Using Crowdsourcing to Support Pro-environmental Community Activism

Abstract

Community activist groups typically rely on core groups of highly motivated members. In this paper we consider how crowdsourcing strategies can be used to supplement the activities of pro-environmental community activists, thus increasing the scalability of their campaigns. We focus on mobile data collection applications and strategies that can be used to engage casual participants in pro-environmental data collection. We report the results of a study that used both quantitative and qualitative methods to investigate the impact of different motivational factors and strategies, including both intrinsic and extrinsic motivators. The study compared and provides empirical evidence for the effectiveness of two extrinsic motivation strategies, pointification – a subset of gamification – and financial incentives. Prior environmental interest is also assessed as an intrinsic motivation factor. In contrast to previous HCI research on pro-environmental technology, much of which has focused on individual behavior change, this paper offers new insights and recommendations on the design of systems that target groups and communities.

Full text

Using Crowdsourcing to Support Pro-Environmental
Community Activism
Elaine Massung, David Coyle, Kirsten Cater, Marc Jay, Chris Preist
Department of Computer Science,
University of Bristol,
Bristol, BS8 1UB, UK
{elaine.massung, david.coyle, kc6678, chris.preist}@bristol.ac.uk, marc.jay@bristolalumni.org.uk
ABSTRACT
Community activist groups typically rely on core groups of
highly motivated members. In this paper we consider how
crowdsourcing strategies can be used to supplement the
activities of pro-environmental community activists, thus
increasing the scalability of their campaigns. We focus on
mobile data collection applications and strategies that can
be used to engage casual participants in pro-environmental
data collection. We report the results of a study that used
both quantitative and qualitative methods to investigate the
impact of different motivational factors and strategies,
including both intrinsic and extrinsic motivators. The study
compared and provides empirical evidence for the
effectiveness of two extrinsic motivation strategies,
pointification – a subset of gamification – and financial
incentives. Prior environmental interest is also assessed as
an intrinsic motivation factor. In contrast to previous HCI
research on pro-environmental technology, much of which
has focused on individual behavior change, this paper offers
new insights and recommendations on the design of
systems that target groups and communities.
Author Keywords
Community activism; sustainability; participatory
urbanism, crowdsourcing; gamification; motivation
ACM Classification Keywords
H.5.m. Information interfaces and presentation (e.g., HCI):
Miscellaneous.
INTRODUCTION
Community activist groups, including pro-environmental
groups, are typically driven by a core group of highly
motivated individuals. These people are often willing to
dedicate a great deal of time and effort to help bring about
desired changes within their community. While they may
derive intrinsic personal satisfaction from their activities,
and enjoy the benefits of participation in a group, there is
often no expectation of, or need for, other extrinsic rewards,
e.g. financial remuneration.
The research in this paper was undertaken in collaboration
with a pro-environmental community activist group called
Close the Door (CTD). This group aims to reduce energy
waste by encouraging shop owners to keep their doors
closed during cold weather. In spite of winning national
environmental advocacy awards and having dedicated core
members, the CTD campaign faces challenges that are
common to many community activist groups. In particular,
they have found that to achieve scalable and sustainable
change, it is not sufficient to rely purely on their core
members. Activist groups also need to develop effective
strategies for drawing on support from casual volunteers.
One particular challenge for CTD is to maintain their
database of shops and monitor the behavior of shop owners
on an ongoing basis. In this paper we evaluate different
strategies through which computer-supported citizen
science and crowdsourcing techniques [6, 7, 36] can be
brought to bear to help in addressing this challenge. We
focus on developing mobile applications (apps) that allow
members of the public to undertake lightweight
environmental data collection. The aim is to increase the
scalability of activist groups like CTD by collecting data on
a much larger scale than otherwise possible and freeing
core members to focus on high impact advocacy activities.
Members of the public are unlikely to be as motivated as
community activists and overcoming public apathy is often
difficult for pro-environmental movements. The question
thus arises: what design strategies can we apply to motivate
people to engage in pro-environmental data collection? The
key contributions of this paper are to implement a set of
mobile data collection apps and then provide both
qualitative and quantitative evidence for the effectiveness
of the different motivational strategies that are applied in
the apps. Overall we developed three apps. The first used
“pointification”, a subset of gamification that uses game
mechanics such as points, badges, and leaderboards to
encourage engagement and competition [11, 27, 38]. The
second offered participants financial incentives to carry out
data collection tasks [29]. The final app did not use any
explicit motivational strategies and acted as a control.
We conducted a study in with each app was used by 16
participants for two weeks. Results show that pointification
increased performance, though not to a statistically
significant level. However, financial rewards led to a
significant increase in the amount of data collected.
Surprisingly, interest in environmental issues and existing
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for profit or commercial advantage and that copies
bear this notice and the full citation on the first page. To copy otherwise,
or republish, to post on servers or to redistribute to lists, requires prior
specific permission and/or a fee.
CHI 2013, April 27–May 2, 2013, Paris, France.
Copyright © 2013 ACM
978-1-4503-1899-0/13/04...$15.00.

tendencies towards pro-environmental behavior did not
correlate with greater data collection. Qualitative data,
collected through semi-structured interviews, allowed us to
investigate such issues in greater details. For example we
found that the factors that encourage people to initially take
part in a community activism project can differ from those
that maintain interest. Also, although motivational factors
impact a person’s use of an app, motivation is complex and
is only half the battle. Designers must also consider the
enabling factors that influence casual volunteers’ use of
crowdsourcing apps.
The research presented in this paper extends HCI research
on pro-environmental technologies in several key ways.
Prior research in this area has largely focused on systems
targeting individual behavior change [13, 17, 21]. This
paper provides new insights on systems targeting groups
and communities rather than individuals. It provides both
quantitative and qualitative evidence on the effectiveness of
different motivational and enabling strategies. By drawing
on this evidence we provide recommendations to guide the
design of crowdsourcing applications that support pro-
environmental community activism.
RELATED WORK
This paper builds on prior research on pro-environmental
behavior change, citizen science and participatory
urbanism, and motivational techniques for crowdsourced
data. We will begin by considering the pro-environmental
literature that has influenced our research.
Pro-Environmental Behavior Change
Prior HCI research on technology-supported environmental
behavior change has been dominated by systems targeting
individual behavior change [13, 17]. Such systems are often
driven by rational choice models, which assume that
individual behavior is driven by self-interest. For example,
rational economic models assume that – given access to the
relevant information – people behave in such a way as to
maximize rewards and minimize costs (although costs and
rewards need not be purely financial), thus adopting
behaviors that are advantageous [17, 21]. An alternative
approach to behavior change, which has received less
attention in pro-environmental HCI research, focuses on
communities rather than individuals. Theories such as the
norm-activation model hypothesize that individual behavior
is strongly shaped by community norms and everyday
practices [17]. Rather than focusing on individual reward,
such approaches recognize the benefits of collective and
coordinated actions. These approaches are beginning to be
applied to pro-environmental behavior [28, 33], and some
studies, such as [30], have shown that these norms can more
positively affect behavior than traditional pro-
environmental messages or messages highlighting money
saving opportunities.
The CTD campaign, and apps described in this paper, aim
to take advantage of both individual (rational choice) and
community (norm-activation) strategies.
Citizen Science and Participatory Urbanism
With the increasing ubiquity of smart phone technology,
mobile data collection through crowdsourcing is increasing.
It has been used in citizen science to collect data in support
of scientific research, including monitoring the spread of
invasive plants through the What’s Invasive app [18] and
monitoring animal population distributions through the
reporting of roadkill [5]. It has also been used
in participatory urbanism, by municipal authorities seeking
to engage people in monitoring and improving their urban
environments. Examples include monitoring noise pollution
[26] and repairing potholes and streetlights [24]. The
boundary between these two genres blurs in applications
such as water quality [23] and air quality [31] monitoring,
where the data both increases scientific understanding and
can also contribute to local management policies. Our work
can be viewed as a form of participatory urbanism, but
rather than supporting the efforts of municipal authorities it
instead supports the efforts of community activists. As
such, it adopts a similar position to the work of Kuznetsov
et al. [25], who engage local citizens with their concerns
around air quality with portable sensing equipment.
Motivating and Engaging Contributors
In common with all crowdsourcing approaches, Citizen
Science and Participatory Urbanism face key challenges in
engaging contributors – namely how to recruit them, and
how to get them to make an active ongoing contribution. In
a review of crowdsourcing literature, Doan et al. [14]
categorize common approaches, including (amongst others)
coercion, payment, intrinsic enjoyment and competition. In
citizen science literature, Kim et al. [23] report the use of
incentives, competition, entertainment and education to
support ongoing engagement. However, neither compared
or analyzed the effect of these different types of approaches
to assess their effectiveness on recruitment or contribution.
Rotman et al. [32] explored the motivation of volunteers in
projects making a scientific contribution and found a
complex network of factors at play, with initial scientific
interest and curiosity moving towards a desire for
attribution and acknowledgment over time.
Focusing on motivation, the factors that motivate
individuals can be viewed as either intrinsic or extrinsic.
People who are intrinsically motivated are willing to do an
activity “for its inherent satisfactions rather than for some
separable consequence” [10]; examples of intrinsic
motivators include interest, curiosity, competence, and
enjoyment [21]. Within the context of sustainability, De
Young [9] posits that behavior that is intrinsically
motivated is more effective for self-sustaining individual
behavioral change. In community activist campaigns such
as CTD, core members are likely to have a high degree of
intrinsic motivation. Extrinsic motivation is carrying out an
activity “to attain some separable outcome”, such as
material incentives or social reinforcement [10, 21].
“Gamification” is an approach that has rapidly gained
prominence as a motivational technique. Deterding et al.

define it as “using game design elements in non-gaming
contexts” [11]. Elements can include point scoring,
leaderboards, goal setting, questing, and artifact collecting.
Gaming techniques have been applied in the domain of eco-
feedback technology. A series of games concentrating on
creating awareness of household energy in adolescents were
trialed through the use of smart meters and mobile phone
applications [2, 19], and participants often compared the
changing ambient display of UbiGreen to gaming levels
[16]. However, these applications focused on the behavior
of an individual, with the ultimate goal of awareness or
behavior change by system users, rather than data
collection. Providing participants with small financial
rewards in return for carrying out basic tasks such as image
identification has also been successfully employing as a
way of attracting and motivating participants, e.g.
Amazon’s Mechanical Turk [29]. This approach has been
found to increase the amount of effort expended [32];
however, it has also been suggested that financial rewards
may reduce intrinsic motivation [10].
The study in this paper provides the first systematic
analysis of the effectiveness of two forms of extrinsic
motivation on environmental data collection performance: a
financial pay-for-results scheme, and a virtual reward
scheme, which uses points, badges and a leaderboard. We
also analyze the effect of intrinsic motivation by examining
if a positive disposition towards environmental behaviors
affects performance. Furthermore, through a qualitative
study, we examine the interrelationships between different
motivators and the other factors affecting the motivation
and performance of people using the CTD app.
THE CLOSE THE DOOR CAMPAIGN
On UK high streets it is common to see shop doors propped
open to encourage potential customers to step over the
threshold. However, during cold weather these open doors
allow heat to escape. A detailed study of typical 150m2 UK
high street shops found that keeping doors closed can
reduce emissions and energy from heating by 30-50% [3].
The CTD campaign was established in response to this
finding. Since its inception in 2007 it has grown to include
chapters in nine UK and two international cities, and it was
named Best Campaign by UK Climate Week 2012.
Considered purely in terms of energy costs, closing doors is
a rational choice for shop owners. However owners fear
that they will lose customers if they close their doors whilst
other shop doors remain open. This is a classic “I will if
you will” environmental problem [34]. The CTD campaign
attempts to combat this problem by sending volunteers to
shopping streets to record whether shops have their doors
open or closed. Owners or managers are then approached to
determine if they want to join the campaign – i.e. keep their
doors closed when heating or air-conditioning systems are
running. Shops that join the campaign are listed on the CTD
website and Facebook page, and are given a door sticker to
publicize their membership. To incentivize participation by
shops, the CTD team also runs a public information
campaign. Members of the public are given information
about the environmental benefits of closed doors and
encouraged to support shops carrying a CTD logo. The
message to the public is: “If a shop will not Close the Door:
don’t shop there, go elsewhere.” By targeting both shop
owners and their customers, the campaign aims to make
closed doors the accepted norm. Ultimately this new norm
benefits both shop owners and the environment.
DESIGN OF THE CLOSE THE DOOR APPS
Through conversations with members of the CTD
campaign, we found that the time spent recruiting and
retaining shop owners is critical to their success. However,
in order for this advocacy to work, members need to collect
data about shops in the local area. This is time and labor
intensive and must be undertaken on an ongoing basis to
check that shops continue to follow the Close the Door
policy. It has placed great strain on the core members of the
campaign. Working in collaboration with campaigners, we
therefore decided to develop mobile apps that allow casual
participants to record shop doors via a smartphone. If
people use our apps on a regular basis it would allow the
CTD campaign to scale up their data collection and would
also free campaigners to focus on advocacy activities.
Rather than evaluating a single data collection app, we
chose to design three iPhone apps. Each app applied a
different strategy to encourage users to collect data. By
comparing the effectiveness of these apps we aimed to
provide evidence of the effectiveness of different
motivational strategies that can be applied to engage casual
participants to collect pro-environmental data. In designing
the apps, we deliberately sought to keep things simple.
Rather than exploring a wide range of functionality and
overlapping motivational strategies, our aim was to
compare and provide strong evidence on the relative
effectiveness of particular strategies.
Our apps are designed to allow people to collect data while
they go about their daily routines. Each app is built around
a core design that uses the map and GPS functionality on an
iPhone. As people move around a city they can open a map
that shows an overlay of shops in their vicinity (Figure 1a).
People can click on shops – represented by door symbols –
and record whether the shop door is open (red) or closed
(green), Figure 1b. The initial shop database was populated
using the CTD campaign’s existing list of shops. If a shop
was not in the existing database a user could add it
manually. A traffic light system was employed to signal a
shop’s status: a consistently opened door would appear as
red, an occasionally open door as yellow, and a consistently
closed door as green (Figure 1a). Participants also had the
option to report any problems, such as duplications,
incorrect shop names, or to record if a store was
permanently closed, allowing the database to be kept up to
date. The statuses of shops were updated in real time to
reflect participants’ ongoing ratings.

Across all our apps, two approaches were used to promote
data quality and help to avoid cheating. Firstly, a system of
independent verification by a second user was used to
validate new shops. Thus, new shops only appeared on our
maps when validated by two independent users. Secondly,
participants were only able to add new shops or to record
doors within 200 yards of their present location and could
only rate a shop once each day.
Our first app acted as a Control. It used the full
functionality described above. Participants using this app
were asked to record as many shop doors as possible and
could also add new shops to the database.
The second app – the Virtual reward app – again used the
full functionality described above, but also incorporated
pointification techniques. Participants earned points and
badges for their contributions. They received 2 points for
rating a shop already in the database and 15 points for
adding a new shop (but only after it was independently
verified by a second participant). Participants could also
earn up to 15 badges (gold stars) worth 15 points each.
Badges were earned for activities such as using the app for
five days in a row or for rating a shop on the weekend.
Participants could keep track of their point score and
badges and get information on how to earn new badges
through a status screen (Figure 1c). The Virtual app also
included a leaderboard. This showed people their position
and point score relative to other Virtual app users. The
Virtual app therefore augmented the Control app with
extrinsic rewards (points and badges) and a source of
extrinsic motivation for the participants: the desire to
achieve a high score and a high position on the leaderboard.
The leaderboard listed usernames rather than real names.
The Financial app was an adapted version of the Virtual
one. It also had a leaderboard, but in this case participants
received a financial reward based on their points score and
position on the leaderboard (see experimental procedure for
reward values). The leaderboard showed a real-time tally of
the amount of money each participant would earn based on
this running point total (Figure 1d). Participants also earned
money for each badge achieved. The Financial app
therefore augmented the Virtual app with another source of
extrinsic reward and motivation: a payment related to their
relative performance in collecting data.
Participants
Participants were recruited through a mixture of sources: a
university bulletin board, an online classified service, and
local environmental sustainability groups. Participants were
told they could earn gift vouchers for taking part in a study
that would involve using an iPhone application and was
undertaken in collaboration with the CTD environmental
campaign. Overall we recruited 48 participants: 60%
female, aged 17-59, with a mean age of 27 (SD = 10.4).
Experimental procedure
Our participants were randomly assigned to three groups –
Control, Virtual and Financial – with 16 participants per
group. In all cases, participants were asked to record data
on as many shop doors as possible while going about their
everyday routines for a period of 2 weeks. The Control
group were advised that they would receive a £50 voucher
in return for participation. The Virtual group were also
advised that they would receive a £50 voucher for
participating in the study, but in addition that virtual points,
badges and a leaderboard would be available to track their
individual performance. The Financial group were told that
what they earned (in vouchers) would depend on how much
they used the app relative to other participants. They were
told they would receive £2 per badge earned, plus a share of
a fixed pot based on their relative performance on the
leaderboard. The size of the pot was set so that the total
amount available to the participants was equivalent to £50
per head, i.e. equivalent to the total payment to other
groups, but shared based on badges earned and relative
performance on the leaderboard.
All the participants completed a preliminary online
questionnaire to assess their attitude to environmental
sustainability. Questions were adapted from Diekmann and
Preisendörfer’s study regarding environmental behavior
discrepancies [12], and covered both attitudes and
(a)
(b)
(d)
(c)
Figure 1. Screenshots from left to right: (a) the map view shop doors, (b) the submission screen, (c) badges, collected and
available, (d) the financial leaderboard with points and financial tally.

Figure 2. A map showing some areas of the city mapped by
participants.
Shops
Added
Total
Points
Total
Badges
Control
221
3158
92
Virtual
274
4492
90
Financial
618
13112
165
Table 1. The total doors recorded, new shops added, points
and badges earned by each group.1
Figure 3. Points scored by the 16 participants in each group.
frequency of pro-environmental behaviors, e.g. recycling
and energy/water conservation. The question took the form
of a five-point Likert scale. Past studies in environmental
psychology have shown that an individual’s general
environmental attitude is not a guaranteed predictor of
whether they will behave in a pro-environmental way [22,
35]. For that reason, we calculated an environmental
disposition score for each participant by summing their
responses to the behavior questions only. This provides a
measure of their intrinsic motivation to engage in
environmental activities. The app was then distributed
through Apple’s TestFlight software, which relied on the
participants to actively install the relevant CTD app on their
iPhone, mimicking real-life deployment. The study then ran
for two weeks. The participants’ usage of the app itself was
recorded through the software, providing a breakdown of
1 Note that the “shops added” totals include those that were not
independently verified by a second participant within the 2 weeks, and
so did not score points.
points and badges earned, and measurements such as the
number of retailers manually added by each participant and
the amount of time the app was used. The Control app kept
track of the points and badges that the participants would
have earned from their behavior to allow for comparison
between the three groups, but this information was not
revealed to participants. At the end of the two-week test
period, a follow-up survey was sent to all participants to
obtain an overview of their opinions on the app. Finally,
based on total points, two high, two mid-range, and two low
scorers from each group were selected for a semi-structured
interview. This delved into further detail about app usage
and motivational factors. Ethical permission was granted by
the University of Bristol Ethics Committee.
QUANTITATIVE RESULTS AND ANALYSIS
Overall, the three apps proved very successful in allowing
us to map and monitor shops. Over the two-week period,
participants made 6674 individual recordings and added
1113 new shops to the CTD database. As illustrated in
Figure 2, this resulted in a detailed mapping of key
shopping streets in the city center. Table 1 shows the total
points scored by participants, together with the number of
new shops added and badges earned. As can be seen, the
Financial group achieved the highest points totals, with the
Virtual group second and the Control group third.
Data quality was very high across all experimental
conditions. Using Google maps and local knowledge, we
checked a random 10% of all added shops (including
unverified ones) across the three groups. All but one was
found to be accurate. The remaining one was unverifiable,
not necessarily wrong.
Figure 3 plots the points earned by participants in each
group. The behaviors within each group do not follow a
normal distribution, so the non-parametric Kruskal-Wallis
Test was carried out to test for differences between the
groups. This showed a significant difference across both
points earned and shops added to the database: (Points:
χ2(2, n = 48) = 9.79, p = 0.007; Shops added: χ2(2, n = 48) =
21.48, p = 0.001). We then conducted post-hoc Mann-
Whitney U Tests, with Bonferroni correction, to further
assess the differences:
Control / Virtual
Points: U = 125.0, z =-0.11, p = 0.91, r = 0.016
Shops added: U = 122.5, z =-0.22, p = 0.83, r = 0.032
Control / Financial
Points: U = 49.0, z =-2.98, p = 0.003, r = 0.43
Shops added: U = 18.5, z =-4.16, p = 0.001, r = 0.60
Virtual / Financial
Points: U = 64.5, z =-2.40, p = 0.017, r = 0.35
Shops added: U = 30.0, z =-3.72, p = 0.001, r = 0.54
The points collected and shops added by participants in the
Financial group were significantly greater than those in the
Control and Virtual groups. There was no significant

Figure 4. Z-scores for the 16 participants in each group.
difference between the Control and Virtual groups. The
effect size (r) between the Financial group and the Control
and Virtual groups ranges from medium to very large.
Though there is no significant difference in performance of
participants of the Virtual and Control groups, visual
inspection of Figure 3 suggests a different distribution of
performance in these groups. The higher scorers in the
Virtual group outperformed the Control, but the lower
performers were comparable or lower than their Control
equivalent. This is reinforced by the observation that,
although the mean point score in the Virtual group (281) is
greater than that of the Control (197), the median of the
Virtual (112.5) is less than that of the Control (197). To
investigate this further we plotted the Z-scores of all
participants to test the relative comparative performance of
individuals in each group (Figure 4). Results show that that
the Virtual and Financial groups followed a roughly similar
trajectory, with the top three or four participants
outperforming the others. The Control group, with the
exception of the highest scoring participant, follows a
flatter trajectory showing a more even spread of behavior.
The highest-scoring participant from the Control group
earned a score over three standard deviations from the mean
and, per the extreme studentized deviate method, her result
can be considered an outlier (critical value of Z = 2.58, p
<0.05). Overall the Z-scores suggest that competition, both
in the Virtual and Financial groups, motivated the higher
performers, but may have demotivated some of the middle
performers. This issue was investigated further our semi-
structured interviews.
It appears that the option of collecting badges did not have
a significant effect on participants’ behavior. In fact, the
Virtual group earned two fewer overall than the Control,
despite the Control group being unaware of what they were
earning. Furthermore, it was predicted that a majority of the
Financial participants would earn the full set of 15 badges,
worth £30.00 in online vouchers. However, only 25% did
so. The interviews helped elucidate the badge discrepancy;
many participants reported being unaware of how to earn
the badges despite prior instruction: “I wasn’t really aware
of when you would get a badge. I think you could press
them to find out what you needed to get but I didn’t go in
that far, I just saw that the badge wasn’t highlighted.” As a
result, the motivating effect of deliberately earning a virtual
badge could not be adequately assessed.
Effect of Existing Environmental Tendencies
In addition to testing extrinsic motivational factors, our
study was also designed to examine whether pre-existing
environmental tendencies would function as an intrinsic
motivator and influence app usage. We therefore performed
a correlation analysis within each group comparing
participants’ environmental disposition scores, from the
initial questionnaire, and their point scores. The results
showed no significant correlation: Control r = 0.16; Virtual
r = -0.30; Financial r =0.18. This is in line with previous
studies [1]. Again, this finding is explored further in our
qualitative study.
QUALITATIVE RESULTS AND ANALYSIS
The results and analysis presented in this section are
derived from the follow-up survey completed by all
participants, and from semi-structured interviews with 18
participants, six from each group. The interviews were
audio recorded and transcribed. We then undertook a
thematic analysis [4]. The following sections address the
key themes that emerged.
The Usability of the Apps
A vast majority of the participants felt that the app was
intuitive and easy-to-use. For example: “It was easy to use,
only two taps on the screen were needed to report the status
of a shop door.” Some participants expressed frustration at
the GPS receiver not accurately locating their position. This
occasionally prevented them for rating a shop. Overall,
however, it is unlikely that basic usability issues impacted
on the results of our study.
Motivators and Enablers
In considering the design of systems to target individual
behavior change, Fogg emphasizes the importance of both
motivation and ability [15]. A similar trend emerged in our
thematic analysis. However rather than the term ability, we
use the more holistic term “enablers” to address the joint
impact of app functionality and contextual factors (e.g.
lifestyle) on app usage. Alongside motivational factors,
enablers proved critical to participants’ engagement.
Motivator: Performance-Related Financial Incentives
As presented above, the Financial group gathered a
significantly greater quantity of data/points, confirming that
an extrinsic financial incentive tied to contribution served to
motivate users far more than either the extrinsic motivation
of leaderboard position or intrinsic motivation to contribute
to an environmental activity. However, the qualitative data
suggested that this came with a cost. Participants in both the
Control and Virtual groups reported 100% agreement, when
asked in the follow-up survey, whether they would be
willing, without financial compensation, to continue using
an app like CTD to help a community organization on an
ongoing basis. This decreased to 75% of those within the
Financial group, corroborating Deci’s hypothesis [10] that

intrinsic motivation, such as helping a charity, is reduced
when performance is explicitly linked to financial rewards.
Motivator: Competition
Quantitative results suggested that the participants near the
top of the Virtual and Financial leaderboards were actively
competing for the top position, and that this competitive
aspect can provide extrinsic motivation to maintain
engagement with the app. The interviews backed this up,
with some high scoring participants reporting feeling
encouraged by competition, and indicating they were
willing to use the app more than they would have
otherwise: “The competition was an extra element that I
wasn’t expecting with the app … I would check it [the
leaderboard] even if I knew I wasn’t going to go out that
day to see if people were coming up near me, and if they
were then I’d be like, ‘Right I’ve got to get on with it’.”
Conversely, distant competition, for participants who were
hundreds or thousands of points behind, had the opposite
effect and served as a very clear demotivator. One low-
scoring participant described this, stating: “I gave up as I
felt as though there was never a chance to catch the
leaders. When I saw that someone had stacked up hundreds
of pounds all my motivation dissipated. Games rely on
positive feedback for the player to want to continue. This
de-motivated me massively so for me the game failed.”
Several of the participants interviewed adopted “self-
gamification” strategies, in which they set goals or
challenged themselves to record more than previously. Two
high-scoring participants in particular exhibiting a tendency
to self-gamify: “I think being kind of competitive with
myself, like ‘Today I’ll do a few more than I did yesterday,
maybe I’ll walk up that street because I’ve not been up
there before so I don’t know what’s up there’.” This trend
has also been found in previous research on eco-feedback
technologies, such as smart meters, which are not explicitly
gamified [37]. Whilst this issue was not specifically
addressed by the app designs used in this study, it would be
interesting to investigate how designs that encourage self-
gamification can be used to motivate participants who find
head-to-head competition to be demotivational.
Motivator: Community Norms
As well as a source of potential competition, leaderboards
can also provide feedback regarding the behavior of the
community, and may therefore result in “norm activation”
in participants, another source of extrinsic motivation. We
have some evidence of this occurring. For example, activity
on the leaderboard acted as a motivational trigger for a
participant who initially delayed using the app: “I think it
[seeing people with points] probably did make me go, ‘Oh,
okay, people have got started I should probably make sure
that I make an effort to do it when I go out’. Yes, that
probably did give me a bit of a kick.”
Despite the anonymous leaderboard, some participants also
expressed a desire for what we can call “acceptable
mediocrity”, i.e. not a desire to be the best through
competition, but to show that they had put some effort into
using the app. For example, a participant felt the
leaderboard “was an indication obviously on how much
other people were using it and I wanted to make sure I was
sort of in the middle or top half rather than the bottom
end.” Overall, the simple pressure or desire to be seen as an
active member of the community, by having an acceptable
presence on the leaderboard, regardless of score, appeared
to affect app usage: “It made me feel bad because I realized
that people had used it more than I had … I felt guilty that I
hadn’t started using it so then I used it.” It is interesting to
observe that this occurred even though the leaderboard in
our study was anonymous and participants were not known
to each other.
Motivator: Environmental Interest
As shown in quantitative results section, environmental
tendencies were not a guarantee of engagement or app use.
The qualitative responses supported this, showing that both
high and low-scoring participants were equally likely to
describe pro-environmental tendencies (e.g. another low-
scoring participant: “I notice doors being open and closed,
and I make sure when I leave a shop I try and close it,
although some shops are like, ‘No, you can’t close the
door.’”) Furthermore, one participant informed us they had
previously volunteered for the CTD campaign, yet was a
low-scoring member of the Virtual group. However, the
topic of the app cannot be dismissed in its entirety as it can
serve as a threshold motivator, encouraging people to
engage initially. Regardless of their final score, the idea that
an easy-to-use mobile app allows a participant to make a
small contribution of time and effort to support an
environmental cause was cited by several participants as a
positive element, and given as a reason for signing up in the
first place. e.g.: “Definitely the whole environment thing,
that definitely motivated me because I like to think of myself
as the kind of person, I might not be destined for greatness
but I can have a part to play, a little role that helps do
something great, helps to make a change, and all I have to
do is use an iPhone app.”
Enabler: Lifestyle
Interviews suggest that participants’ lifestyles played a
significant role in this use of the app. For some, their
lifestyles acted as an enabler to active participation, while
others’ lifestyles reduced their opportunities for data
collection. A common factor shared by the highest scoring
participants from each group was that they were not in
regular employment during the testing period, and therefore
could use the CTD app as and when they saw fit. For
example, the high-performing outlier in the Control group
was on maternity leave: “… when I’m out with her she
tends to sleep when I’m walking and it gives me something
to do so I walked out of my way a lot of the time just to
make sure I would catch as many shops as I could on the
way.” In contrast, one low-scoring participant reported
using the app only when walking to or from work, or while
on breaks: “Well, most of the time I use it because I work

on a retail estate so I used it when I was on my break, up
and down because all the shops are within meters of each
other.” Another expressed enthusiasm about the CTD app
in terms of its purpose to reduce energy waste, but
admitted: “I didn’t have that many opportunities … to go
out to different places to have a look … to go out to the
different shops to try it”.
Overall these responses reveal that while people may be
intrinsically motivated in terms of supporting a cause, at the
same time they can be “disabled” by the lack of
opportunities to actually use the app. Given this,
organizations needing data collection on a compressed
timescale may wish to seek out participants who are time
rich, even if they do not seem to be part of a target
motivational group.
Enabler: Technology
The use of smartphones to collect data cannot be
overlooked as a possible threshold motivator to initially
attract participants. Several participants mentioned that our
recruitment advertisement specifically asking for iPhone
users attracted their attention: “My partner works in IT
anyway so I’ve got a soft spot for new bits and pieces and
obviously I’m a bit of a MAC fan so it went together.”
As previously mentioned, technology did serve as a disabler
on occasions where the GPS receiver was not accurate, and
prevented participants from recording data. Some
participants also expressed concerns that the app used their
mobile data connection and might cause them to exceed
their monthly data allowance. Allowing the data to be
stored and uploaded at a later time would avoid this
problem, but it could also affect the pointified elements if
the real-time nature of the leaderboard was influenced.
Further research is needed to determine whether this delay
would have an impact on app engagement.
Enabler: Weather
Wasting heating energy by leaving doors open in cold
weather is the very raison d’être of the CTD campaign.
However, the changing weather itself came up in the
interviews as affecting a participant’s ability to conduct the
task. Good weather acted as an enabler, encouraging people
to spend more time outside and around the city. However,
there is also evidence that some participants avoided using
the app if they felt a shop was justified in leaving a door
open on an unusually warm day during the trial.
Cold and wet weather acted as a disabler. This was due to
one of four reasons: (1) the participant decided not to go out
because of the weather; (2) the participant used a car or
public transport instead of walking, precluding use of the
app; (3) the participant did not want to get the phone wet;
and (4) participants went out but were in a hurry due to the
weather, so did not use the app. This reliance on weather
must be taken into account if an environmental data
collection app is to be successful. For example, a mobile
app designed to record blocked drainage in the streets
during severe wet weather is unlikely to be used frequently
enough to be useful, as the necessity of recording in the rain
is unappealing to participants.
Behavior and Awareness
The CTD app was developed to investigate how a mobile
app could be used to scale up the activities of a pro-
environmental community activist group. It did not seek to
alter behavior or raise the participants’ awareness of an
environmental issue. Nonetheless, there is evidence in the
interviews that it did have some affect. One-third of the 18
interviewees reported reading the CTD website as a direct
result of taking part in the study. These were spread among
groups, but tended to be those who had a stronger
environmental disposition in the initial questionnaire.
Furthermore, two participants claimed to have even
changed their shopping habits to favor shops with doors
closed: “Using the app made me much more aware of my
surroundings as I walk around the city. It has made me
consider the environmental impact of choices I make when
shopping. I also find I am more inclined to visit a business
whose door is closed!”
Whether this is a lasting, or even genuine, change is
unknown. However it does highlight another potential use
of such data collection apps: as a way of drawing attention
to the issue in a wider population and attracting more
permanent volunteers to the organization. Those who enjoy
the app or have an interest in the subject may be willing to
get further involved and take on additional responsibilities.
RECOMMENDATIONS
Based on the quantitative and qualitative results presented
in the previous sections, we can make recommendations to
researchers and organizations, such as CTD, wishing to
exploit the potential of digital data collection.
1. Seek those whose lifestyle is likely to enable them to
participate.
2. Use passion for a cause as a threshold motivator, but do
not assume it acts as an engagement motivator.
The people in our study who were mobile but not “goal
directed” in their mobility made the greatest contributions –
the parent walking with a pram being our archetypal
example. Hence a good recruitment strategy for CTD would
be to target online discussion boards for new parents, who
are likely to have a lifestyle that would allow ongoing
contribution, and use an environmental message as a
threshold motivator to attract those with compatible
intrinsic motivation to volunteer. This is likely to be more
effective than recruitment through an environmental forum,
which may initially yield many enthusiastic volunteers who
are motivated to cross the threshold to participation but then
do little or no data recording due to lifestyles incompatible
with ongoing contribution.
3. Make competition available, but easy to ignore.
It is clear that close competition among leaders is
productive, but also clear that it demotivates those not in
the leading group. Applications such as foursquare provide

multiple and customizable leaderboards, allowing a
participant to choose who they compete with. Some have
weekly resets, ensuring that an existing player cannot build
up a long-term, unassailable lead. Additionally, we could
envisage the use of ‘adaptive leaderboards’ to provide
competition when it is likely to be motivating. This would
consist of a set of leaderboards (e.g. day, week, etc.) by
default in the background, a given board only moving to the
foreground for a participant able to engage in close
competition on it. Alternative approaches to doing this, and
assessments of effectiveness, are areas for further research.
4. Provide information regarding ‘community norms’ in a
way which motivates desired behavior.
A number of interviewees were motivated by “doing their
bit” as opposed to being top. Based on this, it is perhaps as
important to “normify” an app as it is to “gamify” it. For
example, providing the average amount of data collected by
participants may encourage some to try to meet this figure,
even if they cannot compete with the leader. Recent
research in the area of domestic energy reduction has also
provided evidence to support this approach [30, 33].
5. Use financial motivation carefully.
It is clear from the quantitative data that a “pay for results”
approach is a powerful motivator and resulted in more data
being collected. However our interviews also suggested that
it decreased participants’ willingness to use the app in the
future, with only 75% of Financial group participants
expressing willingness in comparison with 100% of other
participants. Hence this short-term gain may be offset by a
longer-term loss of engagement. For that reason, it may be a
better use of financial resources to adopt an approach
combining a small payment as a threshold motivator [8] and
funding prizes and rewards linked to both gamified and
normified achievements. Exploring the size, quantity and
structure of alternative reward mechanisms and their impact
on behavior is a fruitful area for future research.
CONCLUSIONS AND FUTURE WORK
In this paper, through quantitative and qualitative
approaches, we have assessed the impact of different
motivating factors and design strategies on the performance
of subjects collecting data for the CTD pro-environmental
campaign. In our trials a pay-per-results financial reward
mechanism resulted in significantly increased data
collection, though there is also evidence to suggest it may
reduce long-term intrinsic motivation to participate in such
activities. We also found that pointification increased
performance, though not to a statistically significant level.
Our analysis suggested that this was because close
competition acted as an effective motivator of those who
were high ranking, resulting in their increased performance
relative to the Control group. However, this effect was to
some extent offset by reduced performance of lower
ranking participants because of the demotivating effect of
distant competition. We also identified that a leaderboard
can have the effect of inducing social norms to encourage
performance. Finally, we showed that intrinsic motivation
to carry out environmental actions was not correlated with
performance, identifying qualitative evidence that an
appropriate lifestyle had a bigger impact. Based on our
findings, we made recommendations for organizations
wishing to design and use mobile apps to support
community activism. Note that, though our trial participants
were paid, all but the last of these recommendations apply
equally to organizations recruiting unpaid volunteers.
As this paper is amongst the first to use both quantitative
and qualitative techniques to explore motivators and
enablers for pro-environmental data collection, there remain
many topics for further research in addition to those
addressed here. Firstly, our trial was not long enough to
explore how participants’ behavior changes over time, so
valuable work could be conducted to examine motivation
over a longer time period and what can be done to support
continued engagement. Secondly, our participants were a
priori unknown to each other and remained anonymous on
our leaderboards. As a result, we have not explored the
motivating effect of competing with one’s own social
group, something that foursquare provides as an option. Nor
have we explored the effect of allowing teams of
participants to compete together – something that the FoldIt
citizen science game [20] has used to good effect. Also, we
have not investigated the effects of gamification approaches
beyond simple pointification. In particular, we believe the
intrinsic pleasure of carrying out an enjoyable activity for
its own sake is an important motivator, which our work has
(consciously) not explored, due to our focus on simplicity
in the app design. Our qualitative results suggest that “self
gamification” – allowing participants to set themselves
challenges and compete against their prior performance –
may act as a strong motivator for some participants. Each of
these areas warrants further exploration.
Finally, and perhaps most importantly, research is needed
on effective design strategies to “close the loop” between
casual volunteers and the “targets” of the community
activist campaigns – in our case shop owners. This could
have several positive effects. Firstly, by providing feedback
to the participants on the ongoing effectiveness of the CTD
campaign, such as newly signed-up shops or a measure of
ongoing carbon emissions saved, intrinsic motivation may
be increased as volunteers may gain a greater sense of
contributing to a project with real impact. Secondly, by
providing data on local shops that keep doors shut, users
could search for shops that support the CTD campaign and
place pressure on retailers to change their behavior. Finally,
by providing data to the retail community, such as maps of
local areas highlighting those shops who do maintain a
doors-closed policy, retailers can see the spread of a
community norm which is both environmentally and
economically beneficial, potentially resulting in an
increased speed of uptake of the norm. Through ongoing
research on these issues, digital technology may further
contribute to the spread of pro-environmental behavior.

ACKNOWLEDGEMENTS
We would like to thank Sue Pollard and Phil Holtam from
the Bristol-branch of the Close the Door campaign, Darren
Hall from Bristol Green Capital, and the participants of the
study. This work was part funded by UK Research Council
Digital Economy grant number EP/1000151/1.
REFERENCES
1. Bamberg, B. How does environmental concern influence
specific environmentally related behaviors? J. Environmental
Psychology 23 (2003), 21-32.
2. Bang, M., Gustafsson, A. & Katzeff, C. Promoting new
patterns in household energy consumption with pervasive
learning games. Persuasive 2007, 55-63.
3. Basarir, M. & Overend, M. Interim Report on the Energy
Appraisal of Retail Units: Assessing the effect of open doors
on energy consumption and thermal comfort. University of
Cambridge (2010). CUED/D-STRUCT/TR232.
4. Braun, V. & Clarke, V. Using thematic analysis in
psychology. Qual Research in Psych 3, 2 (2006), 77-101.
5. California Roadkill Observation System, Annual Report
(2011): wildlifecrossing.net/california. Road Ecology Center
& Information Center for the Environment at UC Davis.
6. Cohn, J.P. Citizen Science: Can Volunteers Do Real
Research? BioScience 58, 3 (2008), 192-97.
7. Cooper, S., Khatib, F., Treuille, A., Barbero, J., Lee, J.,
Beenen, M., Leaver-Fay, A., Baker, D. & Popović, Z.
Predicting protein structures with a multiplayer online game.
Nature 466 (2010), 756-760.
8. Crowston, K. & Prestopnik, N.R. Motivation and data quality
in a citizen science game: A design science evaluation.
International Conference on System Sciences 2013, in print.
9. De Young, R. Changing behavior and making it stick: the
conceptualization and management of conservation behaviour.
Environment and Behavior 25, 3 (1993), 485-505.
10. Deci, E.L. Effects of externally mediated rewards on intrinsic
motivation. Journal of Personality and Social Psychology 18,
1 (1971), 105-115.
11. Deterding, S., Sicart, M., Nacke, L., O'Hara, K. & Dixon, D.
Gamification. using game-design elements in non-gaming
contexts. ACM CHI EA 2011, 2425-2428.
12. Diekmann, A. & Preisendörfer, P. Environmental Behavior:
Discrepancies between aspirations and reality. Rationality and
Society 10 (1998), 79-102.
13. DiSalvo, C., Sengers, P. & Brynjarsdóttir, H. Mapping the
landscape of sustainable HCI. ACM CHI 2010, 1975-1984.
14. Doan, A., Ramakrishnan, R. & Halevy, A.Y. Crowdsourcing
systems on the World-Wide Web. Communications of the
ACM 54, 4 (2011), 86-96.
15. Fogg, B.J. A Behavior Model for Persuasive Design.
Persuasive 2009.
16. Froehlich, J., Dillahunt, T., Klasnja, P., Mankoff, J., Consolvo,
S., Harrison, B. & Landay, J.A. UbiGreen: investigating a
mobile tool for tracking and supporting green transportation
habits. ACM CHI 2009, 1043-1052.
17. Froehlich, J., Findlater, L. & Landay, J. The design of eco-
feedback technology. ACM CHI 2010, 1999-2008.
18. Graham, E.A., Henderson, S. & Schloss, A. Using Mobile
Phones to Engage Citizen Scientists in Research. Eos Trans.
AGU 92, 38 (2011), 313-15.
19. Gustafsson, A., Bang, M. & Svahn, M. Power explorer: a
casual game style for encouraging long term behavior change
among teenagers. ACE 2009, 182-189.
20. Hand, E. Volunteer army catches interstellar dust grains.
Nature 466 (2010), 685-687.
21. He, H.A., Greenberg, S. & Huang, E.M. One size does not fit
all: applying the transtheoretical model to energy feedback
technology design. ACM CHI 2010, 927-936.
22. Hines, J.M., Hungerford, H.R. & Tomera, A.N. Analysis and
synthesis of research on environmental behavior. A meta-
analysis. Journal of Environmental Education 18 (1987), 1-8.
23. Kim, S., Robson, C., Zimmerman, T., Pierce, J. & Haber, E.M.
Creek watch: pairing usefulness and usability for successful
citizen science. ACM CHI 2011, 2125-2134.
24. King, S.F. & Brown, P. Fix my street or else: using the
internet to voice local public service concerns. Conference on
Theory and Practice of Electronic Governance 2007, 72-80.
25. Kuznetsov, S., Davis, G., Cheung, J. & Paulos, E. Ceci n'est
pas une pipe bombe: authoring urban landscapes with air
quality sensors. ACM CHI 2011, 2375-2384.
26. Maisonneuve, N., Stevens, M., Neilssen, M., Hanappe, P. &
Steels, L. Citizen noise pollution monitoring. International
Conference on Digital Government Research 2009, 96-103.
27. Malone, T. Heuristics for designing enjoyable user interfaces:
Lessons from computer games. ACM CHI 1982, 63-68.
28. Mankoff, J., Matthews, D., Fussell, R. & Johnson, M.
Leveraging social networks to motivate individuals to reduce
their ecological footprints. HICSS 2007, 87-96.
29. Mason, W. & Watts, D. Financial incentives and the
“performance of crowds”. SIGKDD Explorations 11, 2 (2008),
100-108.
30. Nolan, J., Schultz, P., Cialdini, R., Goldstein, N. &
Griskevicius, V. Normative social influence is underdetected.
Personality and Social Psych Bulletin 34 (2008), 913-23.
31. Paulos, E., Honicky, R.J. & Hooker, B. Citizen Science:
Enabling Participatory Urbanism. In Handbook of Research on
Urban Informatics: The Practice and Promise of the Real-time
City, M. Foth, editor (2009).
32. Rotman, D., Preece, J., Hammock, J., Procita, K., Hansen, D.,
Parr, C., Lewis, D. & Jacobs, D. Dynamic changes in
motivation in collaborative citizen-science projects. ACM
CSCW 2012, 217-226.
33. Studley, M., Chambers, S., Rettie, R. & Burchell, K.
Gathering and presenting social feedback to change domestic
electricity consumption. Workshop on Persuasion, Influence,
Nudge & Coercion Through Mobile Devices 2011.
34. Sustainable Consumption Roundtable, “I will if you will”:
towards sustainable consumption. 2006, Sustainable
Development Commission & National Consumer Council.
35. Wiegel, R.H. Environmental attitudes and the prediction of a
behavior, in Environmental Psychology, N.R. Feimer and E.S.
Geller, editors (1983), 257-287.
36. Wiggins, A. & Crowston, K. From Conservation to
Crowdsourcing: A Typology of Citizen Science. International
Conference on System Sciences 2011, 1-10.
37. Woodruff, A., Hasbrouck, J. & Augustin, S. A bright green
perspective on sustainable choices. ACM CHI 2008, 313-22.
38. Zichermann, G. & Cunningham, C. Gamification by Design:
Implementing Game Mechanics in Web and Mobile Apps.
Sebastopol, CA: O’Reilly Media (2011).