ABSTRACT

Wearable devices
have been a pivotal interest in the field of HCI and become a trend for individuals these days. While the individuals are using this technology heavily on observing and detecting their body after exercising; nevertheless, people have aroused their doubt on the presenting data that shows in the wearable devices due to the unclear methodology. Furthermore, the concern has also arisen how the computing system calculates the information on fitness application, such as the counted footstep or the burnt calories, matching to their ongoing activities during the individual exercises. To discover and address the challenge and apprehension that people have, this paper presents an ethnographic study by interviewing seven students from The University of Queensland who have experienced using the exercise tracker. This unpacks how the individuals measure and perceive the showing data on the activity tracker. Additionally, we illustrate a set of insightful aspects on the reasons why the participants want to do the exercise, discussing the health, attractive, and delightful reasons and experiences human beings have while they are doing exercise. Through the finding and the discussion, the implication of the study suggests refinement approaches to the current issues that the tracker has, and what the individuals perturb. These suggestions will allow the developers and designers who develop the wearable trackers to leverage the findings they overlooked.

Keywords: Activity tracker; Ethnographic research; HCI; Social engagement

Duration
4 months // 2020
Role
Team Leader and Researcher
Collaborators
Yawen Deng
Peiwen Li

INTRODUCTION

In light of the recent events that Apple has released its six-generation of Apple Watch, activity trackers nowadays are used for many purposes throughout the world. More than twenty percent of the people of the United States today have worn the activity trackers or the wearable fitness device regularly [6]. Some people uses the tracker to record their daily step count while the others might be using the tracker to log their beats per minute (BPM) of the heart rate or the burnt calories after doing the exercise. Activity tracker has become an essential that interweaves our daily life [1, 5, 7]. In the field of Human-Computer Interaction (HCI), researchers have shown an increased interest in self-tracking and pervasive technology. For example, it is argued that social media platforms, such as Instagram and Facebook, should make a close connection between the individuals and the personal informatics [1]. Also, extensive researches have shown the need of the integration of the data on multiple tracking devices [2, 4]. The literature has diagnosed the hurdles while the individuals are using the activity trackers, mentioning the data interpretation and the data authorisation may be two decisive factors why the individuals do not want to use the wearable device [4]. While there are many benefits of using the activity tracker, one of the major issues in many research concerned that the individuals do not trusted the presenting data on the activity tracker due to the inaccurate information and unclear methodology [2, 3, 4]. However, previous HCI research have not dealt with the unreliable data on the activity tracker and proposed the issue on how to offer the users a sense of security while presenting the information that correlates to the users’ physical activity.

To address the identified issue, this study provides a set of opportunities to discover the incentive of why the individuals want to fitness. We found that the individuals do the exercise with the activity tracker comprising a set of co-existing goals - improving the health and shaping their body, sexual attraction, and the delightful and emotional experience during the time they are doing the exercise. More importantly, we discovered that the individuals are interested in receiving the reward sent from either the activity tracking application or social media platform [1, 4 ,5]. People who have the habit of posting their food on social media are likely to receive and send the social support in the hope of getting positive feedback socially [1]. This triggers individuals’ interests to use the activity tracker. That being said, we also found that some individuals, especially for people who do not do the exercise regularly, only use their own mobile-based application to log the data instead of using the wearable devices. As the goal that the participants set are completed, we discovered that the continuity of doing the exercise becomes a habit to the participants, and they are likely to continue using the activity trackers. This strengthens and supports the argument presented by the other distinguished authors [7]. Creating upon prior works on the implication of the research and study from the findings and discussion section, we have designed a mix of design approaches, allowing the developers and the designers who work in the wearable device industry to explore the possible solution in the hope of addressing the discovered issues.

LITERATURE
REVIEW

Our study of how people perceive the wearable device and activity trackers regarding their strength and the limitation was informed by prior research on the barriers that people may experience during the time they are using the trackers, including the fitness tracker, and the sleeping tracking technology. More recent existing-literature has focused on the delivery of the integration either from the multiple tracking devices or the social media platform. These provide the insights of how people enmesh and interweave the tracking technology to their daily life. And they are the following:

The Inaccurate Information

The works of literature on activity tracker have highlighted the importance of incorrect data. A review of 24 participants discusses the barriers of the activity tracker has shown that the trackers cannot record accurately, especially for measuring the non-step activities [2]. The study also shows that people would likely to have a relatively holistic activity tracker, compared to most activity trackers these days can only calculate the specific statistics. With regards to the study discussing the accuracy on the sleeping tracker, Zilu and Bernd [3] explored that as the sleeping tracker relies on the accelerometer data to measure whether the participants have the sleeping issue, the way that the participants make the premeditated or unintentional movement during the time they are sleeping may lead the inaccuracy of data aroused. More importantly, one research has argued that the participants are highly concerned the accuracy, even though most participants considered that it is acceptable for not showing the guaranteed statistics precisely due to the calculation method [5]. Other research has conducted a qualitative study and specifically identified the current challenges that sleeping tracking technologies have [4]. They found the data inconsistency when twelve users utilise more than one sleeping tracking technologies simultaneously when they are sleeping. As the users compared the result congruency from multiple sources on their trackers, the authors found that the participants wondered what the methodology that the wearable devices set to measure their sleeping quality. What even worse is that some trackers are not able to permit the hope of adjusting the incorrect data. This unclear design has created a sense of insecurity and doubts to the users. One study has shown the expression from the existing literature that the participant was surprised by how easy the step count can climb up and be counted [7]. This indicates the limitation of the current issues that the activity trackers have.

The Integration of the Personal Informatics

Much of the tracking technology research has focused on investigating and evaluating the integration of personal data. One paper has suggested the social comparison to reduce the barriers to tracking technology [2]. Based on the research, they argued that the participants tend to swap using the same tracker since they want to share their informatics of the physical activity to their friends and peers. This shows the frustration that the individuals are expecting a cross-platform tracking technology that will give them a social support network online. Additionally, Wanyu et al. [4] have illustrated the hurdles with the sleeping tracking technologies regarding data manipulation. They found that the data export and the integration tool are two pivotal challenges that the sleeping tracker user encounters. This also shows that the individuals are vigilantly looking for a proper application or tool to transmit the data on their devices. Besides, a previous study has explored the relationship between the personal informatics and the social media platform [1], arguing how people who take health food journey or diary on Instagram, a photo-based application can be used across smartphone, tablets, and personal computer. This provides convenience for people to log their eating habit, and they can always review their visual-appealing notes on multiple devices.

Social Support

Social support has stood a pivotal role in the field of tracking technologies. Numerous studies have attempted to explain that people are eager for having social engagement through the tracking device. For example, previous research has established that the reason why the participants are fascinated using Instagram to record their eating habits, allowing them to have a social interaction through leaving comments or getting feedback from their friends or followers [1]. This provides social and emotional support for participants to use social media actively. Also, the study has shown the findings that social engagement can support people using self-tracking technology; this can encourage users to have a behaviour change. Another study has discovered that the activity tracking in an employer-sponsored health program offers the social influence and interaction between the users [7]. With regards to the support feature for the participants, some literature has shown the influence of the reward system either from the tracking application or the social support from the peers. One study shows that collecting rewards has triggered the participants’ interest in doing physical activity [5]. This not only gives them physical benefits such as the gift voucher but also receive mental advantages such as the sense of achievement and accomplishment. Another study has illustrated that the scores on the activity tracker can help people to have a positive competition with their peers socially [2]. With the score and reward collection display socially in the user’s network, this indubitably encourages the users to do the physical activity regularly in the hope of getting a good result to show off to their peers.

METHOD

We used contextual inquiry [8] under ethnography methods to explore people’s opinions about our research question: How to make the users believe that the displayed data on the wearable device are correlated and matched to their activities? For investigating this research question, three different methods were used to collect both quantitative and qualitative data from participants. The contextual inquiry methods of interviews [9], cultural probes [10] and observations [11], fulfill the say-make-do model [12]. Data were collected from what the users were saying, doing and making, providing a comprehensive understanding of the user's behaviour and attitude about accuracy of exercise data on wearables.

In the first phase, participants were given three wearable devices to finish an outdoor walking task (Make). This activity aimed at providing participants a better engagement to understand the wearable devices tracking exercise data. In order to minimize the performance differences among trackers, these three devices were chosen from the same company, Huawei, with different models, as shown in Figure 1. Besides, in order to minimize the bias from different parts of body movement, participants were asked to wear them on the same arm, as shown in Figure 2. In addition, three devices were used in order to set one as a reference group and compare with the other two devices. An observation of participants’ behaviors and interaction with the wearable devices was conducted during the walking task (Do). This observation took place at the same time when participants performed the walking activity, with two researchers observing.

Figure 1: Wearable device models

Figure 2: Wearable devices on the same arm

In the second phase, interview questions were asked to understand participants’ experience of using wearable devices and exercise habits (Say). This approach was chosen because participants could articulate their feedback on the understanding and using trackers, and it provided insights on user behavior of recording data when doing exercises.

Participants

Seven participants (four female and three male) in Brisbane took part in our cultural probe activities and interviews. The age range of these participants were from 20 to 30 years old. In order to get useful feedback about accuracy of exercise data, we only recruited participants who have a regular habit of doing exercises, as shown in Table 1.

Table 1: Participants demographics

Data Collection

Quantitative exercise data recorded by three wearable devices from walking activity were synchronized through Bluetooth to the mobile phone application of Huawei Health. Qualitative data of observations and interviews were audio recorded and taken on the notes.

Data Analysis

The quantitative data from cultural probe activity were compared to find out the biased data deviated from the other two trackers. Only common fitness data on all three trackers were compared. Number of times of the deviation on each device under different fitness items is shown in Table 2. The qualitative data of audio recordings from interviews were transcribed and were analyzed through a thematic method using affinity diagram (Figure 3).

Figure 4: Affinity Diagram

RESULTS

The findings gathered from data analysis can be categorised into four aspects: exercise motivations, recording habits, trust of accuracy and perception about fitness data.

Motivations of continuously doing exercises

To study the exercise habits of people who exercise regularly, we investigated the frequency and the types of exercises participants do. All the participants would do exercise at least once a week. The types of exercises are various based on personal preferences, which can be categorized in either aerobic or anaerobic exercises. Apart from one participant who was enthusiastic about outdoor exercises, all the other participants (85%) prefer indoor exercises. The initial triggers to do exercises are to keep either physically healthy or mentally healthy. After a long period of persistence on doing exercises, these groups of people have explored their own pleasure and enjoy the process. The motivations of keep doing exercises are strongly personally related. Their exercise routines still exist without a fitness tracker.

The wearable device is not the main factor that affects my exercise, it is just a system that assists my exercise. If I do not have a wearable device, I will still do sports. (P1)

The persistency of doing exercises can have a positive effect on users.

He [Men] doesn't do it for attracting females, he does it for himself. And sports can't do without you over the time... because it only shows effects when you keep doing it. It won’t work if you give up halfway. (P5)

One of the factors mentioned by participants is the emotional delightfulness brought by fitness.

... If I feel good after running 2 kilometers, I want to keep going. I used to have an experience about the records that I challenged myself yesterday. I set the same goal for 2 kilometers and I feel much better after a long time running. I received incentives for doing it consistently. (P2)

Because when you stop, you become fat again... to make me more confident, much happier. When you have better body shape, you will become happier. And doing sports will secrete dopamine, which will also make you happy. When you study all day, you can release pressure and relax from exercise. (P4)

Another factor is that the exercise habits have been integrated as a part of their daily life. Positive feedback on their bodies makes them continue doing exercises.

I have heard a saying that there is a DNA in people’s bodies that motivates you to do exercise. If you don’t do exercise, your body function will degrade. It won’t let you degrade because you need to use your body to hunt. (P6)

Because you can see your own effect. It is a long-term thing. If there is no feedback during this period, you will feel very disappointed and give it up. As it will give you a feedback of how much you did today for a group of data, so if I do not achieve it tomorrow, it doesn’t work. I want to do better. (P5)

Fitness Data Recording Habits

From the interviews, we find only half of fitness participants have an experience in using wearable devices. These participants used to check fitness data in a high frequency, either immediately after exercising, or do it whenever they have time. While they do not currently use wearable devices, it is because after users understand how their body and exercise data goes, only important fitness data needs to be recorded.

As time goes, the base changes little, and the weight does not change much, so the diet data becomes no longer important to be recorded, and the diet data no longer affects the weight data. (P3)

I can check it whenever on my phone, but I haven’t used these old ones recently. You can see this Huawei Health app on my phone, it records not only the exercises data, but also weight, heart rate and sleep information. (P7)

The other half participants have never used such wearable devices we provided in the cultural probe activity. These participants use either built-in functions or applications in mobile phones to record fitness data.

I use the built-in health app on my phone… it will record automatically in the normal time. It only records the simple data like step and distance. It doesn’t have calories and heart rate, not so much information. (P4)

I used to download an app which is named Keep, I used Keep to record during that month and went to the gym. Keep would record my weight every time I went to the gym… I saw the data from Keep after doing every exercise, you put in the time and types of exercise, it will calculate the calories you burnt. You fill in your own exercise style in this app, it can calculate how many calories you burnt. (P2)

I use my mobile phone to record data. I measure the daily weight data and record it in the notepad of the phone, and other data will not be recorded. Sometimes I also record my waist measurement data. (P3)

Some of them do not have the habits of recording it.

I don’t like to use devices to record data. When I feel tired, I will stop running. (P6)

Accuracy of Fitness Data

According to the quantitative data recorded by wearable devices from seven participants, there was not exactly the same data for all three devices under each fitness item. From the user's perspective, most participants believe the fitness data is accurate in a certain range.

I have participated in a 21km marathon before. I have worn it and saw its pace information. I think its data is quite accurate. (P1)

I think the data is accurate in general but may not be accurate to one step. (P5)

After the fitness data from the wearable devices were present to participants, the biased data were selected as they thought it was deviated from what they actually achieved. In terms of the tolerance range of each fitness item value, Table 2 has summarized the number of times fitness data deviated from the actual activities they conducted. Apart from the duration item, which was controlled manually by users, the other fitness items have deviated at least once. Comparing among three wearable devices, all the three trackers have deviations in different fitness items.

Table 2: Number of times fitness data deviated

For people who believe the data, they understood that as a device, it sometimes makes mistakes.

It is based on your exercise, even if you are not walking, it will be counted when you hold it and shake it. Sometimes it is more exaggerated. (P5)

One participant doesn’t believe the accuracy of the fitness data on wearable devices at all and chose to record data manually.

I think the calculated data is not accurate, so I do not use software to record the data. I think the app that records exercise data is not very accurate in calculating calories, and it is not proportional to my exercise volume. Maybe it is because I only do yoga, the consumption of the calories is inside the body, it does not affect the volume of exercise. It’s not as obvious as other anaerobic sports in the gym. (P3)

The reasons behind participants’ trust about fitness data varied. For those who trust the data, even if the data is not accurate to a digit, it is still worthy of reference

I think most people believe it. If they don’t believe it, they won’t check it or use it. (P5)

I do not very care about the data from the wearable device, the data is just a reference value. Unless I am a professional athlete, I will care about this data, I will consider the speed. (P1)

The price among different devices might also be a factor of the accuracy.

I find different devices various, but I think the expensive ones might be more accurate. (P7)

Perception about Trackers

There are three main aspects that affect users’ understanding of the accuracy of fitness data on trackers. Firstly, users want to know the basic principle of the calculation methods for each fitness item. It is not the algorithm behind it, but the simple reasonings of how it comes from body movements.

I want to see how it is calculated, the reasoning or basic principle behind the number. For example, it will present the calories burnt, I want it show how it is calculated below in a simple way, such as time multiplied by what...I don’t want to wear it. It makes me very uncomfortable. (P3)

This data is based on your weight, height and body fat rate to calculate how many calories you will burn after doing this exercise. The methodology should be clear. (P1)

I’m not sure if it includes the data when I take the bus and it moves slowly. I don’t know the mechanism of how it calculates the distance and step... the information present should let me understand how it works. It can express when they advertise it. (P4)

Secondly, information provided should be more realistic and easier to understand, especially for users who don’t have ideas about the fitness data.

I have no idea about the distance before. Now I know that one circle of the Great Court is very short. I used to think it should have 1 or 2 kilometers. I feel more real now. (P2)

I think it doesn't need to quantify the accurate data but can set some levels to distinguish them in a certain range, such as level 3 or level 5. (P6)

Thirdly, customization on the fitness data should be available so that users can focus on their own needs in exercises. One participant mentioned the rewards information, such as encouraging statements, were more important to convince her.

I believe the data analyzed by these apps is a kind of comfort. When I see these data, I can feel that I have really exercised. When I felt very tired after exercises, I chose to believe the number given by the devices…I want to see the random encouragement, even if it sounds very silly and lovely, such as “you’re great!” or some witty words. I won’t have the desire to share my achievement on social media with others every time, but I still need to get approval from others that I did this. (P2)

DISCUSSION

All participants think that the methodology of the data calculating affects users to believe the accuracy of data analysis. Some interviewees do not trust the data analysis derived from the methodology of the data calculating, they consider that the algorithm of the data is not completely displayed on the application, and the users cannot understand how the data is calculated to result in users do not believe in the data and doubt the accuracy of the data. For example, a tester clarified that she used an app to record data (the number of steps and distances), the data was reviewed after the participant doing exercise. The participant doubted how the app recorded the data and was confused about the accuracy of the data. Data analysis should be combined to provide users with more accurate data based on exercise methods and exercise factors. Data recording can be combined with customization so that the system can display more professional data analysis [3]. The activity trackers always involve people’s daily lives and people’s needs [5]. Data recording can be selectively recorded according to the sports-related elements and the data values that the user is paying attention to at the time. In this way, users can see their own personal data analysis. In addition, the process of data analysis should be displayed every step in the activity trackers, which allows users to view the analysis of the data at any time and can enhance the user's trust in the analysis of the data.

On the other hand, the researchers believe that the data displayed and compared on different wearable devices can more accurately observe the diversity and accuracy of data analysis in this study. After the participants walked 1 kilometre, the data will be displayed on three different wearable devices. However, the data analysis values in these three wearable devices are different, and this difference is subtle in most data analysis. Some testers pointed out that if the difference in test data analysis is not particularly large, the tester can still accept to some extent of the data difference. In addition, the maximum heart rate data has the biggest difference in data analysis among three different wearable devices. The data in this Huawei honour watch is different from the other two. These three wearable devices are all from the same company and their functions are the same. Based on these factors, this difference in data may be due to internal hardware.

Initial Design Concept

In the previous parts, we have discussed the reasons why people do not trust data on wearable devices. The next part will show how to make users trust the wearable device data analysis and how to make users more willing to use the wearable devices in the design.

Design for Data Accuracy Analysis

During the research process, many participants expressed that they did not believe in the research data of wearable devices because they could not see the transformation process. In addition, research illustrated that the inaccuracy of data is related to people's habits, and wearable devices cannot detect people's activity habits, which leads to data deviation [4]. In addition, we discussed that the small data gap may be due to hardware issues in wearable devices. According to the previous section, we discussed that some participants accept slight data deviations. These testers believe that the deviation of wearable devices is a normal phenomenon and the data will not be completely accurate.

To reduce errors when moving with the wearable tracker, setting approximate values is a solution, such as setting ABCD levels. Each degree of motion represents a value so that the error value can be reduced. In addition, data customization is very important. This will analyze each user's data more systematically. Customization means that users have different motivations for doing exercise so that the data they want to see is different. In our interview, we found that each user pays different attention to different aspects of the data displayed by wearable trackers. As a result, users can choose the aspects of the system they care about to understand how the data changes without having to see all the data presented. A tester expressed that the participant would like to see results on the doing exercise. If the wearable device tracker does not provide the participant results continually, the participant will be disappointed and refuse to do exercise. Based on this finding, wearable device trackers should be designed to allow users to see the change process of the data at any time, and the user can also view the data analysis report generated in the past. This design can facilitate the user to compare the data and see the change of the data. The user can see the change process of the data at any time, and the user can also view the data analysis report generated in the past. This design can facilitate the user to compare the data and see the change of the data. The user sees the data algorithm clearly, which can make the user more convinced in the accuracy of the data.

Design for Data Integration

According to a result of the study illustrates that people are willing to share their information in social platforms and they are also like to check their information in different devices [1]. Moreover, the exporting data and data integration are challenges for the wearable trackers [4]. These issues try to be solved by design. The Internet of Things connects everyone and everything. People can connect information through the network so that information can be shared on the network. pervasive computing (ubiquitous computing) connects information and different devices. These two technologies inspired me to design a system, which exports data in different devices. People can use the data of doing exercises continually in other trackers, which makes it easy for people to move around in different environments and the recording data can be updated all the time. Furthermore, people can share their data to their friends on various social platforms.

Design for Community

From some papers, the social function is one reason for users using wearables devices. With the development of electronic devices, users' cognition of them is no longer just a machine to record data. People like to use electronic devices to record data and share their data with others through social platforms [1]. Therefore, creating a community is an essential function to assist people sharing their information to others by wearable devices.

Two feedback from the interviews indicate that the data improves people to do more exercise. A participant thought that the recording data of the wearable devices would allow the tester to see the daily limitation of doing exercise. Another Participant indicates that that data improves the participant to break the previous sports records. Furthermore, according to a study, users tend to challenge targeted activities, and people will complete the exercise in the process of unconsciously using the tracker. Users in the same challenge can also make friends in sports [4]. From these findings, the community function could set the same exercise goal for users, so that users with the same needs can do exercise together. Users can use wearable devices to record exercise data and test whether the common goal is achieved. Users can also make friends in the community function.

During the research, two participants expressed that the rewards or aesthetic interface setting will attract them to do some exercises. The encouraging words can enhance participants to motivate people doing physical activity. If the smartphones can display some encouraging words during their doing exercises, they feel encouraged and are more willing to do exercise. Smartphones should add some lovely encouraging words to promote users doing exercises. A paper records that the aesthetic appearance and rewards impact the increasing usage of the trackers [2]. The find is also mentioned in another article, which demonstrates that people like to receive rewards while using an activity tracker [5]. These data illustrate that the aesthetic interface setting and rewards occupied an important status in improving people using activity trackers.

Limitation & Future Work

Although we have analyzed the reasons why users do not believe the data analysis results of the wearable trackers and the reasons for data deviation, small deviation of data is inevitable. We try to change the level of data classification to reduce the deviation value of data. This approach only makes the data hierarchical, so that people can see the level of the reaching amount of physical exercises after doing exercise, rather than the exact data value. As a result, the data can be slightly skewed. For the future work plan, we will study how to make the wearable device tracker can better sense the movement frequency of users, so that it can analyze the data more accurately.

CONCLUSION

This article focuses on researching how to convince users of wearable device data analysis. We have prepared three different wearable devices for the data analysis of the user's doing exercise. 7 participants were invited to wear these three wearable devices to walk the same distance, and we analyzed all the data of these participants’ walking. In addition, we used interviews to understand the testers’ perceptions of the wearable devices and whether they believed in the data analysis of wearable devices. We found that people do not trust that the data analysis of the wearable devices stems from the user's invisible complete data algorithm process. However, participants believed that it was inevitable that there would be a small error in the wearable device, and they could accept the difference of data analysis in a small range.

Based on these studies and analyses, we have made several design approaches to address the finding issue. First and foremost, the main purpose of the design is to enable users to have more trust in the data analysis of wearable devices; the major reference is to create an overview of the dashboard that the user has exercised, presenting hierarchically and conducting data analysis according to users' habits. Last but not least, we have collected users’ other needs for wearable devices, which is also considered in the design aspects. These will enable the designers or the developers of the activity tracker to create a better device in the future.

REFERENCES

1. Bill Gaver, Tony Dunne, and Elena Pacenti. 1999. Design: Cultural probes.
Interactions 6, 1
(Jan./Feb. 1999), 21–29.

2. Chia-Fang Chung, Elena Agapie, Jessica Schroeder, Sonali Mishra, James Fogarty, and Sean A. Munson. 2017. When Personal Tracking Becomes Social.
Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems.

3. Daniel Harrison, Paul Marshall, Nadia Bianchi-Berthouze, and Jon Bird. 2015. Activity tracking: barriers, workarounds and customisation.
ACM Digital Library,
617–621.

4. Dhaval Vyas, Zachary Fitz-walter, Erica Mealy, Alessandro Soro, Jinglan Zhang, and Margot Brereton. 2015. Exploring Physical Activities in an Employer-Sponsored Health Program.
Proceedings of the 33rd Annual ACM Conference Extended Abstracts on Human Factors in Computing Systems - CHI EA ’15.

5. E. B.-N. Sanders. 2000. Generative Tools for Co-designing.
Collaborative Design
(2000), 3–12.

6. Emily A Vogels. 2020. About one-in-five Americans use a smart watch or fitness tracker.
Pew Research Center.

Retrieved from https://www.pewresearch.org/fact-tank/2020/01/09/about-one-in-five-americans-use-a-smart-watch-or-fitness-tracker/

7. Hanna Kallio, Anna-Maija Pietilä, Martin Johnson, and Mari Kangasniemi. 2016. Systematic methodological review: developing a framework for a qualitative semi-structured interview guide.
Journal of Advanced Nursing
72, 12 (June 2016), 2954–2965.

8. John Rooksby, Mattias Rost, Alistair Morrison, and Matthew Chalmers Chalmers. 2014. Personal tracking as lived informatics.
Proceedings of the 32nd annual ACM conference on Human factors in computing systems - CHI ’14.

9. Mary Elizabeth Raven and Alicia Flanders. 1996. Using contextual inquiry to learn about your audiences.
ACM SIGDOC Asterisk Journal of Computer Documentation
20, 1 (February 1996), 1–13.

10. Vesa Jääskö and Tuuli Mattelmäki. 2003. Observing and probing.
Proceedings of the 2003 international conference on Designing pleasurable products and interfaces - DPPI ’03

11. Wanyu Liu, Bernd Ploderer, and Thuong Hoang. 2015. In Bed with Technology.
Proceedings of the Annual Meeting of the Australian Special Interest Group for Computer Human Interaction on - OzCHI ’15.

12. Zilu Liang and Bernd Ploderer. 2016. Sleep tracking in the real world.
Proceedings of the 28th Australian Conference on Computer-Human Interaction - OzCHI ’16.