where people are continuing the conversation by highlighting and responding to this story.

Week 1

Right after a 16 hour flight from halfway across the world, I rushed into two back to back classes and then was informed that there was a team meeting over dinner when all I wanted to do was sleep.

But the pho at Tan Lac Vien did help me recover from jet lag I think.

Surprisingly, I hadn’t worked with Catherine, Kate or Saumya in the previous semester so I was happy to get a chance to work with them. Based on class discussions, I know that there are certain areas that we agree upon and cetain points of view that we hold common. But I also know that there are some things where we have very strong differing opinions.

Making a contract was a great way for us to get to know each other’s working styles, areas of strength and possible points of conflict. We’re surprisingly well balanced as a team though and it’s great that everyone wants to use this project as a way to move slightly out of their comfort zones.

The project brief is still not fully crystallized in my head though but that’s always the case this early on I think.

Week 2

To make sure that we didn’t misinterpret or miss anything in the brief, we deconstructed it and created a map of questions, interesting areas and possible research areas. This was really helpful to ease into the territory definition.

We explored a couple of different areas — home, retail, workplace, education, healthcare. Almost by a process of elimination rather than informed decision making, we went to autonomous cars. I’ve come to realize that I’m eager to move onto the next step as soon as the current one seems resolved but I’ve realized that others on the team like more time to reconsider the options and I need to make an effort to be patient, considerate and encouraging of that.

Working on the territory map wasn’t terribly difficult since we all are excited about the topic and we seemed to have narrowed it down quite a bit. There’s so much scope for human-computer symbiosis and conversational interfaces in autonomous cars. Plus, since we are exploring it through the ‘softer’ side of trust and comfort, I think it makes for an interesting space for user research (probably a completely different kind of user research from that happening on the other side of campus).

I’m slightly concerned about the ethical aspect of it. By designing a way to make autonomous cars more comfortable, are we encouraging it? Do I personally want people to be completely trusting of autonomous cars? Will they ever be that safe? We’re convincing ourselves and others by saying it’s inevitable so we might as well make it a smoother transition to autonomous cars.

Week 3

I think with every week, I get more excited about the project.

There’s so much stuff to look into and people to talk to!
Just today Catherine and I were talking to a PhD student who’s working on the autonomous car project at CMU and one thing he said struck me as particularly interesting -

When he first sat in a self driving car, he felt extremely uncomfortable in the car since he wasn’t in control. The steering wheel turning on its own was the most scary part since he didn’t know when it was going to stop. This is despite him knowing how to read the visual graphics that are shown on the screen. But after he wrote the code and algorithms for a component and they were testing that out in the car, he felt perfectly fine since he knew that the code was good.

It’s been great talking with experts in the field — both designers and scientists. We’re still trying to wrap our heads around a target user group. I’m partial to a general population that has been rendered immobile due to health, age, financial status and what it means for them to have access to this car.

The second thing we’re thinking of is the car as an environment — besides verbal, what other kinds of conversations and media could there be?

Plus, I’m excited about prototyping these experiences!

Week 4

Realizing that we needed to narrow out focus, we built 3 personas and scenarios each.

Persona/Scenario 1:
Elderly man looking at the car as mobility
Robert picks up a phone and asks for an autonomous car for the day. When it arrives at his house, he gets a phone call telling him it’s ready and where it’s parked (NLP used for 2 way communication). He gets in and says where he wants to go. With him in the drivers seat, the car starts moving. Sensing that Robert might fall asleep during the long ride, music starts. Robert calls his friend from the car and they chat about another one of their friends. Sensing this, the car joins in the conversation and asks if he would like to go visit that friend too. Robert and his friend think that’ll be a great idea.

Persona/Scenario 2:
Mom with kids looking at the car as a workspace
Kara is always stressed about driving her kids around. She and her children have a crazy schedule and rarely get a chance to spend quality time with each other. With the new autonomous car they’ve bought, Kara hopes to be able to use the time travelling to have more meaningful conversations with her kids. She discusses homework assignments and they talk through it so that by the time they get home, the kids can finish it up.

Persona/Scenario 3:
Busy professional looking at the car as a relaxing space

We finally decided to focus on EMTs’ experience in an autonomous ambulance. We had a great first interview and I’m glad that the entire team was able to make it. We captured lots of notes (all over the glass walls as well). And figured out the broad problem spaces. There’s a lot of them and through the rest of the interviews, we tried to confirm them and narrow down the scope. We haven’t been able to talk with too many EMTs but on the positive side, because we’re also talking with doctors and ER nurses, we have a much more holistic view of the entire system and possible intervention points. There’s definitely a very strong use case for hands free, conversation based decision help.

Week 5

Preparing the presentation was an interesting experience. We certainly gave ourselves more time to work on it and build a story. The session with Jon Zimmerman was helpful in figuring it out — as designers we get really caught up with the intricacies of the arc of the research. Because it’s our insights are a revelation to us, we assume that how we got there is critical too. But in this presentation we really tried to focus on building a scenario. Interestingly, when constructing the slides, we felt we were being repetitive but the feedback we got was that there was a bit of a leap from the research to the design insights. From the class discussions that followed, I think we were trying to focus on building a case for autonomous ambulances (which we needn’t have done) and for CUI and human computer symbiosis. I’m glad we decided not to talk too much about our initial general problem space. We also worked on scripts (which I personally think works great because it allows for room to improvise rather than fumbling over communicating the basics).

I am a little concerned about the added layer of complexity with autonomous ambulances. We now need to extrapolate very specific problems based on the current situation, imagine a future state with autonomous vehicles and anticipate and solve problems with that added layer of communication.

Week 6

This week I’ve been pondering about applying transition design to the studio project — aspects of theories of change, principles of living systems and the one that intrigues me the most-futuring. I don’t know how to synthesize the two classes. I ‘get’ both but for some reason, they seem to exist in different parts of my brain and I’m having a hard time applying it directly.

In one of the videos in our learning module, Peter Schwartz talks about the importance of having a vision and yet being willing to accommodate that your initial hypothesis was wrong and that it’s much more expensive to stick with a wrong decision.

For the project we’ve spent a lot of time scoping down and deciding on the nature of the problems that we want to tackle — what does it mean to design with human-computer symbiosis v/s for it.

We’ve also been having a hard time scheduling appointments for the workshops but it looks like now we’re finally having potential participants responding.

Week 7

Week was crazy and not very productive for the project and team. Because Confluence. But we did plan out a workshop and piloted it with the engineer we had spoken with earlier. The collaging exercise to find characteristics of an ideal team mate was surprisingly successful. The mapping of the task flow according to a journey map was less helpful because he assumed or hoped that it would all be completely automated. But the flexible modelling gave us some good ideas. It might make more sense to incorporate role playing with flexible modeling for EMTs.

Week 8

Over the weekend Ideo conducted a prototyping workshop. My key takeaways from that were — protoype from day 1, prototype quick and dirty, deep dive into a few key interaction points and work on them iteratively till the micro-interaction stage. I also realized how powerful video documentation of these quick and dirty micro interactions is. Just after the workshop, 3 of us were super inspired to generate early ideas.

We had 2 generative research workshops after that,where we got to probe a bit around the ideas we had. Doing a workshop with 3 participants and 4 facilitators is hard! But it makes the synthesis much easier and less convergent (which may or may not be a good thing 12 hours before a presentation).

For me it’s a little hard right now to not think of the design outcome as 2 levels — tactical low hanging fruit like fixing the current forms of the interface and completely futuristic networked infrastructure systems.

We also had a great debriefing session today where we came up with a potential concept direction related to ubiquitous computing and smart fabrics.

Week 9 — Spring Break

Week 10

In a larger group, it’s harder to manage roles and responsibilities. Of course, scheduling time to meet is difficult too but that shouldn’t get in the way of basic things like documentation. I do have a tendency to be take on the ‘organizer’ role and was trying to stay away from it but I think that’s exactly what our team needs. So we’ve decided on a more distributed documentation system, we have agendas for every meeting and we know what we’re doing in the next week.

Thinking about the worst case and best case scenarios really helped us ideate. We found that our key themes were naturally reflected in the problems and solutions we noted down.

Later, we went through the class’ feedback on our presentation and debriefed about the ridealongs. Lots of interesting areas — especially the one about transition spaces and the feeling of brotherhood came up. The next time we met, we re-looked at our insights and implications in the last presentation to see where the nuances were and what areas were not as important as our participants in the generative research said they were. Going back and forth between the last couple of weeks made sure that we didn’t miss anything.

We had a wide range of ideas, scenarios and roles that the CUI could take on. I’m still a bit concerned that we’re imagining the CUI as another ‘being’ or ‘computer’ and not as a way of communicating with an existing one. It’s a tricky space given the brief and though it is open to interpretation, I think we need to be aware of that when we frame our final presentation.

Week 11

We had a very informative session with our participants. Surprisingly what we had thought would work (stress relief), weren’t as easily accepted. They were much more focused on tangible, near future solutions to their problems. I’m thinking about this in general about user-centered design. It’s so compelling and tempting to solve problems but I’m starting to think that designer should actually play a much bigger role than just ‘fixers’ of a broken system. There may be a way in which design activities are framed and communicated to participants, or just about which phases users are involved in. If the same users are involved from the exploratory to the validation phases, they do expect to see their ‘solutions’ worked on but as designers, we may decide to extrapolate from their solutions and re-frame problems.

Adding to this complexity is the fact that with technology, we already know what the solution should do and also need to reframe problems to fit the solutions while at the same time expanding the scope of the solution presented.

I think we’re on track for the next deliverable, and I’m looking forward to this non-linear discovery and experimentation phases.

Week 12

We haven’t been too productive during the week this time. We did have an interesting time prototyping conversations and realized some nuances while roleplaying — for me, the biggest discovery was the importance of eye contact during a conversation and creating non-verbal cues. When you’re having a ‘conversation’ with Siri, it’s just you and the phone. Siri is also reactive rather than proactive. This relegates her to the role of a secretary. A proactive assistant should know when to interrupt (and how to interrupt) and when to recede (but still show that she is listening).

Another approach is to not try to make the computer conform to our social norms. It’s a hard question to answer and there are so many ways of answering it. I also started creating a set of heuristics that not only outline the ‘frame for good’ but also clearly emphasize the benefits and limitations of using a CUI in different circumstances.

Week 13

We had been preparing for this round of workshops for a long while now. I’m glad that aspects from our very first workshop like the journey map and personality of a team mate are reflected in what we’ve prepared.

Based on the feedback from our scenarios, we came up with a list of features and mapped it onto the user journey to see where it would fit best for a concept video and for the validation workshop. So we had to achieve a balance between complexity and communicating the concept.

This one was the most structured workshop we’ve ever conducted and had some great positive and negative responses. For me, writing up the script and mapping out the flow of interactions has been a great learning experience. As a team though, we could have stayed more agile and iteratively prototyped the physical interactions and multiple levels of feedback.

We did all take a ride in Kate’s car and play out the scenario to iron out the kinks and learnt a lot from that session too — that you don’t always need experts to test concepts.

Week 14

We conducted the workshop on Friday and had only a day to pull together a draft of the presentation on Monday before refining it over the next night for Wednesday. Initially we bit off more than we could chew and were trying to present all the details that we had thought of. So we pulled back and perhaps in our eagerness to present within the allotted 10 minutes, we didn’t present enough context and detail. Kevin had the interesting feedback that the concepts weren’t specific enough to EMTs. Though we did get very EMT-specific feedback from our workshops, we hadn’t integrated those feature details into the video.
We have quite a lot to do in the next couple of weeks and hopefully we’ll be able to get there as a team.

Week 15

This past week, we’ve spent quite a lot of time thinking about how to strategically present our ideas. There’s two questions there — about providing context and framing the concepts; and the concepts themselves. We realized that the concepts we had narrowed down to after all our research didn’t look EMT-specific enough. I suppose that’s also the challenge with user-centered design. We had all the same assumptions before heading out for research — that they want to stay in constant contact with the hospital, that they need assistance with administering drugs, that they need explicit stress relief. But none of these proved true. It would have made for exciting design solutions that would have been very EMT specific but they were clearly not needs.

We’ve now decided to focus on other features that we’ve not tested as rigorously with participants but that show the value proposition of the system and fit in within a framework.

Week 16

We’ve had so many revisions and frameworks that we tried using to present our solution it’s hard to keep track at this point. After spending hours discussing the minute details of the interactions, the kind of artificial intelligence and the personality of Kara, it comes down to how well we can build a narrative not just around the work we did but also the value of it. Thinking from the point of view of the Microsoft liasons has been a great learning experience. After all, presentations have to be catered to the audience.

We had the video sketch blocked out last week but once the storyboards we stitched together with the audio, the feedback we got was that it was confusing. We were trying to do too much. Plus, given that our ‘deep dives’ of the interactions were also videos, it seemed to be a lot of video to sit through in a 10 minute presentation. We decided to prioritize the deep dives and later think about how and if we wanted to integrate the concept video into the final presentation.

Week 17

It’s finally done! The feedback we got during the final presentations was great. After Peter and Bruce gave us their feedback, we revised the slides multiple times and tested it out with people who didn’t know that much about our particular project. Because we didn’t have an emotional concept video, we needed to make sure that the initial slides of the presentation set the right context dramatically enough. Simon’s feedback that we could have spent a little more time discussing the problem and showing how messy their current infrastructure is was really helpful. Also, Irina’s comment about the usage of data as a business proposition was great. Kevin’s feedback about talking a bit about the actual implication of the system — if it could be retrofitted onto existing systems or if new systems would need to be created from scratch. Peter’s comment about disagreement between human and computer was something we had resolved in our earlier scales of human and computer autonomy and also resolved in the fact that the ambulance is just smart and not autonomous. We presented our concepts to the EMS personnel who had been our participants earlier and surprisingly, our conversation with them led to answers for a lot of the questions raised the previous day.

The project was very exciting, and I think it gave us the opportunity to explore conversational user interfaces, multimodal interactions and research methods in order to design for a highly specialized user group. My thoughts on user centered design have become more nuanced and I’d like to think I’ve learnt to craft better presentations. Personally, I’ve learnt so much from my teammates and my design process (not ‘the' design process since there are so many) has evolved tremendously. The project was challenging, working with the same team for an entire semester was tough at times but I think we’ve come out of it with a lot many avenues to explore and ideas to pursue in future work.

Amazon might be considered as a digital only service but the fact that you order, anticipate and receive physical products makes it a digital-physical hybrid. Users approach it with a different need or intention than pinterest but expect the same level of stimulation. Since the interaction involves a monetary transaction, I think the ‘Interact’ section is a lot more detailed and explicit.

Shopping at a grocery store is as physical and mundane as it gets for me. Like with Amazon, I’d only go there when I need something, and not expect to be ‘delighted’. Maybe that’s a good opportunity for in-store service experiences. I was surprised mapping this out because I assumed that the Amazon experience would simulate the the in-store experience but I guess they leveraged technology to create a more satisfying and personal experience. Being able to navigate time and distance to build in a more social experience is also a positive, as compared to a physical store

Documenting service experiences was originally published in Design for Service on Medium, where people are continuing the conversation by highlighting and responding to this story.

As soon as you enter Dobra Tea, you encounter the reception that also serves as the tea making station and product display. It smells wonderful, there’s yellow lights on wood, exotic carpets and ’s a very visceral experience from the moment you put one foot in the café.

The host/waiter/chef hands you a menu (it’s a binder full of teas! Definitely implies that these guys know their teas) and a tiny brass bell that you ring when you’re ready to order. For me this was an important artefact and my perception of what it implied changed as I experienced more of the service.

Initially, I thought it was cute and quaint, a nonintrusive and non-hurried way of ordering. The sentiment seemed in sync with the pace and mood set up by the décor. But when I had to ring it, it felt embarrassing to expect service at the ring of a bell so to speak. The less formal way of making eye contact and smiling to ask for help rather than ‘order’ seemed more natural than the use of props that solidified power dynamics.

The other interesting thing about Dobra was the use of spaces to create nooks and niches that allowed for different kinds of behaviour and interactions. Most of the people who were working on laptops or studying were sitting at the tables and chairs and those who were engaged in conversations with other people or reading leisurely were sprawled on the diwans and couches. Also, people who sat at the back seemed to stay longer (maybe they wanted to, that’s why they sat at the back) than those who sat closer to the entrance.

The interaction from a customer’s point of view was pretty straightforward. I’ve mapped it as a task flow

Walk in — Find a spot (an excuse to explore the space, figure out your own intentions and how you want to spend your time) — Unpack — Decide what to drink+eat (‘read’ the menu (artefact), validate authenticity+evaluate if expectations match the content) — Order (ring a bell (artefact)) — Wait — Drink+eat+work/talk/daydream — Pack up — Go to the payment counter — Pay (with a card, you interact with a person, then tablet (artefact), then person — leave

From an employee’s point of view –

Greet customer — Hand them the menu — Hear for the bell — Get the order — Make the tea/food — Serve the tea/food — Discretely check up on the customer — Manage payment — Clear table — Do dishes — Check inventory — Order inventory

Infrastructure involved –

Making tea, cleaning etc doesn’t happen at the back. Infrastructure for power, plumbing, maintaining safety and hygiene regulations have to be considered.

The decision to conceal (heating) or reveal infrastructure and processes through visual or sound or both (washing up) is a very deliberate one aimed at creating and selling an experience by appealing to our rational and irrational mind.

Documenting a service was originally published in Design for Service on Medium, where people are continuing the conversation by highlighting and responding to this story.

What is Interaction Design?

Interaction design is the orchestration of environments, technologies, services to make it (more) conducive for people to communicate with each other and have better experiences.

My definition of interaction design hasn’t changed too much. In fact, in reading it again now, I can see parallels between it and my final paper for this class.

I have a slightly different take on the popular perception that design is about problem solving. It was evident from our Interaction design studio projects that design is not so much about problem solving as it is about problem framing. Solutions and fixes feel more immediate and concrete but finding the right problem and asking the right questions are more critical.

Interaction design is also so tightly integrated with communication design –not just in the graphic design sense. Good and honest communication (vague, I know, but it doesn’t have to be efficient or clear), is so important when asking questions, teasing apart the problem space, facilitating conversations and presenting solutions and receiving feedback. Perhaps that is the simplest and broadest way to describe interaction design — as a conversation. Good interaction design is a good conversation, it doesn’t have to be a resolution.

In my previous definition I skirted the issues of technology and ethics. Both have their own implicit political baggage. I’ve learnt through the course of the readings that technologies have flavour and affordances. It’s important to understand what it is and what it’s trying to do, what it can do and what kind of infrastructure systems it gives rise to or that support it. Initially I was adamantly against the techno-centric view of design and I still am not in support of a techno-deterministic society but I’ve come to realize that it’s a tool like any other and as designers, it’s imperative to be in the know. That way, you always have the option to subvert it. But if it’s not part of your personal design ideology at all, that can be problematic.

Anthropologist for Intel describes Big Data as a person - InformationWeek

My previous definition restricted interaction design within a system but I now think that design can operate at two levels — where it works within the system but affects it and the other is at the system level itself. I’m interested in seeing how to work with it at this level. How do interaction designers become part of this larger process? How do we prototype and iterate? How do we scale up and scale down, maintain the 30,000 feet (5.68 miles doesn’t have quite the same impact) perspective and the 1 foot perspective? What does it mean to craft systems for interactions and not just the interactions?

What was Interaction Design Interaction design is about managing the behaviour of entities (human… was originally published in Interaction & Service Design Concepts: Principles, Perspectives & Practices on Medium, where people are continuing the conversation by highlighting and responding to this story.

I was interested in trying to figure out my changing diet. Whenever I move to a new place, my diet changes drastically. I love good food but it’s not always top priority. I think I ususally eat healthy but I can be a glutton. The relationship between my work hours and the amount (and type) of food I eat is strange and I haven’t really been able to pinpoint a pattern.

I was also interested in the idea of my eating habits changing if I knew I was capturing as data — like people being more active when they wear activity tracking devices.

First I started taking photos of my meals. I had the time and locations.

I then started weighing them

It was impossible to do this on a regular basis so I just maintained an excel sheet. Plus I eat a lot more than I plan to in the morning.

So I started an excel spreadsheet.

The excel spreadsheet had to be reorganized

This were the initial sketches exploring form using visual variables.

I saw this at Phipps and realized that on it’s own, the data doesn’t say much. The composition of meals and the time are pretty important for a healthy diet. But what is the benchmark? What is a healthy diet? What kinds of ‘diets’ do people follow? Contrasting these to my actual meals would be interesting and would accentuate the problems with my diet.

Exploring forms using visual variables. I looked at forms other than pie charts — and ways of representing time. I also considered interactions that viewers could have with the data and the layers of information that they could possibly see.

To explore the form in digitally, I started defining the visual variables as graphics.

I looked at how shadows could possibly be used to show that the day was progressing and that meals were transitioning from breakfast to lunch to dinner.

At this point, though the visual components could represent the data, I didn’t think it looked enough like food.

I was pretty sure I wanted to show the temporal aspect and the ‘missed’ meal times using actual time so I constructed an elaborate time conversion system that would scale every 2 hours to 4 seconds.

The first row represents the ideal meals and the second row, the actual for each day (every column is a day). Though this does provide a contrast between the ideal and the actual, it was very confusing. Also, the video format allows for less interactivity.

I was trying really hard to move away from the literal representation of food (like color and photograph of the actual meal) to a more abstract representation. But the flat, clinical pie charts felt too far along the spectrum towards abstraction.

The intention of using the photographs was to create a texture and visual interest in the forms of the icons. In doing so, I still had to co-relate it to the colors of the ‘Ideal’ meal. I thought the outline would help but it wasn’t visible enough. Adding an overlay of color just muddled the message and the color.

Changing the images to make them more like textures did help. Using grayscale images also did reduce the visual strain but it didn’t look exciting enough.

The added layer of information in the number of rings is the number of people I ate the meal with.

I cropped several different images to see which would work best as textures and colors without additional graphic elements to communicate food groups.

A description of my project — the process and structure

An interactive Axure prototype of the visualization. You can view it on your browser using this link.
http://auzdg2.axshare.com/#p=home

Here is a walkthrough of the interactions

Food diary is an interactive log of everything I ate from Oct 19 to Nov 8, 2015. It’s a visualization of my eating habits with a focus on time and proportion of food groups consumed at each instance rather than amount. More information is revealed at each interaction that adds and changes meaning. Grain and vegetable is healthy right? Nope, it’s pizza. I wanted to see if and how recording and visualizing data about myself would force me to consciously make healthier decisions.

Visualizing Patterns was originally published in CD Studio Documentation on Medium, where people are continuing the conversation by highlighting and responding to this story.