Mission Possible: Using visual feedback to improve physical activity in children


This paper describes the deployment of a novel ubiquitous behaviour change system for social interaction and reflection amongst school children. For four weeks, a class of schoolchildren (Year 5) was monitored with Fitbit activity monitors and their daily physical activity was visualised on a custom ambient display. In addition, video segments describing mission-based activities were shown on tablet devices to the children at the start of each week. The ambient display would indicate if they performed better than the previous day. We describe how the system was designed and developed, present findings from the in-the-wild study, and provide design guidelines for future studies.


It is well established that peadiatric obesity is associated with numerous health implications in later life (Freedman 2007). Despite evidence to suggest that the prevalence of obesity has plateaued in recent years within the UK (Boddy 2010) and internationally (Rokholm 2010), there is no evidence of a decline, and a high proportion of children remain at risk of morbidity. Physical activity and sedentary behaviour are key variables implicated in childhood obesity due to their influence on energy balance (Rowland 2004). Current physical activity guidelines recommend children between 5 and 18 years of age to engage in at least 60 minutes moderate-to-vigorous physical activity every day (Department of Health 2011). Despite this, children, on average, are insufficiently active (Hills 2011) and engage in excessive sedentary behaviour. Specifically, only 41 percent of boys and 30 percent of girls in Wales meet these recommended guidelines1. Moreover, according to the Department of Health, more than 30 percent of 5 to 12 year old children in the UK are obese, with Wales leading at 36 percent2.

Many interventions have been conducted to reverse childhood overweight and obesity, employing a variety of strategies to enhance levels of habitual physical activity and reduce time spent in sedentary behaviours. Schools have been identified as a key context to implement such physical activity promoting interventions, given that children spend, on average, 40 percent of their waking time there (Fox 2004). Despite this, school-based interventions have been conducted with varied success (Summerbell 2012), which could be attributed to the different intervention strategies and variable methodological quality, such as lack of objective measurements of physical activity (Mountjoy 2011). Furthermore, interventions targeting reduced sedentary behaviour tend to discourage highly valued behaviours, such as engagement with technology. Therefore, there is a need to integrate such technological behaviours into future interventions. Some interventions have sought to do this. Specifically, ambient displays, also known as glanceable displays, which are peripheral, aesthetically pleasing displays of information that support awareness of some data, can be utilised to make information visible in an appealing and socially interactive manner. They are designed to be looked at occasionally without distracting us from our activities (Rogers 2010).Consolvoetala et al. (Consolvo 2008, Consolvo 2008a) integrated an interactive mobile fitness application with a glanceable display finding that those individuals utilising an awareness display maintained their physical activity levels better in comparison to those with no ambient display. However, such devices have inherent problems, such as monitor placement (Trost 2005), and have not been incorporated into a community-based settings, especially targeting children. Therefore, the aim of the present pilot study was to utilise ambient displays in order to provide near real-time visual feedback on physical activity levels during school time.

  1. http://www.bhfactive.org.uk/young-people-key-facts/index.html

  2. http://www.dh.gov.uk/health/2012/04/obesityfacts/

Personal Activity Monitoring Devices/Apps

The simplest of self-monitoring devices are pedometers. Such devices have long been of interest to health promotion professionals for their ability to encourage more active lifestyles (Baker 2008, Copelton 2010). It has been shown that their use can lead to increased walking (Kaminsky 2013, Thomas 2006) and moderate weight loss (Richardson 2008). However, health promoting physical activity interventions have usually used pedometers alongside other resource intensive support and promotion activities, such as classroom training and face-to-face motivational sessions (Bravata 2007, Kang 2009). Pedometers exist either as dedicated devices whose sole purpose is to measure step count, or as embedded features in other equipment such as mobile phones. This dedicated nature of pedometers also presents barriers to their use, particularly for those less committed to physical activity. Users have to be committed enough to fitness and lifestyle-change to both purchase the pedometer and then remember to wear or carry it. Furthermore, questions of fashion, design and convenience can deter people from using such devices (Consolvo 2006).

In addition to collecting physical activity-related data, some personal monitoring devices also include feedback on performance. Examples include: i) the Nike+ iPod Sport Kit1 MP3 player, which tracks individual exercise levels and interrupts music to verbally report on progress; ii) the Pediluma (Lim 2011), a shoe accessory that tracks wearers’ physical activity and produces a visualisation of activity levels by varying the intensity of LED-lit custom-built enclosures that contain Arduino boards; and iii) systems such as foot pods2, Jawbone UP3, Misfit Shine4 and FitBit5 monitor various activities throughout the day and synchronise the data almost in real time to a mobile app or a web-based interface, where they are used either in an online coaching system or as an input to a virtual competition with a computer-generated partner.

Mobile-phone based pedometer apps6 have the advantage of being embedded in equipment that people already own and keep attached to them. Smartphones offer various sensors to infer and record physical movements throughout the day and provide a convenient platform to give feedback about activity levels. Their in-built display and communication capabilities allow users not only to view their feedback but also to share it with others. As a result, smartphones are increasingly being used to address the problem of sedentary lifestyles (Consolvo 2008, Fogg 1998, King 1999). Some mobile apps also allow the diarisation of physical activity data in order to provide additional motivation for physical activity (Slootmaker 2009). That said, to accurately measure and validate physical activity levels, there needs to be a consistent monitor placement (Trost 2005), which is highly unlikely when integrated into mobile devices.

Social sharing and team working

Sharing physical activity data with friends and peers and participating in team-based physical activity games have been an emerging research theme in HCI and non-HCI literature in the past few years. The following sections review some examples published in the HCI literature that have applied the power of participating in physical activity as a team, social sharing and peer support.

Social fun and games

Playfulness and enjoyment are important for the achievement of changes in physical activity levels because they stimulate positive emotional states that help to motivate increased physical activity (Blythe 2005). Several physical activity games attempt to make use of this effect. In one such game, Fish’n’Steps (Lin 2006), users are presented with a fish avatar whose growth, emotional state and behaviour reflects the participant’s recent physical activity. Moreover, Fish’n’Steps includes behavioural goals in a team-based game, with the team-members being responsible for the health and growth of their fish in a shared fish tank. Consistent with the findings in Blyth et al. (Blythe 2005), the evaluation of Fish‘n’Steps showed rather than provoking increased physical activity, an unhappy avatar (sad or unhealthy looking fish in this example) could cause users to simply stop using an app. In Neat-O-Game (Fujiki 2007), a wearable accelerometer provides data that is used to control an avatar that represents the player in a virtual race. Multiple players can participate in Neat-O-Game. Winners are declared on a daily basis and players can use activity points that they have gained to receive hints in mental games such as Sudoku that are included in the app. In a different example, rather than trying to motivate individuals to reduce the time spent on sedentary activities, Berkovsky et al. (Berkovsky 2009) focus on integrating physical activity into the predominantly sedentary activity of computer gaming. Berkovsky et al. (Berkovsky 2009) integrate a novel game design called “Play, Mate!” into an open source game called Neverball and raise players’ motivation by increasing the difficulty of the game and including elements of physical activity. Berkovsky et al. (Berkovsky 2009) conducted an experimental evaluation of Neverball involving 180 users aged 9 to 12 unfamiliar with this game. They divided participants into two groups: 90 played the sedentary version of Neverball and 90 played the active version of Neverball. The experiment showed that children performed more physical activity and decreased sedentary playing time when they used the active version of Neverball. Furthermore, children did not report any reduction in their enjoyment of playing.

Social games, which encourage physical activity, are not limited to desktop or fixed platforms. Such games are rapidly growing in the smartphone market. For example, Ahtinen et al. (Ahtinen 2010) designed a team-based mobile wellness app called “Into”, which visualised the number of steps for its users with an analogy of a virtual trip from one place to another. As the team members took steps, the application combined the achievements of each team member and visualised the combined progress as a trip between the departure and destination places. Ahtinen et al. conducted a qualitative pilot study with 37 participants (a total of 12 people in four teams) over a period of one week. The more participants took steps, the quicker the line between the places turned to green (achieved goal). The line between the departure and destination places reflected the true distance between the places in the physical world, and respectively the users needed to take as many steps together as the real distance required. The findings from this study showed that setting departure and destination places and viewing up-to-date progress between them can motivate individuals. However, Ahtinen et al. did not report on the impact of feedback and or goal-settings on physical activity.

Sharing data

There is an abundance of research in HCI literature relating to social sharing as a tactic to increase awareness and to promote physical activity (Anderson 2007, Consolvo 2006, Harries 2013, Toscos 2006). For example, Shakra (Anderson 2007), Houston (Consolvo 2006), and Chick-Clique (Toscos 2006) not only allow individuals to self-monitor and set personal goals but also allow groups of friends to share performance data using mobile platforms. These systems integrate social influence through social facilitation and social support. bActive, a smartphone app (Harries 2013), employs a social norms approach, showing people what other people do, in order to influence them. Harries et al. investigated this approach amongst 152 young to early middle-age men through a rigorous and large-scale field trial. Unlike most similar active-lifestyle apps, bActive does not need to be activated prior to participating in physical activity and needs no special additional equipment. As a result, it requires minimal initial commitment from the user. The longitudinal data analysis of 6-week randomised control trial of the bActive shows that the number of steps the 22-40 year-old participants walked was 64% higher, on average, if they used the bActive app.

Foster et al. (Foster 2010) included the social sharing feature in Step Marton, a Facebook app, to create social and competitive context for daily pedometer readings in order to motivate physical activity at the workplace. They studied two versions of the app (social vs. individual) with 10 nurses (1 male) over a period of 21 days and claimed that the total number of steps taken was significantly higher when participants used the social condition (\(Z= -2.5, N=10, p=0.013\)).

Unlike games such as Fish’n’Steps (Lin 2006) where the competition is explicitly introduced, Houston (Consolvo 2006) and Chick-Clique (Toscos 2006) users do not explicitly compete with one another but can view and comment on the progress of their peers. During a three-week evaluation of Houston with young female friends (N=13) Consolvo et al. claimed that the sharing groups were significantly more likely to meet their goal (\(t=2.60, p < 0.05\)). They also reported that average daily step counts increased from the baseline week to the two weeks for seven participants in social sharing groups with goal sharing: daily averages exhibited increases from between 5% to 61% extra steps (median increase: 30%); the average daily step count increased from 180 to 4,587 steps/day (median: 2,234). Shakra (Anderson 2007) tracks the daily physical activities of people, using an Artificial Neural Network (ANN) to classify different activities throughout the day. In a short-term study (10 days) of the prototype that shared activity information amongst groups of friends and or co-workers (an average of three people in three groups), Anderson et al. reported that awareness encouraged reflection on, and increased motivation for, daily activity. In Chick-Clique (Toscos 2006) the step counts of teenage girls were recorded by pedometers and automatically transmitted to peers’ and friends’ PDAs for sharing. Some teenagers participated in this study expressed concern with regards to negative effects of competition i.e. excessive physical activity levels damaging their friendship. However, some said that it helped them to become more comfortable about talking about physical activity engagement with their friends.

Although there are many examples in the HCI literature which have included social sharing to promote physical activity behaviour change, the effectiveness of this method is unconvincing. For example, the relatively short but big sample size in the bActive study did not find any evidence to suggest that social norms are an essential component of such an app; instead the study found that the impact of feedback limited to individual performance exhibited no significant difference to that of feedback that also included social data. The relatively short data collection periods and small sample size utilized raise questions about the claims made for Step Marton, Houston and Chick-Clique. Moreover, given the lack of precision both in terms of assessing baseline levels and the limited time period over which Houston study data were collected raises doubts over the efficacy of the conclusions drawn, and the sustainability of the changes observed. As with the Shakra pilot study Anderson et al. only observed some encouragement among ‘buddies’ and, in some cases, strong competition but they did not do and or report any quantitative or qualitative analysis on physical activity measures.

Goal-setting and Feedback

Goal-setting is considered a key feature of technologies intended to encourage physical activity (Consolvo 2006). The most prominent approach to determining physical activity goals can be found in Chick-Clique (Toscos 2006), where users set their own daily step-count goals. In this example, the locus of control remains with the user, who has to determine the degree of change that he or she wants to make. One potential problem with this approach is that it relies on the user to set an appropriately challenging goal. If users set goals are either too difficult or too easy, they can fail to inspire change (Pearson 2012). To avoid setting inappropriate goals, Houston (Consolvo 2006) and Fish’n’Steps (Lin 2006) automatically create each user’s step-count goal based on their baseline step-count. UbiFit Garden (Consolvo 2008, Consolvo 2008a), a mobile fitness app, consists of a fitness device, an interactive application and a glanceable display. The goal attainment in this system is tracked through the glanceable display (i.e., how well the garden is looked after). The interactive application includes detailed information about an individual’s physical activities and a journal where activities can be added, edited, and deleted. The study showed that participants who had an awareness display were able to maintain their physical activity level better than did participants who did not have such a display.

Feedback in Chick-Clique was presented to users in the form of text messages, whereas Houston, Fish’n’Steps and UbiFit Garden provided visual feedback to the users when daily goals were achieved. Feedback varied and ranged in complexity from an asterisk annotation in Houston to the development of animated characters in Fish‘n’Steps. The motivation behind the development of a character goes beyond the straightforward mechanism of feedback, aiming to cultivate “a strong internal locus of control through care of pet or plants”. In the case of UbiFit Garden the individual’s level of physical activity was reflected in the “liveliness” of a garden environment visualised on the individual’s mobile phone screen.

Other researchers have started to investigate the potential for the use of ambient persuasion, in which visual feedback about physical activity levels is presented peripherally, such as through an augmented mirror1 (Fujinami 2008) or using tablet-based information visualisation methods (Fan 2012). In all the technologies reviewed in this paper, feedback either through a visual or a music/haptic display has been present and has shaped a significant part of the choice architecture depending on the negative or positive nature of the feedback. In the other hand, goal-setting is rarely used as a standalone motivational tool. More commonly, it is framed within the context of games (Ahtinen 2010, Berkovsky 2009, Gasser 2006, Lin 2006) or social awareness applications (Consolvo 2006, Toscos 2008) that explicitly introduce aspects of teamwork, competition and social facilitation into the process of behavioural change.

  1. A technology that uses a one-way mirror to achieve an intuitive augmented reality environment

Mission Possible: Ubiquitous Social Goal Sharing

This paper presents a pilot study using ambient displays to provide near real-time visual feedback on physical activity in a school environment.


All children within Year 5 (9-10 years old) from Bigyn Primary School in Llanelli were invited to take part in the study (n=32). Available resources dictated that this was the maximum number of participants that could be recruited to pilot the intervention, thus statistical methods were not used to determine samples sizes and no control group was measured. Written informed parental consent and participant assent were received from all children (100% participation rate).

Outcome Measures

Ethics approval to conduct the study was obtained from Swansea University’s A-STEM (Applied Sports Technology Exercise and Medicine Research Centre) ethical advisory committee. After securing the co-operation of the school, all Year 5 children were invited to take part via an information sheet and parents were provided with a consent form to complete if they wanted their child to participate. Additionally, children provided informed assent. Baseline data collection measures were completed in April 2013 and post-intervention measures were completed after the 4 week intervention period in June 2013. Children completed the Physical Self-Perception Profile for Children (PSPP-C) (Whitehead 1995). Stature and sitting stature to the nearest 0.1cm (Seca Ltd. Birmingham, UK) and body mass to the nearest 0.1 kg (Seca Ltd. Birmingham, UK) were measured using standard techniques (Lohman 1988). Body mass index was calculated (body mass (kg) / stature2 (m2)) and BMI z-scores were assigned to each participant (Cole 1995). Waist circumference was measured to the nearest 0.1 cm using a non-elastic anthropometric tape and measurements were taken at the narrowest point between the bottom of the ribs and the iliac crest. All measurements were undertaken by the same trained researchers. The 20m shuttle run test was conducted to provide an estimate of cardiorespiratory fitness (CRF). This test has been widely used in children of similar age (EUROFIT 1998, Stratton 2007, van Mechelen 1986). The total number of completed 20m runs was used as a marker for CRF. Data were analysed using a mixed “between-within” analysis of variance (ANOVA), with group as a between-participant factor and time as a within-participant factor. Statistical significance was accepted at P<0.05. All statistical analyses were conducted using IBM Statistics 21 (SPSS, Chicago, IL). All data are presented as means \(\pm\) SD. Statistical significance was accepted when P \(\leq\) 0.05.


The design of Mission Possible draws on the principles discussed in the previous section, but differs in a number of key ways from most of the ubicomp technologies. Firstly, Mission Possible allows school children to reflect on the process of increasing their activity levels for each day of each mission in a playful way. Children have to work as a team and ensure that a representative of their team is wearing the Fitbit; there is no requirement for data entry (as is needed for diarisation). Secondly, children get the reward of seeing their activity data without having to make any initial effort, or to remember to switch the recording function or the display on and off. They just have to clip on the Fitbit - the ambient display is already part of their schoolroom environment.

A third difference concerns goal-setting. Formal goal-setting, training and coaching are replaced in Mission Possible by users’ engagement with the information on the ambient display in terms how their team is performing and includes performance of the other teams. As a result, rather than feeling that they are engaging in a formalised exercise program (e.g. physical education), children are allowed to respond to this information in whatever way they wish. As argued by Thaler and Sunstein (Thaler 2009), behavioural feedback forms part of the choice architecture that nudges behaviour. In this case, the feedback from the ambient display nudges children to be little bit more active than a day before; having the LED display right at the front of the classroom allows them to occasionally glance at their performance data during the day.

The fourth difference concerns the structure of the social interaction. Almost all of the work reviewed in the social sharing section focused on participants remotely contributing to their team’s goal. For example, Into, Fish’n’Steps, Houston, Shakra and Chick-Clique individual members of teams have a physical activity monitoring device, set goals as an individual or a team and work toward those goals remotely through a digital medium (no face to face interaction with team members happens). Mission Possible breaks away this individualistic digital bubble phenomenon 1 and allows children through shared Fitbits and the ambient display, to be more creative, playful and thoughtful of each other.

Although it differs from other ubicomp technologies in the respects listed, Mission Possible shares with them the desire to be interesting and fun to use. To this end, Mission Possible draws on the experiences and lessons of other systems. It gives positive reinforcement (learning from the success of UbiFit Garden and Houston and from the problems experienced by Fish’n’Steps); similar to Houston, Chick Clique, UbiFit Garden, Into and Shakra, provides opportunities for teams to reflect on their activity. Finally, like the social gaming and social data sharing features of Fish’n’Steps, Houston, Chick-Clique, Shakra and Into, the social norms information within Mission Possible is designed not only to prompt increased physical activity, but also to encourage engagement with the feedback i.e. the displays of their team and others’ activity levels. The intervention design and content were informed by formative work with the children involved, thereby employing a user-centred design. The children were divided into four groups, where each group generated ideas for possible missions. These included ideas like:

  • playing tag with laser guns, bouncy balls or velcro stars

  • who can do the assault course the fastest

  • “aliens vs. cowboys” or “zombies vs. humans”

  • capture the flag

  • hide and seek using climbing wall

We selected feasible ideas from the lists, and generated descriptions of the missions based around a “secret agent” type theme:

  • Mission 1: Assault course - “Captain Cybernetic wants you to infiltrate Dr. Tempus’s lair. However, Dr. Tempus has placed various obstacles in your path. Navigate the obstacles to get to his lair as fast as possible.”

  • Mission 2: Code hunters - “A video clue is shown on the iPad, where you have to find a hidden token. The token must be returned to the teacher, who will reveal the next video clue the following day. The tokens from Monday to Thursday will be needed to find the final token on Friday.”

  • Mission 3: Capture the flag - “Two teams; where each team tries to get the other team’s flag while protecting their own.”

  • Mission 4: A race against time - “Captain Cybernetic has gained access to a high-power laser beam that is powered by your feet. You and your teammates need to work together to rack up as much steps as possible. Your step count will be used together to power the laser beam and destroy the evil Dr. Tempus’s lair.”

Design and methodology

Children were divided into 10 groups, where each group was assigned a uniquely distinguishable colour to represent their group on the ambient display. Flexible LED lighting was installed along the bookshelf in the front of the classroom.

The display was connected via a microcontroller to the school’s computer network. Activity monitor data were retrieved from the Fitbit website and displayed with moving colour segments, where different colours were used to represent the various teams. The teams could make their colour segment go faster by increasing their number of steps. The team’s colour would start to flash if the spent more time being active than the previous day.

For the missions, ideas were elicited directly from the schoolchildren, with a brainstorming activity performed at the school. These ideas were then refined by the project team and turned into four activity-based missions.

Hardware design

Ambient displays located in a person’s environment have been shown to change their behaviour by providing live feedback (Fan 2012, Harries 2013). These displays serve as decorative visual art pieces intended for reflection.

An ambient display, consisting of a 4m-long lighting strip with 240 individually controlled LED lights, was designed and constructed by Swansea University and installed in the classroom. Data collected from 10 Fitbit activity monitors were visualised on the lighting strip with different lighting patterns. The ambient display connects via a network cable to the Internet in order to download the Fitbit data.

To visualise the performance of each of the 10 groups, uniquely distinguishable moving colour segments were used. Each colour segment had a speed betwen 0 and 10, based on the group’s performance. To calculate the speed of each group we used the following formula:

\[v = 5 \frac{x}{y} + 1\]

where \(v\) is the speed at which the segment is moving, \(x\) is the accumulated number of steps taken that day, and \(y\) is the number of steps taken on the previous day. On the first day of each mission, \(y\) was set to 1000 as a default value.

To provide additional encouragement for the children, we used the intensity level (lightly active, fairly active and very active) while they were performing the missions. We compared the number of minutes they spent at the ‘very active’ intensity level to complete the mission on that day with the previous day. If they were doing more vigorous exercise than the previous day, their group’s colour segment would start pulsating.

A DVD outlining the various missions, together with a Teacher’s Guide mission pack, was provided to the teacher, and the mission videos installed on a set of tablets belonging to the school.

Measurement of physical activity

The Fitbit Zip (Figure \ref{fig:Fitbit}) contains a tri-axial accelerometer to record time-stamped physical activity data, including number of steps and physical activity according to four levels: sedentary, lightly active, fairly active and very active. It can be clipped to various locations on the body. For the study we asked participants to clip the device to the inside of their trouser pocket. The Zip is designed to be small and unobtrusive, making it more acceptable for use by children than other activity monitors which are often large and bulky. A recent study (Adam Noah 2013) has indicated that the Fitbit is reliable and valid for activity monitoring (in terms of step counts).

Data from the activity monitors were uploaded automatically to Fitbit’s servers as the children returned from completing the mission for the day, via a Bluetooth Low Energy (BLE) dongle connected to a computer in the classroom. Data was accessed using the Fitbit API, analysing the results and storing them on the university’s server. The results were then downloaded by the Arduino (see Figure \ref{fig:arduino}), connected with an Ethernet shield to the internet via the school’s network. Note that the activity monitor was worn from the start of the activity until the end of the activity, and not the entire day.

The Fitbit trackers syncs to the Fitbit servers every 15-20 minutes, as long as they are within 15 feet (approximately 5 metres) of the BLE dongle. Each Fitbit tracker has its own ID, which means that one BLE dongle can sync multiple devices. Whenever new data is available on Fitbit’s servers, it is synchronised with our university server. It takes approximately 1-2 minutes to transfer all the activity data from the Fitbit servers. The Arduino polls the university server every 5 minutes for the results to be displayed on the ambient display.

  1. Also called Mindless Technology by Professor Yvonne Rogers in her keynote speech at INTERACT 2013 http://www.interact2013.org/Keynotes

Fitbit Zip activity monitors \label{fig:Fitbit}

\label{fig:arduino} Hardware used for ambient display: Arduino microcontroller with Ethernet shield, connected to WS2811 RGB LED strip (60 LEDs/m)

Ambient display installed against a bookcase in the corner of the classroom


To complete all four missions, the children spent approximately 3,290 minutes (equivalent to 55 hours). They were able to take 159,549 steps during this period of time.

For Mission 1, the teacher recorded the time (in seconds) required for each team to complete the assault course using a stopwatch, as shown in Figure \ref{fig:mission1}. As the Fitbit for the yellow team was malfunctioning on the Tuesday, they did not compete that day. Figure \ref{fig:fit1} shows the amount of physical activity performed (in minutes) by each team for each day of the week, where the stacked bar is showing the level of physical activity. The number of steps recorded by the Fitbit is also plotted on the same graph, and can be read on the second y-axis to the right of the graph.

In Figure \ref{fig:mission2} the time to find the token for Mission 2 (in minutes) is shown for each team. On the Monday the silver team spent a large amount of time (23:10) compared to other days in order to find the token. As such the data point was shortened to make the graph more readable. The data recorded by the Fitbit is shown in Figure \ref{fig:fit2}.

The number of points scored during Mission 3, recorded by the teacher and grouped by team, is shown in Figure \ref{fig:mission3}. The related Fitbit data is shown in Figure \ref{fig:fit3}. Figure \ref{fig:mission4} shows the time taken (in seconds) for each team to complete Mission 4, as recorded by the teacher using a stopwatch. The Fitbit data for Mission 4 is shown in Figure \ref{fig:fit4}.

Overall, there were no significant differences in any physical self-perceptions over time (P>0.05). Results are shown in Table \ref{tab:psppc}.


Table. Physical self-perceptions. Note: Values reported as mean \(\pm\) SD
Sport competence Physical condition Body attractiveness Perceived strength Physical self-worth Global self-esteem
Baseline \(2.96 \pm 0.13\) \(2.95 \pm 0.12\) \(2.81 \pm 0.14\) \(2.91 \pm 0.11\) \(3.07 \pm 0.12\) \(3.26 \pm 0.11\)
Post \(2.88 \pm 0.14\) \(2.88 \pm 0.14\) \(2.97 \pm 0.16\) \(2.89 \pm 0.12\) \(3.10 \pm 0.13\) \(3.31 \pm 0.13\)

\label{fig:mission1} Time to finish the assault course (Mission 1), grouped by team

\label{fig:fit1} Mission 1 - amount of physical activity; number of steps

\label{fig:mission2} Time to find the token (Mission 2), grouped by team. The bar for data point 23:10 was shortened to make the graph more readable.

\label{fig:fit2} Mission 2 - Amount of physical activity; number of steps for each each group

\label{fig:mission3} Number of points scored during the mission (Mission 3), group by team

\label{fig:fit3} Mission 3 - Amount of physical activity; number of steps for each each group

\label{fig:mission4} Time in seconds for each group to complete Mission 4, grouped by team

\label{fig:fit4} Mission 4 - Amount of physical activity; number of steps for each each group


This study underscores the value of games in not only increasing children’s awareness of their physical activity levels, but encouraging them to increase them. Such an approach can foster long-term behavioural change and therefore future research should seek to incorporate longer intervention periods. The present pilot study, as opposed an observational or randomised controlled trial, reports the effects of the design choice and offers specific component guidance, which could be refined for future research.

Mission 1

As shown in Figure \ref{fig:mission1}, the fastest teams completed the assault course in less than 80 seconds, but were active for around 10-20 minutes during the session. This shows that the time spent engaging with the activity itself does not have to last the whole session to be effective. For this activity, the time spent very active (as shown in Figure \ref{fig:fit1}) does not seem to be increasing or decreasing significantly over the week, probably due to the total engagement time being so short.

While some teams (e.g. red, purple) seem to have improved over the course of the week, this was not the case for all teams. Care should be taken if the metric “complete activity faster than previous day” is used as a trigger for a motivator, where the motivator is e.g. a short animation or motivational piece of text.

Mission 2

The time spent finding the token (Figure \ref{fig:mission2}) is quite variable across the different days of the week, given the nature of the activity. However, completing the mission quickly (e.g. the green team on Monday, Wednesday and Thursday) did not necessarily mean that they would spend less time being active during the entire session.

One outlier is the silver team, which took an unusually long time (23:10) to find the token on the first day. This is also reflected in their step count and amount of physical activity, as shown in Figure \ref{fig:fit2}.

Mission 3

This mission stimulated the largest amount of physical activity during the session compared to the other missions. Teams consistently spent at least 10 minutes active during the session, with high levels of being fairly active and very active.

Mission 4

In this mission, where teams were specifically asked to try and increase their step count compared to the previous day, the amount of physical activity performed fluctuates throughout the week. The time recorded as being very active is less than with Mission 3.

Implementation issues

One of the biggest obstacles during the installation of the ambient display was trying to establish a connection to an internet server via the school network. As we were using an Arduino microcontroller with an Ethernet shield, it was quite challenging to negotiate network authentication via the school’s proxy server. If possible, it is recommended to use a GPRS/GSM shield or similar when installing Internet-connected prototypes in-the-wild, as the existing network infrastructure could cause unexpected problems.

It can also be challenging to come up with ten easily distinguishable colours using RGB LEDs. When mixing RGB colours using LEDs, the colour black is not available, and it becomes difficult to tell the colours brown and orange apart. Figure \ref{fig:colours} show the final colour palette we came up with.

Other practical considerations include that if you want the teacher to provide data, give a template with the required format. This helps to not have to unnecessarily transcribe data, for example converting “1 min 48 sec” to “00:01:48”.

\label{fig:colours} Colours used on the ambient display to represent the ten teams, with hexadecimal colour code used in parentheses


Though some children’s BMI decreased, this was not statistically significant. In order to ascertain a more accurate interpretation of the intervention effect, the intervention would need to run for longer, have a comparison group, and have a greater sample size. Moreover, future interventions should consider intervention fidelity and incorporate a more thorough process evaluation. That said, teacher and children’s feedback was that all children exposed to the intervention enjoyed participating in physical activity, regardless of whether they previously enjoyed such physical activities.


Funded by the Creative Exchange Wales Network (CEWN). This work was also supported by the UK Engineering and Physical Science Research Council [EP/G059063/1]. A DVD containing the weekly mission video segments, together with a Teacher’s Guide mission pack, were designed and created by our creative partner, Redhead Consultancy. We would also like to thank everyone at Bigyn Primary School who assisted us during this study.


  1. J. Adam Noah, David K. Spierer, Jialu Gu, Shaw Bronner. Comparison of steps and energy expenditure assessment in adults of Fitbit Tracker and Ultra to the Actical and indirect calorimetry. Journal of Medical Engineering & Technology 37, 456-462 Informa UK (Informa Healthcare), 2013. Link

  2. bibliography entry failed to display
  3. Ian Anderson, Julie Maitland, Scott Sherwood, Louise Barkhuus, Matthew Chalmers, Malcolm Hall, Barry Brown, Henk Muller. Shakra: tracking and sharing daily activity levels with unaugmented mobile phones. Mobile Network Applications 12, 185–199 Springer-Verlag New York, Inc., 2007. Link

  4. Graham Baker, Stuart R. Gray, Annemarie Wright, Claire F. Fitzsimons, Myra A. Nimmo, Ruth Lowry, Nanette Mutrie. The effect of a pedometer-based community walking intervention “Walking for Wellbeing in the West” on physical activity levels and health outcomes: A 12-week randomized controlled trial. International Journal of Behavioral Nutrition and Physical Activity 5, 44–44 BioMed Central, 2008. Link

  5. Shlomo Berkovsky, Jill Freyne, Mac Coombe, Dipak Bhandari, Nilufar Baghaei. Physical activity motivating games: you can play, mate!. 411, 273–276 In OZCHI. ACM, 2009. Link

  6. Shlomo Berkovsky, Dipak Bhandari, Stephen Kimani, Nathalie Colineau, Cécile Paris. Designing games to motivate physical activity. 350, 37 In PERSUASIVE. ACM, 2009. Link

  7. Mark Blythe, Kees Overbeeke, Andrew F. Monk, Peter C. (Eds.) Wright. Funology - From Usability to Enjoyment. 3 Springer, 2005. Link

  8. L. M. Boddy, A. F. Hackett, G. Stratton. Changes in fitness, body mass index and obesity in 9-10 year olds. J Hum Nutr Diet 23, 254-9 (2010).

  9. Dena M. Bravata, Crystal Smith-Spangler, Vandana Sundaram, Allison L. Gienger, Nancy Lin, Robyn Lewis, Christopher D. Stave, Ingram Olkin, John R. Sirard. Using pedometers to increase physical activity and improve health: A systematic review. JAMA 298, 2296-2304 (2007). Link

  10. T. J. Cole, J. V. Freeman, M. A. Preece. Body mass index reference curves for the UK, 1990. Arch Dis Child 73, 25-9 (1995).

  11. Sunny Consolvo, Predrag Klasnja, David W. McDonald, Daniel Avrahami, Jon Froehlich, Louis LeGrand, Ryan Libby, Keith Mosher, James A. Landay. Flowers or a robot army?: encouraging awareness & activity with personal, mobile displays. 54–63 In Proceedings of the 10th international conference on Ubiquitous computing. ACM, 2008. Link

  12. Sunny Consolvo, David W. McDonald, Tammy Toscos, Mike Y. Chen, Jon Froehlich, Beverly Harrison, Predrag Klasnja, Anthony LaMarca, Louis LeGrand, Ryan Libby, Ian Smith, James A. Landay. Activity sensing in the wild: a field trial of UbiFit garden. 1797–1806 In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. ACM, 2008. Link

  13. Sunny Consolvo, Katherine Everitt, Ian Smith, James A. Landay. Design requirements for technologies that encourage physical activity. 457–466 In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. ACM, 2006. Link

  14. Denise A. Copelton. Output that Counts: Pedometers, Sociability and the Contested Terrain of Older Adult Fitness Walking. 127–141 In Technogenarians. Wiley-Blackwell, 2010. Link

  15. Physical Activity Health Improvement Department of Health, Protection. Start Active, Stay Active: A report on physical activity from the four home countries’ Chief Medical Officers. (2011).

  16. EUROFIT. European Test of Physical Fitness. Committee for the Development of Sport, Council of Europe, 1998.

  17. Chloe Fan, Jodi Forlizzi, Anind Dey. A Spark Of Activity : Exploring Informative Art As Visualization For Physical Activity. In 14th International Conference on Ubiquitous Computing (UBICOMP ’12). (2012).

  18. BJ Fogg. Persuasive computers: perspectives and research directions. 225–232 In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. ACM, 1998. Link

  19. Derek Foster, Conor Linehan, Ben Kirman, Shaun Lawson, Gary James. Motivating physical activity at work: using persuasive social media for competitive step counting. 111–116 In Proceedings of the 14th International Academic MindTrek Conference: Envisioning Future Media Environments. ACM, 2010. Link

  20. K R Fox, A Cooper, J McKenna. The school and promotion of children’s health enhancing physical activity. Journal of School Health 23, 338-358 (2004).

  21. D. S. Freedman, H. S. Kahn, Z. Mei, L. M. Grummer-Strawn, W. H. Dietz, S. R. Srinivasan, G. S. Berenson. Relation of body mass index and waist-to-height ratio to cardiovascular disease risk factors in children and adolescents: the Bogalusa Heart Study. Am J Clin Nutr 86, 33-40 (2007).

  22. Yuichi Fujiki, Konstantinos Kazakos, Colin Puri, Ioannis Pavlidis, Justin Starren, James Levine. NEAT-o-games: ubiquitous activity-based gaming. 2369–2374 In CHI ’07 Extended Abstracts on Human Factors in Computing Systems. ACM, 2007. Link

  23. Kaori Fujinami, Jukka Riekki. A Case Study on an Ambient Display as a Persuasive Medium for Exercise Awareness. 266–269 In Proceedings of the 3rd international conference on Persuasive Technology. Springer-Verlag, 2008. Link

  24. Roland Gasser, Dominique Brodbeck, Markus Degen, Jürg Luthiger, Remo Wyss, Serge Reichlin. Persuasiveness of a Mobile Lifestyle Coaching Application Using Social Facilitation. 3962, 27-38 In PERSUASIVE. Springer, 2006. Link

  25. T. Harries, P. Eslambolchilar, C. Stride, R. Rettie, S. Walton. Walking in the Wild – Using an Always-on Smartphone Application to Increase Physical Activity. In 15th International Conference on Ubiquitous Computing (UBICOMP ’13). (2013).

  26. A. P. Hills, L. B. Andersen, N. M. Byrne. Physical activity and obesity in children. British Journal of Sports Medicine 45, 866-70 (2011).

  27. Leonard A. Kaminsky, Jason Jones, Katrina Riggin, Scott J. Strath. A pedometer-based physical activity intervention for patients entering a maintenance cardiac rehabilitation program: a pilot study. Cardiovascular Diagnosis and Therapy 3 (2013). Link

  28. Minsoo Kang, Simon J. Marshall, Tiago V. Barreira, Jin-Oh Lee. Effect of pedometer-based physical activity interventions: a meta-analysis.. Res Q Exerc Sport 80, 648-55 (2009). Link

  29. Phillip King, Jason Tester. The landscape of persuasive technologies. Commun. ACM 42, 31–38 ACM, 1999. Link

  30. Brian Y. Lim, Aubrey Shick, Chris Harrison, Scott E. Hudson. Pediluma: motivating physical activity through contextual information and social influence. 173-180 In Tangible and Embedded Interaction. ACM, 2011. Link

  31. James J. Lin, Lena Mamykina, Silvia Lindtner, Gregory Delajoux, Henry B. Strub. Fish’n’Steps: encouraging physical activity with an interactive computer game. 261–278 In Proceedings of the 8th international conference on Ubiquitous Computing. Springer-Verlag, 2006. Link

  32. TG. Lohman, A.F. Roche, R. Martorell. Anthropometric standardization reference manual. Human Kinetics, 1988.

  33. M Mountjoy, L B Andersen, N Armstrong, S Biddle, C Boreham, H P Bedenbeck, U Ekelund, L Engebretsen, K Hardman, A Hills, S Kahlmeier, S Kriemler, E Lambert, A Ljungqvist, V Matsudo, H McKay, L Micheli, R Pate, C Riddoch, P Schamasch, C J Sundberg, G Tomkinson, E van Sluijs, W van Mechelen. International Olympic Committee consensus statement on the health and fitness of young people through physical activity and sport. British Journal of Sports Medicine 45, 839-848 (2011).

  34. Erin S. Pearson. Goal setting as a health behavior change strategy in overweight and obese adults: A systematic literature review examining intervention components. Patient Education and Counseling 87, 32 – 42 (2012). Link

  35. Caroline R. Richardson, Tiffany L. Newton, Jobby J. Abraham, Ananda Sen, Masahito Jimbo, Ann M. Swartz. A Meta-Analysis of Pedometer-Based Walking Interventions and Weight Loss. The Annals of Family Medicine 6, 69-77 (2008). Link

  36. Yvonne Rogers, William R Hazlewood, Paul Marshall, Nick Dalton, Susanna Hertrich. Ambient Influence : Can Twinkly Lights Lure and Abstract Representations Trigger Behavioral Change ?. 261–270 In Proceedings of the 12th ACM international conference on Ubiquitous computing (Ubicomp’10). (2010).

  37. B Rokholm, J L Baker, T I A Sørensen. The levelling off of the obesity epidemic since the year 1999 - a review of evidence and perspectives. Obesity Reviews 11, 835-346 (2010).

  38. T.W Rowland. The childhood obesity epidemic: Putting the ’dynamics’ into thermodynamics.. Pediatric Exercise Science 16, 87-93 (2004).

  39. S. M. Slootmaker, A. Paw M. J. M. Chin, A. J. Schuit, W. van Mechelen, L. L. J. Koppes. Concurrent validity of the PAM accelerometer relative to the MTI Actigraph using oxygen consumption as a reference. Scand J Med Sci Sports 19, 36-43 (2009). Link

  40. G. Stratton, D. Canoy, L. M. Boddy, S. R. Taylor, A. F. Hackett, I. E. Buchan. Cardiorespiratory fitness and body mass index of 9-11-year-old English children: a serial cross-sectional study from 1998 to 2004. Int J Obes (Lond) 31, 1172-8 (2007).

  41. C D Summerbell, H J Moore, C Vögele, S Kreichauf, A Wildgruber, Y Manios, W Douthwaite, C A Nixon, E L Gibson. Evidence-based recommendations for the development of obesity prevention programs targeted at preschool children. Obesity Reviews 13, S129-S132 (2012).

  42. Richard H. Thaler, Cass R. Sunstein. Nudge : improving decisions about health, wealth, and happiness. Penguin Books, 2009. Link

  43. Lyndall Thomas, Mark Williams. Promoting physical activity in the workplace: using pedometers to increase daily activity levels. Health Promot J Austr 17, 97-102 (2006). Link

  44. Tammy Toscos, Anne Faber, Shunying An, Mona Praful Gandhi. Chick clique: persuasive technology to motivate teenage girls to exercise. 1873–1878 In CHI ’06 Extended Abstracts on Human Factors in Computing Systems. ACM, 2006. Link

  45. Tammy Toscos, Anne Faber, Kay Connelly, Adity Mutsuddi Upoma. Encouraging Physical Activity in Teens: can technology help reduce barriers to physical activity in adolescent girls. In Proceedings of Pervasive Health ’08. (2008).

  46. S. G. Trost, K. L. McIver, R. R. Pate. Conducting accelerometer-based activity assessments in field-based research. Med Sci Sports Exerc 37, S531-43 (2005).

  47. J R Whitehead. A study of children’s physical self-perceptions using an adapted physical self-perception profile questionnaire. Pediatric Exercise Science 7, 132-151 (1995).

  48. W van Mechelen, H Hlobil, H Kemper. Validation of two running tests as estimates of maximal aerobic power in children. European Journal of Applied Physiology and Occupational Physiology 55, S03-S06 (1986).

[Someone else is editing this]

You are editing this file