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�+.)$9&D81..1$:D"00)$*+55&.&5$-"50+/"5$+#($Q+#$ 7+/.1K6$ $ !"#$%T&%'%% H4$".3-#%UH115V%G1835$%($?)-151#3$:%(1%J$/+?$%($$-%P-$4#6%N:$% A+#&1.$O&00"#)$!+#10$31+()$3E551..$%1+.1)$%1#$M"N+#$+#($2E/+#4$=E"6$ % !"#$%TB%'%% =+/#3-#%!$125$%".%M14E%"-/%>.)$4%()34/'!"4.6%W1?".31-:R% 9+R$I6$S&.5"#)$A1'1/$O"501')$:8+E#$I+N5"#$+#($:&F"#$T((&5"#% % The Augmented Shopping Trolley: An Ambient Display To Provide Shoppers with Non-Obvious Product Information Jon Bird, Vaiva Kalnikaité and Yvonne Rogers Pervasive Interaction Lab The Open University Milton Keynes, MK7 6AA, UK {j.bird, y.rogers}@open.ac.uk, vaivak@gmail.com ABSTRACT information about the global consequences of their The Augmented Shopping Trolley consists of an ambient consumer decisions [2]. Our goal is to provide ‘non- handlebar display connected to a scanner. When a shopper obvious’ nutritional, ethical and environmental product scans an item the handlebar lights up to provide them with information, that is, information that is not immediately information about the product, such as its nutritional, obvious from an item’s packaging or label, in a form that is ethical or environmental attributes, that are not obvious as salient as the features that typically inform consumers’ from its packaging or label. The system is designed to decision making. The Augmented Shopping Trolley (Figure seamlessly integrate with a shopping experience: it uses 1) is designed so that it fits as seamlessly as possible into a familiar supermarket technologies; it keeps both of a supermarket shopping experience. We use familiar shopper’s hands free; and the simple ambient display supermarket technologies: augmenting a standard shopping facilitates the ‘fast and frugal’ decision-making typically trolley by attaching a scanner and embedding an ambient observed in a supermarket. Our initial lab-based study display in the handlebar. This gives our system two shows that the display can be understood at a glance and advantages over using mobile devices to provide product used to select items based on a product’s nominal properties information. First, the trolley scanning technology is faster (for example, it is organic), ordinal properties (for example, [4] and second, because the ambient display is built into the it has low, medium or high food miles), as well as a trolley handlebar a customer’s shopping experience is not combination of the two at the same time. Where as usability disrupted by having to repeatedly access and store a mobile was the focus of our initial design, ethical issues have come display. Underhill [10, see chapter 4] emphasizes the to the fore as we develop the system for use in importance of having both hands free during shopping. supermarkets and we discuss how these are influencing our design. Author Keywords Persuasive technologies, ambient display, shopping, product information, ethics. ACM Classification Keywords H5.m. Information interfaces and presentation (e.g., HCI): Miscellaneous. INTRODUCTION In a supermarket, shoppers tend to make snap judgments based on just a few salient cues (low price, recognized brand and attractive packaging) and they rarely take time to read product information labels [7]. However, recent Figure 1. The Augmented Shopping Trolley display consists of consumer surveys indicate that shoppers want more 16 LEDs embedded in the handlebar, each of which can be set to green, red or orange Permission to make digital or hard copies of all or part of this work for Our approach to designing an effective ambient display, personal or classroom use is granted without fee provided that copies are first outlined in [9], is motivated by studies of ecological 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, rationality which investigate how people make reasonable or republish, to post on servers or to redistribute to lists, requires prior decisions given the constraints of limited time, information specific permission and/or a fee. and computational resources that characterize most real Copyright © 2011 for the individual papers by the papers' authors. world situations [6, 8]. This research indicates that most Copying permitted only for private and academic purposes. This volume is published and copyrighted by the editors of PINC2011.. natural decision making is made on the basis of ‘fast and (idle state) to a half second sweeping movement of orange frugal’ heuristics – short-cut strategies where people ignore that indicates scanning is in progress. There is then a beep, most of the available data and instead focus on the most as typically heard at a checkout counter, to signal that useful information and process it quickly. Often people scanning is completed and the display then changes to a make a decision based on a single reason as this strategy is new state that provides relevant information about the quick and simple and avoids having to weigh up trade-offs product. If the display is configured to show a nominal between multiple and potentially conflicting options. This property of the product, then it flashes green if the property approach is not rational in certain environments, namely, is present and shows the idle state if it is not. If the display those where available pieces of information are is providing ordinal information about the product, the approximately equally useful. However, in a shopping display employs a bar graph metaphor, with the number of environment, the distribution of information usefulness is red pixels indicating the degree to which an item has a highly skewed, that is, the most useful piece of information property. Specifically, if an item has a low degree of a is a lot more important than the second most useful, which property then pixels 1-3 turn red and 4-16 turn green; if in turn is considerably more important than the third, etc. medium then pixels 1-8 turn red and pixels 9-16 turn green; Our handlebar ambient display consists of just sixteen and pixels 1–13 turn red and 14-16 turn green if the item LEDs. When a shopper scans a product, a few pieces of has a high degree of a particular property. Finally, both non-obvious information, such as whether it contains nuts, these representations can be combined to show the value of is fair trade or has low food miles, are displayed as a salient a nominal and an ordinal property at the same time. In our pattern on the display. study, after a participant selected or discarded an item, the display changed back to the all green idle state. Given that information salience influences a person’s behaviour unconsciously [1], rather than through rational LAB-BASED SYSTEM EVALUATION reflection, this raises ethical concerns about the Augmented 5 adults (1 female, 4 male, aged between 20 and 40) took Shopping Trolley, chief of which is that this system could part in a lab-based evaluation of the Augmented Shopping potentially manipulate people into behaving in ways that Trolley. Each participant completed 12 shopping scenarios they would not otherwise do, and furthermore, that they where they were asked to pick up and scan 5 items of a might not be aware that they had been manipulated. This particular product type and only select those items that met concern, and also issues to do with privacy and clarifying specified criteria. A scanner was attached to the shopping how our system benefits shoppers, form the ethical trolley (Figure 1) but was non-functional and the handlebar considerations that are influencing how we deploy the display was changed using a Wizard of Oz methodology. Augmented Shopping Trolley in a supermarket. On the basis of the changes in the patterns on the handlebar The paper is structured as follows: first, we describe the display, participants had to decide whether to select the display hardware and how it conveys product information; item and place it in their trolley or discard it and place it on second, we describe a lab-based evaluation of the system an adjacent table. Since this was an exploratory study, we that demonstrates the efficacy of the ambient handlebar were intentionally vague about the operation of the ambient display for conveying non-obvious product information; display as we wanted to see whether participants could and third, we describe the ethical issues that are informing understand it intuitively. We only told participants that the the development of the system for use in supermarkets. display patterns would change depending on whether a product had a specific property (yes/no), the degree to AMBIENT HANDLEBAR DISPLAY DESIGN which a product had some property (high/medium/low) or a The handlebar display was designed to provide shoppers combination of the two. Participants were allowed to scan with salient and easy to read information about a scanned the items as many times as they wanted and in any order, product’s nominal properties (for example, whether it is before they made their decision about whether to select a organic or contains nuts), its ordinal properties (for particular item. We used 4 product types: milk; breakfast example, if it has low, medium or high food miles), as well cereal; wine; and juice. Each shopping scenario used one of as a combination of the two at the same time. We the product types and participants were asked to select from constructed the display by attaching 16 bicolour LED units 5 different items. For example, select those bottles of wines to a piece of wood inside a transparent plastic tube (Figure that meet the specified criterion (fair trade) and put them in 1). This replaced the plastic handlebar in a standard the trolley, and place the others on the discarded items shopping trolley. The LEDs are controlled using 2 table. Each of the items was a real product but we masked TLC5940 chips (Texas Instruments) that are driven by an any product information on the packaging and told Arduino microcontroller. In our lab-based study this is participants to only use the handlebar display to decide attached via a USB cable to a laptop running a Processing whether they should select an item or not. The experimenter application. Each LED unit can be set to red, green or playing the Wizard of Oz role sat at a table on which the 20 orange (when both the green and red LEDs are on). Each shopping items were grouped by product type. Each item time a product is scanned, the display changes in the was individually numbered so that the experimenter could following way. First, it goes from an all green background change the display appropriately when the participants use and were able to quickly read it even though they were scanned a particular item. not given explicit information on the meaning of the display patterns. Only two participants scanned items more than In the first 4 shopping scenarios the handlebar display once and this was exploratory activity at the beginning of indicated whether a scanned item had a particular nominal the evaluation when they were seeing how the interface property or not: whether a milk product was organic; worked. whether a breakfast cereal contained nuts; whether a bottle of wine was fair trade; and whether a carton of juice ETHICAL ISSUES AND FURTHER DEVELOPMENT contained added sugar. In 2 of these scenarios the Whereas usability issues informed our initial design, ethical participants had to select items that had a particular considerations are shaping the development of the property and in the other half they had to discard items if Augmented Shopping Trolley for use in supermarkets. This they had a particular property. For example, in the first is because our ambient display not only provides salient shopping scenario participants had to select a milk product product information for shoppers, but also potentially if it was organic and discard it if it was non-organic; in the influences what they purchase. The use of persuasive second shopping scenario participants had to select a technologies raises ethical concerns for many people. For breakfast cereal if it did not contain nuts and discard it if it example, Page and Kray [3] used an online questionnaire to did. investigate people’s views on the ethics of using persuasive In the next stage of the evaluation, the participants technologies to encourage healthy living. 72 participants completed 4 shopping scenarios where the display indicated rated the ethical acceptability of a number of different whether a product contained a low, medium or high value scenarios which varied in 3 different factors: whether a of a particular ordinal property. The task was to select items participant chose to use the technology or an external that had a specified property to a particular degree. agency initiated its use; whether there was a clear benefit Specifically, participants were asked to select milk with a for the participant or not; and the technology used (text medium fat content, cereals with a high sugar content, wine messages to the participant’s mobile phone; public with low food miles and juice with a medium water content. announcements in the participant’s location; Facebook In none of these scenarios were participants asked to messages; restrictions on the participant’s bank account; discard items if they had properties of a particular degree. and electric shocks). The results indicated that the majority The final 4 shopping scenarios tested whether participants of the participants viewed the use of persuasive could understand the display when it simultaneously technologies in most of the questionnaire scenarios as showed information about both a nominal and an ordinal unethical. When there was no clear benefit to the property of a scanned item. Participants were asked to participant, mobile phone were considered the most ethical select milk that was organic and low fat, cereals that persuasive technology. However, approximately the same contained nuts and had a medium sugar content, juice that proportion of participants (40%) considered them very had added sugar and high water content and wine that was ethical or ethical as the proportion that considered very not fair trade and had medium food miles. Only in the wine unethical or unethical when. A large majority of scenario did participants have to reject items on the basis of participants found the other technologies very unethical or information about a nominal property of the product. unethical. In scenarios where the use of a technology would clearly benefit the participant, for example, save their life, USABILITY RESULTS then this usage was considered slightly more ethical than 4 out of the 5 participants were able to interpret the ambient the cases where the technology did not benefit the handlebar display and complete all the tasks without any participant. However, it is not clear whether these mistakes. The other participant made one consistent error in differences were statistically significant. When people were 2 of the first shopping scenarios where the task was to able to freely choose whether to use a persuasive discard items if they had a particular nominal property: they technology or not, then texts, public announcements and selected, rather than discarded, them, but did not repeat this Facebook messages were considered ethical by the majority error in the final shopping scenario which also required an of respondents, in comparison to the situation where the use item to be discarded if it had a particular nominal property. of the persuasive technology was initiated by an external Several participants reported that they found the tasks entity (for example, the UK’s National Health Service). where they had to discard items with particular properties Electric shocks and bank account restrictions were more difficult and it did seem to increase the cognitive load considered very unethical or unethical by the majority of in all participants, resulting in a slightly slower response respondents, even when a participant chose to use them. time (approximately 2 seconds, rather than 1 second for the Page and Kray’s findings seem to concur with a central other conditions). This could be due to the colours used in factor identified by applied philosophical analyses of the display: a nominal property is indicated by a green ethical behaviour, for example, the use of persuasion in blinking display, a colour that many people associate with advertising [5]. Namely, the ethics of an action are positive properties, rather than ones that should be avoided. determined, to a large degree, by the extent to which that All participants reported that the display was intuitive to action impacts on an individual’s autonomy, that is, their capacity to choose how to act and determine their own life. nominal and ordinal properties of a scanned product. Our Page and Kray’s research also highlights that privacy and display is intuitive to use and requires no training. the extent to which a participant benefits are important Participants find it easier to select items when they have issues for determining the ethical acceptability of desirable properties than to not select them because they persuasive technologies. All three of these ethical have undesirable properties. The Augmented Shopping considerations (autonomy, privacy and benefits) are Trolley makes non-obvious nutritional, ethical and shaping the development of the Augmented Shopping environmental product information salient to shoppers and Trolley. facilitates the fast and frugal decision making typically used in a supermarket. Some of the global consequences of To ensure shopper’s autonomy, they will be free to decide selecting particular products can now be made salient to whether they use the Augmented Shopping Trolley and also shoppers at the point of decision making, potentially able to choose which particular non-obvious product facilitating changes in consumer behaviour. We argue that information they want to be informed about. Given that our system is an ethical persuasive technology as it users can configure the system to provide different product enhances the ability of shoppers to buy choose products in information, privacy is not compromised, even though the accordance with their individual values. handlebar will be visible to other shoppers, as they will not understand what particular LED patterns mean. Some of the REFERENCES product information that will be provided by the 1. Cabinet Office and Institute for Government (2010) Augmented Shopping Trolley can clearly benefit a MINDSPACE. Influencing Behaviour through Public participant, for example, nutritional data, whereas other Policy. London: Cabinet Office. information, such as food miles, may not have direct http://www.instituteforgovernment.org.uk/content/133/ personal benefits. In fact, trying to minimize food miles mindspace-influencing-behaviour-through-public-policy may lead, literally, to a personal cost. However, we assume 2. EDS IDG Shopping Report 2007: Shopping Choices: that if participants choose to be informed about a particular Attraction or Distraction? type of product information then they do so because it is of http://www.eds.com/industries/cir/downloads/EDSIDG Report_aw_final.pdf benefit to them and in keeping with their lifestyle choices. 3. Page, R. E. and Kray, C. Ethics and Persuasive We are currently considering how to use the display to Technology: An Exploratory Study in the Context of provide aggregate information about the contents of a Healthy Living. Proceedings of the First International participant’s trolley. The display could indicate how Workshop on Nudge and Influence in Mobile Devices, averaged values of all the participant’s purchases relate to pp. 19-22. some norm(s), for example, is the weekly shop below or 4. Reischach, F., Michahelles, F., Guinard, D., Adelmann, above the average shopper’s food miles. Clearly, there are R., Fleisch, E., Schmidt, A.: An Evaluation of Product normalization issues to be resolved to enable such Identification Techniques for Mobile Phones. In: comparisons to be made. One ethical consideration with Proceedings of the 12th IFIP TC 13 international Conference on Human-Computer Interaction, pp. 804-- this type of display is that even if an observer did not know 816 (2009) what aspect of product information the aggregate display 5. Santilli, P. The Informative and Persuasive Functions of encoded, under certain conditions it could be evident Advertising: A Moral Appraisal. Journal of Business whether a participant was above or below a norm, thereby Ethics, 27--33, 1983. compromising a shopper’s privacy. For example, if the 6. Simon, H. A.: Invariants of Human Behavior. Annual observer had also used the display themselves and the Review of Psychology, 41, 1--19 (1990) colour encoding was fixed. One way to ensure privacy is to 7. Todd, P.M.: How Much Information Do We Need? allow participants to customize aspects of the display, such European Journal of Operational Research, 177, 1317-- as the colour encoding used. A second ethical concern with 1332 (2007) this sort of display is that norms, like salience, typically 8. Todd, P.M., Gigerenzer, G.: Environments That Make Us Smart: Ecological Rationality. Current Directions in influence people unconsciously. To ensure that the Psychological Science, 16(3), 167--171 (2007) autonomy of participants is not compromised it seems 9. Todd, P. M., Rogers, Y. and Payne, S. J. Nudging the important to inform them about the methods used in a Cart in the Supermarket: How much is Enough display and how these typically influence behaviour before Information for Shoppers. In: Proceedings of they choose to use the Augmented Shopping Trolley NIMD2010, pp. 23 – 26 (2010) 10. Underhill, P. Why We Buy: The Science of Shopping. CONCLUSIONS Simon and Schuster: New York. 2009 Our lab-based study shows that participants can rapidly read a shopping trolley handlebar display to determine both Persuasion In-Situ: Shopping for Healthy Food in Supermarkets Ole Kallehave, Mikael B. Skov, Nino Tiainen HCI Lab, Department of Computer Science, Aalborg University Selma Lagerlöfs Vej 300, 9220 Aalborg East, Denmark ole-kallehave@rocha.dk, dubois@cs.aau.dk, ninodk@gmail.com ABSTRACT products, e.g. they cannot understand nutrition labels or Healthy lifestyle is a strong trend at the moment, but at the how much sugar or fat the product contains [5]. Further, same time a fast growing number of people are becoming one of the fundamental problems resides in the fact that we over-weight. Persuasive technologies hold promising are confronted with an overwhelming number of different opportunities to change our lifestyles. In this paper, we food products and it is often difficult to identify and choose introduce a persuasive shopping trolley that integrates two the more healthy ones. Iyengar and Lepper showed in an tools of persuasiveness namely reduction and suggestion. experimental study that consumers were more satisfied The trolley supports shoppers in assessing the nutrition with their own selections when they have fewer options to level for supermarket products and provides suggestions for select from [5]. Schwartz refers to this as the paradox of other products to buy. A field trial showed that the choice claiming that the huge number of choices decreases persuasive trolley affected the behaviour of some shoppers people’s real choice and decision-making [10]. Thus, especially on reduction where shoppers tried to understand people are likely to continue their current routine type of how healthy food products are. On the hand, the suggestion behaviour (as illustrated by Park et al. [8]) and this could part of the system was less successful as our participants potentially prevent them from making healthier choices. made complex decisions when selecting food. Emerging technologies are increasingly being used to alter Author Keywords people’s opinions or behaviour, e.g. smoking cessation [4] Shopping, health, persuasive, supermarkets. or promoting sustainable food choices [7]. Fogg refers to such technologies as persuasive technologies or captology ACM Classification Keywords [3]. Fogg states that contemporary computer technologies H5.m. Information interfaces and presentation (e.g., HCI): are currently taking on roles as persuaders including Miscellaneous. classical roles of influence that traditionally were filled by INTRODUCTION doctors, teachers, or coaches [3]. Research studies within Healthy lifestyles is a hot topic in most Western societies as different disciplines are increasingly concerned with such a rapid growing number of citizens are either over-weight persuasive technologies that may be used to create or or obese, e.g. more than 50% of the adult population in change human thought and behaviour. As examples, Chang Denmark are either over-weight or obese [9]. Over-weight et al. [2] propose the Playful Toothbrush that assists parents problems come from several circumstances, e.g. the lack of and teachers to motivate young children to learn thorough exercise or unhealthy food, but in general people buy and tooth brushing skills while Arroyo et al. [1] introduce the consume food that contains a lot of sugar or fat. Thus, we Waterbot that motivates behaviour at the sink for increased need to alter people’s behaviour and attitude while they safety. Both these examples propose rather simple, yet shop groceries and other food products in supermarkets. potentially powerful input and feedback that aim to inform users of their own behaviour. When supermarket shopping, more studies have shown that consumer behaviour is highly controlled by routine and is Todd et al. [11] illustrate theoretically how nudging could not simply changed or altered [8]. In fact, even if shoppers persuade shoppers to select healthy food products based on want to change their shopping behaviour and patterns, they simplified information to the shoppers in-situ, but call for find it difficult to understand the nutritious values of many empirical understandings of persuasive shopping. We propose a persuasive shopping trolley application called iCART that attempts to motivate change towards more healthy shopping behaviour. First, we outline the idea Copyright © 2011 for the individual papers by the papers' behind the design of the trolley application and then reports authors. Copying permitted only for private and academic from field studies of use on its effects on behaviour change. purposes. This volume is published and copyrighted by the iCART: INFLUENCING SHOPPING BEHAVIOUR IN-SITU editors of PINC2011 iCART is a persuasive application mounted on a shopping trolley that attempts to persuade the shopper’s behaviour and awareness. The system was implemented in C# using Windows Presentation Foundation for the interface and a Reduction reduces complex behaviour to simple tasks in Microsoft SQL server. order to increase the benefit/cost ratio and thereby From our previous research [6], we learned that many influence the user to perform the behaviour [3]. As stated consumers actually attempt to buy healthy products when above, consumers find it difficult to assess the overall supermarket shopping, but often they would find it difficult nutrition level for products. The persuasive trolley reduces to assess the nutrition value or energy level. In fact, several this nutrition value assessment through the simplification in consumers are actually unsure what a healthy food product the Eat Most classification and thereby the assessment now is. Shoppers find it difficult to understand the nutrition becomes a simple task. This is illustrated in figure 2 where information labels on the food products and they usually different products have been classified, e.g. milk as eat less don’t bother consulting this information. Supermarket (middle picture). products and groceries are rather diverse, e.g. ranging from simple non-processed products (e.g. an apple) to more complex processed products (e.g. a pizza). Usually people find it difficult to assess how healthy processed products are. Furthermore, people find it difficult to change behavior and usually choose well-known products while shopping. The overall idea of iCART is that all food products and items in a supermarket can be classified according to nutrition level and this classification will be presented to the user of the trolley every time the shopper puts an item Figure 3: Example of reduction in persuasion: Classification of the cereal product Havrefras as Eat Least into the trolley. For our persuasive system, we adapt the nutrition label initiative called Eat Most from the Danish We colour-coded the three categories with green, yellow, Veterinary and Food Administration. For our purpose, it and red. Figure 3 shows the classification for a cereal provides a simple classification of food products based on product called Havrefras and this product is an eat least the nutrition values of a product. The classification label product. The implementation in iCART reduces the action includes a table for calculating the value of all food of assessing the nutrition value of a product by providing a products. According to the label, all products can be simple classification of only three categories. classified as Eat Most, Eat Less, or Eat Least. The typical use situation could be as follows (illustrated in figure 1): The user walks around the supermarket with the trolley, chooses food products and places them in the trolley (a), the trolley recognizes the product and displays its classification according to the Eat Most label (b), and the system updates the status for the entire trolley on numbers of Eat Most, Less, and Least food products (c). Figure 4: Examples of suggestion in persuasion: Two alternative cereal products that are both Eat Most products. Suggestion means that persuasive technologies have greater power if they offer suggestions at opportune moments [3]. (a) (b) (c) Consumers find it difficult to choice healthier alternatives Figure 1: Illustrating the process of using iCART as they often have limited understanding of the relative Interaction Design levels of nutrition between more products. The persuasive We adapted three persuasive design tool principles from trolley offers suggestions for alternative products (Eat Fogg namely reduction and suggestion [1]. The persuasive Most) within the same product group when the shopper shopping trolley should 1) present or visualize product choices an Eat Less or Eat Least product in the trolley. We nutrition in a simple way and 2) present alternatives to less consider this an opportune moment as the shopper often healthy products. Finally, we decided that the system will find the alternatives in their present supermarket area should be a walk-up-and-use system on a shopping trolley. (as illustrated in figure 4 where two alternative cereals are suggested for the cereal in figure 3). FIELD TRIALS We conducted field trials with the shopping trolley at the Figure 2: Three classifications of the Eat-Most nutrition label local supermarket called føtex. It was rather important to us with eat most (left), eat less (middle), and eat least (right). to understand the use of the system in-situ to facilitate the whole shopping experience. 11 shoppers were recruited through public announcements guided the shopper while shopping. This also had the and we required that they shopped for food products on a advantage that shoppers always knew where to look for the regular basis. The shoppers were between 27 and 58 years nutritious information for all products. Today, this old and represented different kinds of households and information is located on the packaging of the product and worked in diverse job professions. We asked them to fill in thereby distributed in the store. a questionnaire on their supermarket shopping experiences The reduction element of iCART was quite successful. Out prior to the trials. Some of the participants were highly of the 60 food products selected by the participants using concerned with nutritious food while others were less the system, 30 were classified as Eat Less or Eat Least. concerned. The participants were divided into two groups - Thus, half of the selected products were less healthy. In one group used iCART while the other group served as a several cases, the participants were surprised to realize that control group using a regular shopping trolley. We a certain product was less healthy. For example, one of the balanced them in the two groups based on their self- participants chose a bag of carrot buns and got surprised to reported knowledge and attitudes towards nutritious food. see that these buns were Eat Least: “I thought they were Before the trials, we carried out a pilot test to verify and healthy as they contain carrots”. adjust the process and our instructions. Participants were not informed about the purpose of the study in order to On the other hand, several shoppers chose less healthy food minimize study impact and iCART participants were told products and were aware of it – even without the help from about the system but not its focus on healthy food products. iCART. But the classification made them reflect upon their choices and several of them started talking about nutrition The trials consisted of a three parts namely an introduction, and healthy food. One participant said: “But the Eat-Least the actual shopping, and a debriefing. We instructed the classification makes you think and questions whether you participants to shop items from a pre-generated shopping have made the right choice”. From our analysis, it seemed list using their own normal criteria for food selection. Thus, that they acted out of routine behaviour and that they they should try to shop as they normally would. The partially knew the consequences of these choices. This shopping list contained 12 items, e.g. milk, cheese, pate. confirms the findings by Park et al. [8] on changing The list included only general product groups (except for shopping routine behaviour. In summary, the reduction one item) leaving the participants to choose within the element of iCART was quite successful as it raised the group, e.g. cheese where they could choose more 20 awareness of the shoppers on the nutritious level of the different cheese products. They were free to choose in chosen products. which order they would collect the items. The suggestion component of iCART was less successful 303 food items were entered into a SQL database compared to the reduction. The participants changed their representing all items in the store within the groups from choices 3 times out of 30 (10%). This low number was the shopping list. Data collection was done through 1) a somewhat surprising, but shoppers gave several reasons for trolley-mounted video camera that captured verbal this. Some would not change their choice, as they would comments and shopping behaviour and 2) the system rather buy an unhealthy food product that was biodynamic logged and time stamped all user interactions enabling to than buy a healthy product that was not. So the shoppers reproduce action sequences afterwards. The sessions were would implement their own classification schemes based done during normal trading hours and they were not on other aspects than nutrition. Also, some shoppers stated required to check out the collected items. that they never bought any light or zero products, which We evaluated iCART as a Wizard of Oz experiment where often were the products suggested by our system. They said one of the authors acted as wizard implementing the actions that they would rather eat less of the unhealthy products taken by the participant. When a food product was put into than buy a light product. the trolley, the wizard would update this information in the During the field trials, 18 times did the shoppers take a look system. Another person observed the participant while at the suggestions made by iCART, but in most situations shopping in order to facilitate the following interview. The (14 times) they chose not to follow the suggestion. This same procedure was used for the control group, but without indicates that the shoppers are interested in receiving the trolley-mounted display. The total time spent ranged suggestions but the actual suggestions made by the system from 12:08 to 40:28 minutes. Finally, a debriefing session in the situation were not good enough. As illustrated above, including questionnaires and semi-structured interview was they had different objectives when shopping and perhaps conducted immediately afterwards, e.g. they were asked to suggestion functionality should be carefully organized. assess their own session and the collected items. We identified an interesting observation concerning trust to OBSERVATIONS AND DISCUSSION the system. Some users expressed scepticism towards the The five participants using iCART expressed that they suggestion part of the system while none of them really liked the system and they would possibly use it if available questioned the reduction part. Most of them stated that in supermarkets. While food products in supermarkets nutrition labelling whether on the actual product or already have different labels for determining the health or implemented in an interactive system on the trolley should nutritious level, iCART became a personal technology that be controlled and accredited by public authorities. They were more critical when it concerned suggestions than REFERENCES reductions. The problem with suggestion could reside in 1. Arroyo, E., Bonanni, L., and Selker, T. (2005) that it could feel like ads or commercials for other products. Waterbot: exploring feedback and persuasive techniques That could be a potential problem when implementing at the sink. In Proceedings of the 23rd Annual SIGCHI suggestion tools. However, as expressed by one of the conference on Human factors in Computing Systems female participants: “It is cool to be guide. I don’t mind (CHI’05), ACM Press, pp. 631-639 help or receive suggestions, I’m a grown-up who can make 2. Chang, Y-C, Lo, J. L., Huang, C-J, Hsu, N-I, Chu, H-H, my own decisions”. This could imply that to change Wang, H-Y, Chi, P-Y, Hsieh, Y-L. (2008) Playful behaviour designers should focus on providing reduction in toothbrush: Ubicomp technology for teaching tooth complexity of assessing the food product, but they should brushing to kindergarten children. In Proceedings of perhaps not suggest or give recommendations to the user. 26th Annual SIGCHI conference on Human factors in Shopping in supermarkets is noisy and complex and it can Computing Systems (CHI’08), ACM Press, pp. 363-372 be stressing due to several multimodal inputs. We noticed 3. Fogg, B. J. (2003) Persuasive Technology – Using how several participants missed reductions or suggestions Computers to Change What We Think and Do. Morgan on the screen while acting in the environment. Thus, they Kaufmann would actually not receive the information proposed by the system. Also, one participant stated that shopping is private 4. Graham, C., Benda, P., Howard, S., Balmford, J., even though it takes place in a public environment. Bishop, N., and Borland, R. (2006) "heh - keeps me off the smokes...": probing technology support for personal The participants who shopped without the persuasive change. In Proceedings of the Australia conference on guidance appeared to have fewer reflections on nutrition Computer-Human Interaction (OzCHI’06), pp. 221-228 and health. In fact, the iCART participants eventually bought 25 food items classified as Eat Least whereas the 5. Iyengar, S. S., & Lepper, M. (2000) When Choice is other participants bought 34 Eat Least products. The Demotivating: Can One Desire Too Much of a Good difference cannot only be explained in terms of the Thing? Journal of Personality and Social Psychology, suggestion tool implemented in iCART, but the interaction Vol. 79, pp. 995-1006 made them reflect. 6. Kallehave, O., Skov, M. B., and Tiainen, N. (2010) Reducing the Paradox of Choice: Designing a Nutritious CONCLUSION Persuasive Shopping Trolley. In Hasle et al. (eds.) We presented the persuasive shopping trolley iCART that Poster Proceedings of the 5th International Conference guides supermarket shoppers in choosing more healthy on Persuasive Technology (Persuasive 2010), Oulu food products by classifying all products in three groups University Press, p. 25-28 namely Eat More, Eat Less, and Eat Least. Field trials with 11 shoppers showed that iCART proved to provide good 7. Linehan, C., Ryan, J., Doughty, M., Kirman, B., and input on reduction, e.g. reducing the complex task of Lawson, S. (2010) Designing Mobile Technology to assessing whether a product is healthy or less healthy. Our Promote Sustainable Food Choices. In Proceedings of participants noticed when the system classified a product as first International Workshop on Nudge & Influence Eat Least and usually they would start reflecting upon this. Through Mobile Devices, pp. 15-18 Only a few times did this result in change of behaviour 8. Park, C. W., Iyer, E. S., and Smith, D. C. (1989) The where the user changed the original choice. But mostly the Effects of Situational Factors on In-Store Grocery suggestion part of the system was less successful. This was Shopping Behaviour: The Role of Store Environment mainly due to the fact that several participants had rather and Time Available for Shopping. Journal of Consumer specific requirements to their products, e.g. they should be Research. Vol. 15, pp. 422- 433 biodynamic or they never bought light-products. 9. Richelsen, B., Astrup, A., Hansen, G. L., Hansen, H. S., Based on our findings, we see a number of future research Heitmann, B., Holm, L., Kjær , M., Madsen, S. A., avenues. First, rather than optimizing the algorithms behind Michaelsen, K. F., and Olsen, S. F.: The Danish obesity suggestion tools, we propose that we should design systems epidemic that enables shoppers to make their own decisions in-situ. 10. Schwartz, B. (2004) The Paradox of Choice – Why This could require a different approach to reduction. Also, More is Less. HarperCollins Publishers, New York we need to understand the long-term effects of such systems and we plan to conduct more longitudinal studies. 11. Todd, P. M, Rogers, Y., and Payne, S. J. (2010) Nudging the cart in the supermarket: How much is ACKNOWLEDGMENTS enough information for food shoppers? In Proceedings We would like to thank the shoppers from the field trial as of first International Workshop on Nudge & Influence well as reviewer comments on earlier versions of the paper. Through Mobile Devices, pp. 23-26 Towards a Mobile Application to Create Sedentary Awareness Gijs Geleijnse, Aart van Halteren and Jan Diekhoff Philips Research Eindhoven, The Netherlands gijs.geleijnse@philips.com , aart.van.halteren@philips.com ABSTRACT successful persuasive strategies to establish an increased Prolonged sitting time is a potential health risk, not only for amount of physical activity. people with an inactive lifestyle, but also for those who do meet the recommended amount of physical activity. In this Recent medical literature reports that not only an inactive paper, we evaluate SitCoach, a mobile application to nudge lifestyle may lead to adverse health effects, but also people from their seats. The application is targeted to office sedentary behavior itself is harmful. Prolonged sitting time workers. SitCoach monitors physical activity and sedentary is also dangerous for people who meet the WHO guidelines behavior to provide timely feedback by means of of 30 minutes of physical activity per day [4,12]. The suggesting sitting breaks. A pilot experiment with a group reduction of sedentary behavior is hence identified as a of 8 users learned that the general awareness of the target behavior that contributes to a healthy lifestyle. importance of sitting breaks is low. Combined with the 6XSSRUW WR FUHDWH DZDUHQHVV RI RQH¶V VHGHQWDU\ EHKDYLRU belief that the ability to take sitting breaks is highly may be beneficial. However, as Owen et al. state in [12], dependent on external factors, a strategy of proposing break ³JLYHQ WKH UHFHQW UHFRJQLWLRQ RI WKLV SKHQRPHQRQ RI WRR reminders may not be the most successful for this target much sitting, there are not yet any recommended clinical group. Future work should focus on raising awareness of guidelines. Commonsense might suggest that it may be the problem and providing insights into personal sitting prudent to try to minimize prolonged sitting with 5 minute behavior. EUHDNVHYHU\KRXU´ Author Keywords In this paper, we describe SitCoach, a mobile application Sitting time, mobile persuasion, sedentary awareness, that assists the user to create sedentary awareness and to physical activity. have regular sitting breaks. Such an application can be combined with additional physical activity promotion ACM Classification Keywords features. To the best of our knowledge, SitCoach is the first H5.m. Information interfaces and presentation (e.g., HCI): prototype mobile application aimed to reduce sitting time. Miscellaneous. Using SitCoach, the goal is to collect insights into INTRODUCTION SRVVLELOLWLHV WR LQIOXHQFH SHRSOH¶V VLWWLQJ EHKDYLRU XVLQJ D In the past years, a substantial amount of research has been mobile device. devoted to physical activity promotion through mobile SitCoach targets office workers, a group which is often also devices. Using the accelerometer embedded in a mobile assisted by break reminder applications on their PCs. Such phone or in a dedicated device, the energy expenditure of applications are developed to prevent their users from the user can be estimated. The user may receive feedback repetitive strain injuries. Although such applications show on his past physical activity level in minutes or burned to be successful in reducing complaints [7], they may not calories. always be pleasant to use [10]. Morris et al. [10] introduced Several strategies have been explored to influence the SuperBreak, which stimulates break compliance for XVHU¶VEHKDYLRUDQGSURPRWHKLJKHUSK\VLFDODFWLYLW\OHYHOV computer usage. Instead of the usual breaks offered by Most notably, the usage of virtual rewards [1,2], social software packages such as XWrits and WorkRave, support [3,9] and goal setting [8] have shown to be SuperBreak offers the possibility to make the break time more productive. By offering the user the possibility to interact with the PC through gestures during the break, break compliance is promoted and the productivity during the break time is increased. Hence, although SuperBreak may increase break compliance for computer work, it does Copyright © 2011 for the individual papers by the papers' authors. not target a reduction in sitting time. Moreover, neither of Copying permitted only for private and academic purposes. This volume the computer packages support break compliance during is published and copyrighted by the editors of PINC2011." other sedentary time, e.g. during meetings or while reading. After describing the SitCoach application in the next chapter, we present a first pilot user experiment to assess the usability of the application. Through a locus of control questionnaire and by means of a semi-structured interview, we gather additional insights on opportunities and techniques to promote sitting break compliance. INTRODUCING SITCOACH SitCoach is an iPhone application that measures physical activity by means of the built-in accelerometer. The application records active time and sitting time at a granularity of one minute. To fight sitting time and inspire people to take a break once in a while, the SitCoach reminds users after a configurable number of in-active minutes via visual, acoustic and tactile messages. Users set their goals in terms of maximum number of consecutive sitting minutes and number of active minutes per day. Identifying Sitting Time Using the built-in accelerometer in the smart phone, the XVHU¶V DFWLYLW\ LV FODVVLILHG LQ DQ DFWLYH DQG LQDFWLYH VWDWH Every sHFRQG D PHDVXUHPHQW RI WKH SKRQH¶V [ \ DQG ] Figure 1. SitCoach main screen. positioning is taken by the accelerometer. These three values are compared with the previous measurement. When A FIRST USER EVALUATION the difference for x,y or z exceeds 0.3 the accelerometer To assess the usability and user acceptance of the recognizes a movement. The 0.3 was determined application, SitCoach has been evaluated with users. This empirically: it is low enough to pick up the walking evaluation also provides insights into WKH SDUWLFLSDQWV¶ movement of the user without getting a false reading from current sitting behavior and their awareness of the other possible movements like a small turn with the chair harmfulness of sedentary behavior. The goal of the study is while sitting. to identify future directions for persuasive applications targeting sedentary awareness. To distinguish walking from other smaller movements like a small turn or just standing up from a chair the movement In the study, the participants are provided with an iPhone will be monitored over a certain interval of time. An with the SitCoach application and are invited to use the empirically determined value of 5 seconds proved to be application throughout a day at the office. At the end of the sufficient. day, a semi-structured interview is conducted, to discuss experiences. Moreover, the participants are questioned Creating Sedentary Awareness about current sitting break habits and the awareness of the To motivate users to become more active, the application importance of such breaks is assessed. Apart from the stores the number of active minutes per day for each of the interview, two questionnaires were handed to the users. This provides a social nudge for users to see how participants: one focusing on the utilitarian and hedonic others are doing and to comply with the social norm. qualities of the application [5,6] and a second one focusing When it is time to take a break, SitCoach emits a tactile on the locus of control that people perceive with respect to (vibration) and an acoustic warning. Users can override the possibilities to reduce their sitting time [13]. acoustic warning. A visual indicator at the main screen Participants shows when a user is moving, giving the user immediate Eight participants (four females) were invited to participate feedback about their current behavior. Figure 1 provides a in the experiment, during one working day. All participants screenshot of the main screen of SitCoach. The green circle were knowledge workers with high computer dependability. indicates that the application has detect that the user is Procedure and Design currently moving and hence the number of active minutes is increasing while in this state. In the state displayed in the The participants were scheduled on a day they described as figure, the user is nine inactive minutes away from a break a typical office day. Per participant, a day was selected reminder. However, if the user is active for a period equal without having appointments outside the office during to the actual time of the sitting break, the break timer will working hours. be reset. In the morning after arriving at the office, the participants received a fully charged iPhone 3G. SitCoach was the only which is installed by default. The others have disabled it. application installed, apart from the standard software. The For a mobile application to create sedentary awareness, the participants were instructed not to use the phone for other perceived control over the sitting breaks should remain with purposes. No SIM card was installed, limiting the the user. functionalities of the phone. The interviews showed that the phone vibration to signal During the intake meeting, the participants were explained break alerts was appreciated as it is discrete and easy to the functionality of the applications and guided through the ignore when needed, for example during meetings. On the features and settings. The standard break timer was set to other hand, the buzzing signal was experienced to be 60 minutes, prompting for a 5 minute break. The standard GLVWUDFWLQJ ³:KHQ , DP ZRUNLQJ , GRQ¶W ZDQW WR EH activity goal was set to 50 minutes. The participants were GLVWXUEHG´ free to change the settings throughout the day. The Locus of Control questionnaire revealed that six out of $URXQG  R¶FORFN LQ WKH DIWHUQRRQ WKH SDUWLFLSDQWV ZHUH eight participants scored low on the internality dimension interviewed based on a list of pre-defined questions on their (scores < 18 on a range from 6 to 36), while the other sitting behavior, sedentary awareness and the SitCoach scored moderate (18 ” score ” 24). This implies that the application. Moreover, the two questionnaires were handed. office workers participating in the study believe that they have little control over their sitting behavior. With overall The Attrakdiff2 questionnaire was presented to assess both higher scores on the powerful others dimension, it is the pragmatic and hedonic qualities of SitCoach [5,6]. His believed that others (colleagues, managers) strongly scores on both qualities are important for the prolonged GHWHUPLQHWKHSDUWLFLSDQWV¶VLWWLQJEHKDYLRU usage of a product. Specifically, the questionnaire measures perceived pragmatic quality, hedonic quality identification The Attrakdiff2 questionnaire results show favorable scores LH GRHV WKH SURGXFW FRQWULEXWH WR WKH XVHU¶V LGHQWLW\ LQ D on the pragmatic dimension, implying that the participants social context?), hedonic quality stimulation (i.e., does the are generally positive about the interaction with the product help to develop skills or knowledge) and SitCoach application. No remarks were made about any attractiveness (is the product good, bad or ugly?). Each of inaccuracies of the application. This suggests that the those four categories contains seven word-pairs on a seven current implementation is well usable to distinguish sitting point semantic-differential scale (e.g. discouraging vs. time from active time. Lower scores were reported on the motivating, complicated vs. simple). hedonic dimensions, most notably on attractiveness. 7RDVVHVVWKHSHUFHLYHGORFXVRIFRQWUROWRLQIOXHQFHRQH¶V Table 1. 8VHUV¶UHVSRQVHVWRWKHORFXVRIFontrol questionnaire. sitting behavior, a locus of control questionnaire was Participant I nternality Powerful Chance assessed [13]. The commonly used questionnaire, others externality developed by Wallton et al., is adapted for sitting behavior. externality The questionnaire measures whether the control over the sitting behavior is determined internally (i.e. self-control; Person 1 Moderate Moderate Low example statement: If I take care of myself, I can avoid long Person 2 Moderate High Moderate sitting periods), by others (e.g. Whenever I feel I sit too much and too long, I should consult a trained professional.) Person 3 Low High Moderate or by chance (e.g. No matter what I do, I 'm likely to have Person 4 Low High Moderate long sitting periods). Results Person 5 Low Moderate Moderate All participants indicated that they were not aware of the Person 6 Low Moderate Low harmfulness of sedentary behavior itself. When taking a Person 7 Low High Low break and getting up from their desk, the participants did so because they were aware of the adverse effects of Person 8 Low High High prolonged computer usage and the healthfulness of physical Some of the participants reported battery problems with the activity. Half of the participants reported to be unhappy smart phone. Although the participants received a fully with the amount of sitting time during a day in the office. charged phone, the battery time was not enough for the Suitable moments to take a sitting break are in between application to run for the whole working day. Hence, in tasks and when feeling less concentrated. The time spent future work, solutions should be researched that take the during such breaks is not seen as productive. energy consumption of the phone into account when The lack of control is seen as the largest source of running such accelerometer-based applications. annoyance with PC break applications. Only one of the The functionality to share the activity minutes on FaceBook participants is using an RSI prevention program on the PC, or other social media was not well received. Similar to the findings of Munson et al. [11], participants did not feel the Avrahami, D., Froehlich, J.E., LeGrand, L., Libby, R., need to bother their social network with such details. Mosher, K., & Landa\-$³)ORZHUVRUD5RERW$UP\" Encouraging Awareness & Activity with Personal, Table 2. 8VHUV¶UHVSRQVHVWRWKHAttrakDif2 questionnaire. 0RELOH'LVSOD\V´3URFRI8EL&RPS-63. Ppn Pragmatic Hedonic Hedonic Attractive 3. Fujiki, Y (2010). iPhone as a Physical Quality Quality Quality ness Activity Measurement Platform. In Proceedings of the I dentification Stimulatio 2010 ACM Conference on Human Factors in Computing n Systems (CHI). 1 High Moderate High Moderate 4. Hamilton, M.T., Healy, G.N., Dunstan, D.W., 2 High Low Low Low Zderic, T.W., and Owen, N. (2008). Too little exercise and too much sitting: Inactivity physiology and the need 3 Moderate Moderate Moderate Low for new recommendations on sedentary behavior. Current 4 Low Moderate Moderate Low Cardiovascular Risk Reports 2(4), 292-298. 5. Hassenzahl, M. (2006) Hedonic, emotional and 5 High High Moderate Moderate experiental perspectives on product quality. In Ghaoui, C. 6 High High High Low (ed) Encyclopedia of Human-Computer Interaction. Hershey: Idea group, 226-272. 7 High Moderate High Low 6. Hassenzahl, M. (2010) Attrakdiff. Retrieved 8 High Moderate High Moderate August 11th 2010 from http://www.attrakdiff.de/en/AttrakDiff-/What-is- AttrakDiff/Scientific-Background. CONCLUSION AND FUTURE WORK In this paper, we presented an application to assist people to 7. Heuvel, S.G. van den, Looze, M de, Hildebrandt, control their sitting behavior. The mobile application V.H., and The, K.H (2003). Effects of software programs combines feedback on physical activity with insights on the stimulating regular breaks and exercises on work-related XVHU¶V VLWWLQJ SHULRGV SitCoach was developed to gain neck and upper-limb disorders. Scand J Work Environ LQVLJKWVLQWRSHRSOH¶Vawareness of their sedentary behavior Health 29(2):106±116. and the user acceptance of a break reminder application. 8. Lacroix, J., Saini, P., Holmes, R. (2008). The relationship between goal difficulty and performance in With SitCoach, we have created an application that detects the context of Physical Activity. IQ0RELOH+&,¶. sitting time with fair accuracy. However, the users involved in the trial showed not to be in the right stage of change to 9. Lin, J; Mamykina, L; Lindtner, S; Delajoux, G; be responsive to the strategies applied in SitCoach. 6WUXE +%   ³)LVK¶Q¶6WHSV (QFRXUDJLQJ 3K\VLFDO Persuasive strategies to stimulate the user to take sitting $FWLYLW\ZLWKDQ,QWHUDFWLYH&RPSXWHU*DPH´8EL&RPS breaks are likely to be more successful after having 2006. established awareness of the adverse health effects of sitting 10. Morris, D., Bernheim Brush, A.J. and Meyers, B.R behavior. This can be done by first providing insights in (2008). SuperBreak: Using Interactivity to Enhance RQH¶ VLWWLQJ EHKDYLRU DQG VXEVHTXHQWO\ VXJJHVWLQJ Ergonomic Typing Breaks. Presented at CHI 2008. opportunities to reduce sitting time. For users who are 11. Munson, S., Lauterbach, D., Newman, M.W., aware of the problem and the adverse effects of their Resnick, P. (2010). Happier Together: Integrating a behavior, the triggers applied in SitCoach may be revisited. Wellness Application into a Social Networking Site. ACKNOWLEDGMENTS PERSUASIVE 2010. This work was funded by the European Commission, within 12. Owen, N., Bauman A., and Brown, W. (2009). the framework of the ARTEMIS JU SP8 SMARCOS Too much sitting: a novel and important predictor of project ± 100249 - (http://smarcos-project.eu). chronic disease risk? British Journal of Sports Medicine 42(2). REFERENCES 1. Consolvo, S., Everitt, K., Smith, I., & Landay, 13. Wallton, K.A, Strudler Wallton, B., DeVellis, R. -$ ³'HVLJQ 5HTXLUHPHQWV IRU 7HFKQRORJLHV WKDW (1978). Development of the Multidimensional Health (QFRXUDJH3K\VLFDO$FWLYLW\´ 3URFHHGLQJVRI &+, Locus of Control Scales. Health Educ Behav 6(1):160 ± 457-66. 170. 2. Consolvo, S., Klasnja, P., McDonald, D.W., MONARCA: A Persuasive Personal Monitoring System to Support Management of Bipolar Disorder Gabriela Marcu Jakob E. Bardram Human-Computer Interaction Institute IT University of Copenhagen Carnegie Mellon University Rued Langaards Vej 7, Copenhagen, Denmark 5000 Forbes Ave, Pittsburgh, PA, USA +45 7218 5311 gmarcu@cs.cmu.edu bardram@itu.dk ABSTRACT multitude of ways. For example, patients and their clinicians MONARCA is a persuasive mobile phone application de- can use the data to determine the effectiveness of medica- signed to support the treatment and management of bipolar tions, find illness patterns and identify warning signs, or test disorder. Behavioral data is monitored through both sensing potentially beneficial behavior changes. Behavioral data and manual patient input, while timely feedback is provided collected could be used to predict and prevent the relapse of based on clinical recommendations to help patients adjust critical episodes. their behavior and manage their illness. This paper presents Despite the plethora of research into personal monitoring the design process behind the MONARCA system and ini- systems targeting behavior change [8], health-related behav- tial findings on the challenge of designing a persuasive sys- ior change (e.g., physical activity [5, 1], diet [9], cardiac tem for the management of bipolar disorder. We discuss rehabilitation [6], and others [3]), and even the management how difficult the design of such technology has turned out of chronic illnesses (e.g., diabetes [7, 11], chronic kidney to be, for two primary reasons: (1) the inherent challenges disease [10], asthma [4]), mental illness has remained rela- of using persuasive metaphors with a complex mental ill- tively unexplored. One explanation for this untapped poten- ness, and (2) the tradeoffs encountered due to varying, and tial is the complexity and variation of a mental illness like sometimes conflicting, stakeholder needs. bipolar disorder, which causes uncertainty in how to manage it. Moreover, there is no simple connection between measur- Author Keywords able parameters and the course of treatment; mental illness is Bipolar disorder, mental illness management, user-centered fundamentally complex and is often tied into physical health design, personal monitoring systems problems as well as social problems. In the MONARCA project we aim to overcome this challenge by developing a ACM Classification Keywords system that, through pervasive data collection and feedback H.5.2 Information Interfaces and Representation: User In- to the patient, supports the treatment of bipolar disorder. terfaces – User-centered design. J.4 Social and behavioral As such, the MONARCA system can be classified as a per- systems: Psychology. suasive technology [2], similar to other persuasive health- related ubiquitous computing systems. The design of such INTRODUCTION persuasive systems is, however, extremely difficult. It is Persuasive personal monitoring systems seem promising for very unclear how feedback should be given to the patient in the management of mental illnesses such as bipolar disorder. order to influence and change behavior. Numerous studies Bipolar disorder is characterized by recurring episodes of have proven that that trying to change unhealthy behavior both depression and mania, with treatment aiming to reduce such as smoking, drinking, or lack of exercise is extremely symptoms and prevent recurrence throughout a patient’s difficult even with the use of intensive counseling. Medicine lifetime. By applying pervasive healthcare technologies to compliance is also a fundamentally hard problem in the treatment of bipolar disorder, we can monitor patients’ healthcare. Therefore, it is quite challenging – some would behavioral and mood data, and provide timely feedback to say naïve – to rely on non-human actors like computers and them in order to help them adjust their behavior. This data mobile phones to be able to change unhealthy behavior. supports the treatment and management of the illness in a In this paper, we describe the user-centered design process and initial findings on the challenge of designing a persua- sive system for the management of bipolar disorder. We discuss how difficult the design of such technology has turned out to be, for two primary reasons: (1) the challenges of using persuasive metaphors with a complex mental ill- ness, and (2) the tradeoffs encountered due to varying, and sometimes conflicting, stakeholder needs. METHOD (based on phone calls and text messages). This data is ab- Patients and clinicians of a bipolar disorder treatment pro- stracted for analysis, to protect the patient’s privacy while gram took part in an in-depth participatory design process. still supporting self-assessment using objective data. They were instrumental in decision-making about features through collaborative design workshops and iterative proto- Historical overview of data typing. Patients participated in semi-structured interviews The patient and clinician will both have access to the data about the treatment and management of their own illness to through a web interface. This will give them the means to further inform the design process. Notes and artifacts from explore the data in depth by going back and forth in time, these design activities were analyzed for 1) an understand- and focusing on specific sets of variables at a time. ing of each stakeholder's motivations and needs, and 2) indicators of tradeoffs that arose in the design of the sys- Coaching & self-treatment tem. Psychotherapy will be supported through everyday rein- forcement in two ways. Customizable triggers can be set to Workshops were held every other week for six months. At have the system notify both patient and clinician when the every workshop, 1-3 individuals attended from each of the data potentially indicates a warning sign or critical state. following three stakeholder groups: patients, clinicians, and Second, after patients are advised by their clinicians about designers. The designers led each three-hour workshop by which actions to take in response to warning signs, they can facilitating discussion about particular design goals and keep track of and review them through the system. issues; system features and functionality; and feedback on mockups and prototypes of the system. During initial work- Data sharing shops, overall goals of the system were introduced from In order to strengthen the psychotherapy relationship data both clinical and technical perspectives. Sharing these per- and treatment decisions are shared between the patient and spectives of the project involved drawing from their respec- his/her clinician. Similarly, sharing data with family mem- tive best practices: both medically and practically, clini- bers or other caregivers empowers the patient to support the cians know what works with patients; and designers are treatment process. Finally, sharing data among patients aware of related systems and technologies. helps with personal coping and management efforts by re- Design activities at workshops began in the early stages assuring patients that they are not alone, and helping them with hands-on brainstorming. We provided materials such see how others manage their illness. as documents summarizing the goals of the system, images of existing tools and methods, large poster paper, writing CHALLENGES WITH A PERSUASIVE METAPHOR materials, scissors, tape, etc. The sketches that came out of One of the main original goals of the user-centered design this initial brainstorming formed the basis for the first process was to design a persuasive system for bipolar pa- mockups. For the rest of the process, at each workshop we tients, which could help them constantly adjust their behav- 1) discussed a few design goals and system features in ior to manage their own illness. In particular, the design depth, and 2) received feedback on the next iteration of the process revealed the following three parameters were cru- mockups. Mockups presented during workshops progressed cial to keeping a bipolar patient stable: from sketches to wireframes to interactive prototypes. 1. adherence to the prescribed medication – i.e., ensuring that the patient takes his or her medication on a daily SYSTEM DESIGN basis The design process resulted in 5 focus areas for a persua- sive system for bipolar disorder: self-assessment, activity 2. stable sleep patterns – e.g., sleeping 8 hours every monitoring, historical data overview, coaching & self- night and going to bed at the same time treatment, and data sharing. 3. being physically and socially active – e.g., getting out of the home, meeting with people, going to work. Self-assessment Subjective data is collected through a mobile phone using a Now – at first glance, this may seem simple, but numerous simple one-page self-assessment form. Less than 10 items studies have shown that each of the above three things are are entered by the patient on a daily basis, including mood, very difficult to achieve for many patients, and achieving sleep, level of activity, and medication. Some items are all three consistently is inherently challenging in combina- customizable to accommodate patient differences, while tion with a mental illness. Hence, the core challenge is to others are consistent to provide aggregate data for statistical create technology that would help – or “persuade” – the analysis. A simple alarm reminds the patient to fill out the patient to do these three things every day. form. Most persuasive health-related Ubicomp systems have Activity monitoring adopted different metaphors with the goal of motivating the Using sensors in the phone, objective data is collected to patient to perform healthy behavior. Examples of such monitor level of engagement in daily activities (based on metaphors include a garden that grows when the person is GPS and accelerometer), and amount of social activity physically active; a fish that grows when the person walks more; and a dog that is happier when the person eats too much time or attention on the clinician's part, the clini- healthy meals. Common to these metaphors is a simple-to- cians would reject it. An example of one such feature was understand relationship between behavior (e.g. exercise) the system suggesting that the patient contact the clinic if and visualizations in the metaphor (e.g. more flowers in the data collected indicated possible warning signs – and mak- garden). ing it easy for the patient to place this call. The motivation behind this feature was to encourage the patient to reach out In the design of the MONARCA project, we tried to adopt for help when needed, but the clinicians ultimately rejected the same strategy of creating a metaphor. In total of 5 dif- the idea because we could not find a reasonable protocol to ferent metaphors were tested and tried out in a series of make the benefits to the patient outweigh the burden on the design workshops. These metaphors included the use of an clinic's resources. Features of the system also couldn't pre- abstract color picture, a landscape with a river, a dartboard, sent a liability for clinicians, so they were more likely to a music equalizer, and a scale. The patients and clinicians reject ideas and limit the role of the system to be on the safe rejected all of these metaphors – one after the other. side. Any kind of text messages or notes written by the pa- tient and made available to the clinic were kept out of our Why did this happen? First we thought that maybe we were design, because we could not ensure that the clinicians just bad at designing the metaphors, and we kept on trying would always read these messages, so we could not make with new ones. But since it turned out to be a persistent them liable for their content. “problem”, we think that something more fundamental was We therefore realized that designing our system with pri- at stake, which was expressed by one of the patients as: marily a clinical focus was limiting. The clinicians we “I do not want my illness to be reduced to a game.” worked with were clearly most comfortable with strategies that they were familiar with, they had evidence for based on We think that this is an important insight into the design of their experiences with patients, and were backed by clinical persuasive technologies for healthcare and self- trials. Deviating from these practices somewhat, and pushing management. Many of the technologies and metaphors re- our clinicians a little bit out of their comfort zone, enabled ported so far deal with personal lifestyle related health us to explore other potential strategies, from the perspectives management, which is fundamentally different from pa- of the patients and the designers. tients with a diagnosed mental illness. We think that the design of feedback to the patient needs to follow another An additional example of a debated feature is reported stress level. A stress level scale was strongly rejected by a clini- pattern other than using a metaphor. cian who argued that stress is not a clinically useful meas- ure, nor is there any clinical definition of stress that would DESIGN TRADEOFFS support accurate data collection. Interestingly, a second cli- During the user-centered design process, we discovered nician was the one who suggested the stress level scale, and several tradeoffs in the design of the system due to conflict- argued for it from a very patient-centered perspective based ing stakeholder needs and motivations. These tradeoffs re- in psychotherapy. This clinician found that external stressors late to the clinical efficacy of the system, the patient’s pri- play a significant part in the mood of her patients, and it was vacy, sustained use of the system, and other issues. In this useful for her to consider a patient's reported stress level section, we highlight two of the primary tradeoffs we dealt when assessing how that patient was doing. She also be- with during the design of MONARCA. lieved that patients would find it useful to assess their own level of stress, regardless of the fact that they would be in- Clinically driven vs. patient driven strategies terpreting its meaning for themselves in the absence of a If a system has a strong clinical focus – meaning that it clinical definition. The patients tended to agree with her, so adopts only clinically proven treatment strategies – it could although this feature was under debate for several weeks, the miss out on patient-driven approaches that may be helpful to designers opted to keep it in the design because enough par- some patients. In addition, the system may also ignore novel ticipants believed there could be personal value in assessing technological solutions that the clinical field has yet to one's stress. evaluate. Since our system was designed for a clinical con- text, it was important that it adhere to clinical practices so The patients were creative in suggesting strategies based on that it could be evaluated as a valid intervention. In addition, their personal experiences. Knowing what behavioral considering clinical practices was crucial in designing a sys- changes have worked for them in the past, and imagining tem to be viable for adoption and acceptance into a patient's what new strategies might work for them, patients explored treatment, which includes everyday use by the patient and technological solutions unrestrained by considerations of occasional use by the clinician. clinical efficacy. This unrestrained creativity was productive during the design process for two reasons. First, it revealed The clinicians that took part in our design activities shared what would motivate the patients to use the system, which is with us scenarios, anecdotes, and commonalities about the critical to adoption and acceptance. Second, it helped us treatment of their patients. We understood the context we realize which measures, though clinically significant, would were developing the system for by understanding the prac- ultimately fail because they were too intrusive for the patient tices of clinicians with their patients. A recurring theme was to collect, or were not interesting enough to the patient to clinicians' limited resources. This turned into a limitation for motivate collection. the functionality of the system, because if something took Egocentric patient bias vs. clinician generalizations ACKNOWLEDGMENTS Although patients provide valuable insights into the experi- This work has been partially funded by the EU Contract ence of living with and managing bipolar disorder, their in- Number 248545 - MONARCA under the 7th Framework put tends to be egocentric, since their knowledge about the Programme. We would like to thank our participants for disorder mostly comes from their own personal experience their contributions to this project and enthusiasm for the with it. Discussions about the amount and type of data to work. collect were complex due to the different experiences and motivations of the stakeholders: clinicians were interested in REFERENCES data they knew to be relevant for assessment based on clini- 1. Consolvo, S., McDonald, D. W., Toscos, T., Chen, M. cal studies or their own experiences treating patients; and Y., Froehlich, J., Harrison, B., Klasnja, P., LaMarca, patients were interested in data they thought would be useful A., LeGrand, L., Libby, R., Smith, I., Landay, J. A. Ac- to themselves personally for self-reflection. To balance these sometimes opposing interests, designers focused on what tivity Sensing in the Wild: A Field Trial of UbiFit Gar- data would be easy and convenient to collect. Without non- den. Proc. CHI 2008, ACM Press, pp. 1797-1806. intrusive data collection methods, the system will be over- 2. Fogg, B.J., Persuasive technology: using computers to loaded with features and burden the patients, who are re- change what we think and do. 2003, Morgan Kauf- sponsible for collecting the data every day. Here, the design- mann. ers play an important role in keeping in perspective the im- 3. Korotitsch, W. J., and Nelson-Gray, R. O. An Over- plications of collecting different amounts and types of data. view of Self-Monitoring Research in Assessment and Treatment. Psychological Assessment, 1999. 11(4), pp. Patients and clinicians disagreed about how to include cus- 415-425. tomizable personal warning signs, which patients would 4. Lee, H. R., Panont, W. R., Plattenburg, B., de la Croix, personalize and track on a daily basis. In addition to the uni- J., Patharachalam, D., Abowd, G. Asthmon: Empower- versal warning signs that we selected with the help of clini- cians to be applicable to most, if not all, patients, we dis- ing Asthmatic Children’s Self-Management with a Vir- cussed including personal warning signs that each patient tual Pet. Proc CHI 2010, ACM Press, 2010, pp. 3583- could customize based on personal symptoms. Clinicians 3588. argued that there should be as few of these items as possible, 5. Lin, J. J., Mamykina, L., Lindtner, S., Delajoux, G., even stating that one personal warning sign was difficult Strub, H. B. Fish‘n’Steps: Encouraging Physical Activ- enough for patients to attempt to track in their daily life. On ity with an Interactive Computer Game. Proc. Ubicomp the other hand, patients argued that having more flexibility 2006, LNCS, pp. 261-278. would allow them to explore multiple warning signs at once 6. Maitland, J., Chalmers, M. Self-monitoring, self- in order to determine which ones applied to them. One pa- awareness, and self-determination in cardiac rehabilita- tient, who had difficulty understanding her illness and could tion. Proc. CHI 2010, ACM Press, pp. 1213-1222. not identify any of her personal warning signs, asked for a 7. Mamykina, L., Mynatt, E., D., and Kaufman, D., R. lot flexibility because she would have no idea what to track, Investigating Health Management Practices of Indi- so she would need to try many different items. The designers viduals with Diabetes. Proc. CHI 2006, ACM Press, found a solution by suggesting that the feature be limited but pp. 927-936. flexible. The agreed upon solution would allow patients the 8. Oinas-Kukkonen, H. Behavior Change Support Sys- option to include as few as one personal warning sign, but tems: The Next Frontier for Web Science. Web Science no more than three. Those patients who would only be able Conf. 2010. to handle one item at a time could customize the system to 9. Pollak, J. P., Gay, G., Byrne, S., Wagner, E., Retelny, show only one at a time. D., Humphreys, L. It’s Time to Eat! Using Mobile Games to Promote Healthy Eating. Pervasive Comput- CONCLUSION ing, IEEE CS, 2010, pp. 21-27. In the design of a persuasive personal monitoring system 10. Siek, K. A., Connelly, K. H., Rogers, Y., Rohwer, P., for bipolar disorder, we ran into several challenges unique Lambert, D., and Welch, J., L. When do We Eat? An to using persuasive technology for the management of men- Evaluation of Food Items Input into an Electronic Food tal illness. Our findings demonstrate that the design of a Monitoring Application. Proc. Pervasive Health 2006, system for bipolar disorder is quite different from that of IEEE. systems that have been explored for other health purposes 11. Smith, B. K., Frost, J., Albayrak, M., and Sudhakar, R. such as nutrition, physical activity, and chronic physical Integrating glucometers and digital photography as ex- illnesses. In this paper we have highlighted some of the perience capture tools to enhance patient understanding main issues that emerged during our design process, includ- and communication of diabetes selfmanagement prac- ing using a persuasive metaphor, balancing clinical- and tices. PUC, 2007, 11(4), pp. 273-286. patient-centered strategies, and dealing with the biases of patient and clinician participants. Our work revealed major challenges due to the complexity of the illness, stigma sur- rounding the illness, and the often-conflicting needs of cli- nicians and patients. Building Persuasion Profiles in the Wild: Using Mobile Devices as Identifiers. Maurits Kaptein Eindhoven University of Technology / Philips Research Den Dolech 2, 5600MB, the Netherlands. m.c.kaptein@tue.nl ABSTRACT make their choice—the current version does not imple- Tailoring — presenting the right message at the right ment social influence strategies and does not adapt to time — has long been identified as one of the core op- its users. portunities of persuasive systems. In this paper we de- scribe a scenario in which an adaptive persuasive sys- Social Influence Strategies tem which identifies users by the Bluetooth key of their Cialdini [2] shows how small changes to messages—such mobile phone is used to promote energy savings. By as the message on the door—can increase their effective- describing this simplistic system and its possible imple- ness. For example, a message in a hotel room asking mentation we identify several key-criteria of adaptive guests to “reuse their towels” compared to a message persuasive systems. stating “Join your fellow citizens in helping to save the environment” led to a difference in towel re-usage of Author Keywords 28.4% [7]. To structure thes types of messages Cialdini [2] identifies six social influence strategies: Authority, Persuasive Technology, Influence strategies Consensus, Reciprocity, Liking, Scarcity, and Commit- ment. The message in the towel re-usage example im- ACM Classification Keywords plements the Consensus strategy: people act like other H.1.2 User/Machine Systems: Software psychology. people do. A message (e.g.) stating that “The general manager of this hotel requests you to re-use...” would INTRODUCTION implement the Authority strategy. These social influ- ences strategies can easily be used to improve upon the CHI2010 attendees were presented with a choice on effectiveness of the paper-sign. entering the conference hotel: A large revolving door provided access to the hotel while next to it was a slid- The final promise of persuasive technologies however— ing door—some things simply do not fit through a re- adapting influence attempts to individuals—will volving door. With the air conditioning in full opera- require some kind of interactive system. While tion revolving doors are efficient at keeping the heat in. adaptation of persuasive strategies to responses by Sliding doors, however, are not. To help save energy a users is mentioned early on in the literature on paper-sign was put up: “Please take the revolving door”. persuasive technologies Fogg [5, e.g.] we are unaware A brief observation proved the paper-sign to be effec- of any actual implementations. tive just over half the time: 60% of the visitors took the revolving door. This scenario, the “Revolving Door Individual Differences Problem”, offers a framework to describe adaptive per- suasive systems. By further elaborating this scenario There is growing evidence that individuals differ in their and exploring a solution we describe the neccesities and responses to influence strategies: Constructs like Need difficulties that arise when designing adaptive persua- For Cognition [1] predict the response of individuals to sive systems. the usage of social influence strategies. More concretely, Kaptein et al. [9] show that usage of influence strategies for individuals who are low susceptible to these strate- The Promises of Persuasive Technology gies can lead to backfiring: for a portion of participants There are three reasons why employing a persuasive sys- in their study compliance to a request was lower when tem might be more effective than the current paper- the social influence strategy was presented. Next to this sign: (1) Persuasive technologies function as social ac- overall tendency to respond to influence strategies, some tors and can use social influence strategies, (2) they can individuals seem more likely to respond to one specific be context aware, and (3) they can adapt to individual strategy—e.g. an authority argument—while others are users [5, 8]. While the paper-sign is probably located more influenced by implementations of other strategies. at the right place and at the right time—when visitors Cialdini et al. [3] shows that there are sizable and stable individual differences in people’s responses to the com- mitment strategy. Similar results have been obtained when looking at the consensus strategy: Self-reported susceptibility to this strategy highly correlates with be- of approaches the visitor has made to the doors and p havioral responses to this strategy [10]. denotes the probability of success: the probability of taking the revolving door. Given M messages one can These individual differences in susceptibility to differ- compute for each individual, for each message, proba- ent persuasive strategies imply that persuasive systems bility pm = km /nm where km is the number of observed should personalize the way in which they attempt to successes after representation of message m, nm times influence individuals. Such a class of systems, which we to a specific visitor. It makes intuitive sense to present call adaptive persuasive systems, are an unexplored area a visitor with the messages with the highest pm . in that we still need to understand how to model, design and build these systems. This paper takes a concrete For a large number of observations N of one visitor this but simple example that encapsulates the quintessence would make perfect sense. However, this will not inform of this problem to discuss how to address these chal- a decision for a newly observed visitor. For a new visitor lenges. one would present the message m for which pm is max- imized over previously observed visitors1 . Actually— SOLVING THE REVOLVING DOOR PROBLEM? given Stein’s result [4]—for every user a weighted aver- Returning to the revolving door problem, let us consider age of the pm for an individual user and those of other what is involved in implementing an adaptive persuasive users—one where the estimated p�m for an individual is system. We need to (A) identify the visitors entering the “shrunk” toward the population mean—will provide a lobby—minimally by giving each a unique ID, and (B) better estimate than an estimate based on observations measure the effectiveness of a presented message. The of a single visitor alone. E.g., if the authority message is Bluetooth key of visitor’s mobile phone could be used effective 70% of the time over all visitors and only 30% for identification [11]. This will capture around 12% of percent of the time for the specific visitor under consid- the visitors entering the lobby. This same identification eration, the best estimate of the (real) effectiveness of method can also be used to measure the effectiveness of the authority message p�A for this visitor is a weighted each persuasive attempt: One Bluetooth scanner next average of these two. to the revolving door and one next to the sliding door could determine which entrance was used by the current Adapting to Individuals visitor. Based on this knowledge about the visitor and To include both the known effectiveness of a message records of earlier decisions a message implementing the for others, and a specific visitors previous responses to right influence strategy can be selected. In the remain- that same message, into a new estimate of message ef- der of this paper, we focus on the mechanism by which fectiveness, pm , we use a Bayesian approach. A com- these strategies can be selected. mon way of including prior information in a binomial random process is to use the Beta-Binomial model [12]. Suppose we have only two messages to show, one imple- The Beta Beta(α, β) distribution functions as a con- menting the authority strategy—“The general manager jugate prior to the binomial. If we re-parametrize the of this hotel urges you to...” (A)—and one implement- beta distribution as follows ing the consensus strategy—“80% of our visitors always use...etc.” (B). The system then needs a mechanism to π(θ|µ, M ) = Beta(µ, M ) choose the message that is most likely to be effective for the current visitor. It is intuitive that for a new vis- where µ = α+β α and M = α + β, then the expected itor the system should present the message which has value of the distribution is given by: E(θ|µ, M ) = µm . lead to the highest compliance for other, previously ob- In our scenario this represents the expected probability served, visitors. If this message is successful then there of a successful influence attempt by a specific message. is no need to try different messages on subsequent visits. The certainty of this estimated success probability is However, when the selected message is not effective, it represented by: might become attractive to present another message on the next visit. This decision logically depends on the µ(1 − µ) V ar(θ|µ, M ) = σ 2 = initial succes probabilities of the messages under con- M +1 sideration, the variance of effectiveness of messages be- tween visitors, and the number of succes’s or failures ob- After specifying the probability of success µm of mes- served for the current visitor. A collection of estimates sage m and the certainty about this estimate σm 2 we can of the effectiveness of different influence strategies for treat this as our prior expectancy about the effective- an individual is called a Persuasion Profile and can be ness of a specific message and update this expectancy used to select the most-likely-to-be effective message on by multiplying it by the likelihood of the observation(s) a next visit. to obtain the distribution of our posterior expectation: p(θ|k) ∝ l(k|θ)π(θ|µ, M ) Formalizing the Adaptation Problem = Beta(k + M µ, n − k + M (1 − µ)) The probability of a single visitor taking the revolving door on multiple occasions can be regarded a binomial 1 This is assuming the error costs—the effects of presenting random variable B(n, p) where n denotes the number the wrong message—are equal for each message. The newly obtained Beta distribution, B(µ, M ), func- A and 50% to strategy B, (2) susceptible visitors, tions as our probability distribution with a new point- A = 40%, B = 90%, (3) visitors susceptible to message estimate of the effectiveness of the presented message B, A = 10%, B = 90%, and (4) visitors susceptible given by: to message A, A = 90%, B = 10%. Table 1 shows k + Mµ an excerpt of the simulated data. Based on these E(θ|k) = simulated data we first compute our population n+M estimates of message effectiveness for each message: Decision Rule p�A = 0.38, p�B = 0.58. Thus, message B—the The Beta-Binomial model described above allows us to consensus message—was most effective. estimate the effectiveness of message m, include prior knowledge, and update these estimates based on new Type User Occasion Mes. A Mes. B observations. A individual’s persuasion profile would 1 1 1 1 0 0 be a record of both the expected success, µm , and the 2 1 1 2 0 0 certainty, σm 2 of different influence strategies. 3 1 1 3 0 1 .. .. .. .. .. .. To determine which message to present next, one could .. .. .. .. .. .. pick the message which has the highest µm . However, if 1000 4 20 50 1 0 σm2 is large this decision might not be feasible given that Table 1. Overview of the simulated data for the 4 differ- the difference between effectiveness estimates might not ent user groups. Columns Mes. A and Mes. B represent be significant. To address this we can choose to show the success of the influence message at that point in time. the message with the highest estimate when this es- timate is “certain enough”—in the binomial case only Next, we simulate for each visitor, each occurrence at once sufficient observations are obtained. In uncertain the doors. We select the message as specified by our situations we can randomly present one of the H mes- decision rule and record the (simulated) outcome. Next, sages which have the highest estimates out of the total we update our expectancy for the selected message and set of estimates of M messages. This decision rule would iterate through all occurrences. To ensure a flexible avoid presenting each new visitor with only the single starting point for each user we set the prior variance of most effective message when responses to messages are each estimate at the first encounter to be high: σA 2 = variant. σB = 0.05. 2 3 Because the Beta distribution is not necessarily sym- Figure 1 shows for four users—one out of each group— metrical the variance σm 2 provides and inadequate start- in separate panels, the estimated probability of success ing point to compute confidence intervals. This prob- of the two messages (left and right side of each panel). lem can be solved using simulations: By generating a In the upper left panel—representing a general insus- number of draws from the specified Beta distribution ceptible visitor —convergence to message B, whose esti- and computing (e.g.) the 20th and 80th percentiles one mated effect is presented on the right side of the upper can compute a empirical confidence interval. The above left panel, is slow: it takes about 40 observations before described decision rule for M = 2 would then result in: B is consistently estimated to be the “best” message.  1 µ1 > P erc(80)2 With higher compliance and/or larger differences in ef- Mselected = 2 µ2 > P erc(80)1 fectiveness of the two strategies convergence is much  faster. The bottom right of figure 1 shows a user from Rand(1, 2) otherwise the visitors susceptible to message A group. For this Thus, if the estimated effectiveness of a message 1, user after 10 observations strategy A is correctly iden- p�1 = µ1 , is higher than the 80th percentile of message tified as the most successful strategy. 2, P erc(80)2 , the system presents message one.2 If the confidence interval of two messages overlap the system Limitations of the proposed solution could randomly present one of these two. There are a number of drawbacks of the proposed Beta-Binomial solution to create adaptive persuasive Simulations systems. Besides the fact that when the number of To explore the presented Beta-Binomial approach in strategies grows the number of necessary occasions the M = 2 scenario we simulated a dataset presenting for convergence will increase, there are three more different visitors observed at multiple points in time. fundamental issues which are not addressed by this The simulated data describes the message success algorithm. First, while including prior information of two different messages for four different groups based on other users, the algorithm described here of visitors with 20 visitors each on 50 approaches does not use a shrunken estimate on each occasion: to the doors. The four groups represent (1) general After including the initial knowledge of the behavior insusceptible visitors—those that respond favorable to of other visitors the model is specific for an individual only 10% of the message which implement strategy 3 One could estimate this variance based on the between- 2 The 80th percentile is an arbitrary choice. visitor variance. 0 1020304050 0 1020304050 Mobile devices—as used in our scenario—provide a core opportunity to serve as an identifier for adaptive persua- lower + mean + upper lower + mean + upper 0.8 0.8 sive technologies. Currently we are operating a system, 0.6 0.6 like the one described here, in real-life and we would 0.4 0.4 like to share our experiences building and deploying this system during the CHI 2011 PINC workshop. 0.2 0.2 0 1020304050 0 1020304050 References Occasions Occasions [1] Cacioppo, J. T. and Petty, R. E. (1982). The need for cognition. J. of Pers. and Soc. Psy. 0 1020304050 0 1020304050 [2] Cialdini, R. (2001). Influence, Science and Practice. Allyn & Bacon, Boston. lower + mean + upper lower + mean + upper 0.8 0.8 [3] Cialdini, R. B., Trost, M. R., and Newsom, J. T. 0.6 0.6 (1995). Preference for consistency: The development 0.4 0.4 of a valid measure and the discovery of surprising 0.2 0.2 behavioral implications. J. of Pers. and Soc. Psy., 69:318–328. 0 1020304050 0 1020304050 [4] Efron, B. and Morris, C. (1975). Data analysis us- Occasions Occasions ing Stein’s estimator and its generalizations. Journal of the American Statistical Association, 70(350):311– Figure 1. Progression of point estimates of the effects of 319. two messages on four different users (the four panels). Within each panel the left side shows the estimated ef- [5] Fogg, B. J. (2002). Persuasive Technology: Using fect of message A, including in gray its 80% confidence interval, and the right shows the estimates for message Computers to Change What We Think and Do. Mor- B. A horizontal section in the estimates of message A gan Kaufmann. indicates that at that point in time the message B was shown and updated. [6] Gelman, A. and Hill, J. (2007). Data Analysis Using Regression and Multilevel/Hierarchical Models. Cam- bridge University Press. visitor. While this provides quick adaptation there is no opportunity to adapt estimates based on [7] Goldstein, N. J., Cialdini, R. B., and Griskevicius, changing population wise trends. Second, since V. (2008). A room with a viewpoint: Using social the estimates for the effectiveness of the strategies norms to motivate environmental conservation in ho- are treated independently there is no way to of tels. J. of Cons. Res., 35(3):472–482. “borrowing strength” [6] based on correlations with [8] Kaptein, M., Aarts, E. H. L., Ruyter, B. E. R., and other strategies. Both of these concerns could be Markopoulos, P. (2009a). Persuasion in ambient in- addressed using a multilevel approach. Finally, the telligence. Journal of Ambient Intelligence and Hu- proposed model provides no method of including prior manized Computing, 1:43—56. believes about the distribution of visitor profiles over a population. [9] Kaptein, M., Lacroix, J., and Saini, P. (2010). Indi- vidual differences in persuadability in the health pro- CONCLUSIONS motion domain. In Ploug, T., Hasle, P., and Oinas- We identified two core necessities of adaptive persua- Kukkonen, H., editors, Persuasive Technology, pages sive systems: a means to identify users and a means to 94—105. Springer Berlin / Heidelberg. measure effectiveness of persuasive attempts. Further- [10] Kaptein, M. C., de Ruyter, B., Markopoulos, P., more, we highlighted a number of challenges associated and Aarts, E. (2009b). Persuading you: Individual with the design of these systems. The presented Beta- differences in susceptibility to persuasion. In 12th Binomial solution is lightweight and functions well in IFIP TC 13 International Conference - INTERACT, simulations with only two messages. More elaborate pages 24–28, Uppsala, Sweden. ACM Press. algorithms which are (1) variant to changing popula- tion trends,(2) allow for relationships between strate- [11] Kostakos, V. (2008). Using bluetooth to capture gies, and (3) enable us to include prior beliefs about passenger trips on public transport buses. CoRR, user profiles should be explored. Given the current state abs/0806.0874. of social science literature on influence strategies we be- lieve that persuasive technologies should tailor the influ- [12] Wilcox, R. R. (1981). A review of the beta-binomial ence strategies they use to their users. We described one model and its extensions. J. of Educ. Stat., 6:3–32. possible—but limited—implementation of such a sys- tem. This, and other, implementations should now be tested empirically. Nudging Users Towards Privacy on Mobile Devices Rebecca Balebako, Pedro G. Leon, Hazim Almuhimedi, Patrick Gage Kelley, Jonathan Mugan, Alessandro Acquisti, Lorrie Faith Cranor and Norman Sadeh Carnegie Mellon University 5000 Forbes Ave. Pittsburgh, PA 15213 USA ABSTRACT or monitored; other times, while aware of ongoing infor- By allowing individuals to be permanently connected to the mation flows, we do not understand their consequences, or Internet, mobile devices ease the way information can be ac- properly assess their risks. Such challenges are magnified cessed and shared online, but also raise novel privacy chal- in mobile scenarios. Therefore, a mobile device user may lenges for end users. Recent behavioral research on “soft” end up sharing information in a manner that goes against her or “asymmetric” paternalism has begun exploring ways of own long-term self interests. helping people make better decisions in different aspects of their lives. We apply that research to privacy decision mak- In recent years, there has been growing interest in using ing, investigating how soft paternalistic solutions (also known lessons from behavioral economics to influence and ame- as nudges) may be used to counter cognitive biases and ame- liorate decision making in situations where cognitive and liorate privacy-sensitive behavior. We present the theoret- behavioral biases may adversely affect the individual [11, ical background of our research, and highlight current in- 16]. This approach is often referred to as soft or asymmetric dustry solutions and research endeavors that could be classi- paternalism, or with the more popular term “nudges.” Soft fied as nudging interventions. We then describe our ongoing paternalism aims at countering and overcoming those biases, work on embedding soft paternalistic mechanisms in loca- so as to assist individual decision making. Our research aims tion sharing technologies and Twitter privacy agents. at applying and extending lessons from the nascent field of soft paternalism to the field of privacy decision making. This Author Keywords paper presents an overview of our research agenda in this Nudge, Privacy, Security, Location Sharing, Mobile Devices, area. First, we introduce the research exploring cognitive Soft Paternalism and behavioral biases in privacy decision making. Then, we examine current academic studies and industry products that ACM Classification Keywords focus on influencing privacy (and security) decision making, and that therefore may be compared to nudging interven- H.1.2 User/machine systems: Human information process- tions. Finally, we discuss how we are integrating soft pater- ing; J.4 Social and behavioral sciences: psychology nalistic mechanisms in our research on privacy in location sharing applications and social networks. INTRODUCTION As mobile devices and applications become pervasive, pri- vacy risks to their users also grow. The accessibility and FROM HURDLES IN PRIVACY DECISION MAKING ease of use of these devices make it easy to casually broad- TO SOFT PATERNALISM cast personal information at any time, from anywhere, to Findings from behavioral economics and behavioral deci- friends and strangers. Without a doubt, users benefit from sion research have highlighted hurdles in human decision and enjoy such streams of information sharing. However, making that lead, sometimes, to undesirable outcomes. The they also expose themselves to tangible and intangible risks: hurdles are often due to lack of information or insight, cog- from tracking by commercial entities interested in exploit- nitive limitations and biases, or lack of self-control [16]. Be- ing personal information for profit, to surveillance or even cause of those hurdles, individuals may end up making de- stalking by malicious parties. However, it is difficult for in- cisions that they later regret. Those decisions may include dividuals to determine the optimal balance between reveal- (not) saving for retirement, (not) eating well, or smoking ing and hiding personal data. Sometimes we are not even cigarettes [11]. They may also include decisions about pro- aware that information about us is being broadcast, shared, tecting too much, or not enough, personal information [3]. Privacy decisions are complex and often taken in conditions of information asymmetry (that is, individuals may not have full knowledge of how much of their personal information is being gathered, and how it is being used). Furthermore, pri- vacy decision making may be overwhelming: the cognitive costs associated with considering all the ramifications of a Copyright ! c 2011 for the individual papers by the papers’ authors. disclosure may hamper decision making [3]. Finally, cog- Copying permitted only for private and academic purposes. This volume is published and copyrighted by the editors of PINC2011. nitive biases may affect one’s propensity to reveal personal 1 information: for instance, heightened control of one’s per- users are interested in protecting their privacy and may even sonal information may, paradoxically, make the user over- pay for it, if appropriate tools and salient, simple, and com- confident about sharing information [5]. pact privacy information are offered. Specifically, one series of studies explored the impact of making information about Paternalistic policies try to solve decision-making hurdles privacy practices on web sites more accessible to buyers. by mandating decisions for individuals. Such policies are of- The results showed that online customers are more likely to ten heavy-handed and generate externalities [11]. Soft pater- shop online from websites that exhibit more protective pri- nalism, on the other hand, avoids coercion; it seeks to steer vacy policies. Additionally, those customers are willing to users in a direction (believed to be more desirable based on pay a premium for privacy. Furthermore, privacy indicators the user’s own prior judgement, or on external empirical val- displayed at the moment an individual is shopping online idation), without impinging on her autonomy. A soft pater- may have an impact on consumer decisions. In particular, nalistic solution, for instance, would consist of making an they increase the willingness to pay for privacy; however, individual aware of the biases, lack of information, or cog- if the indicator is provided only after the shopper has al- nitive overload that may affect her decision. ready chosen the website from which to buy, the user will not change their already-made decisions. The authors find Nudges are tools of soft-paternalism, and may be used to that timing is essential when trying to help people to protect ameliorate privacy (as well as security) decision making [2]. their privacy [17], [6]. Similarly, another study found that Their application to scenarios involving mobile devices is merely priming Facebook users with questions about their particularly appealing. In the case of insecure communica- online disclosure behavior and the visibility of their Face- tion channels, or covert data collection through a mobile de- book profiles was sufficient to trigger changes in their dis- vice, a nudge may take the form of an alert that informs the closure behavior [13]. Application interface design is also user of the risk. In the case of mobile devices that store sen- important, and should help users notice when changes in sitive information (which could be accessed by strangers if context generate changes in information flows and then help the phone was misplaced), a nudge might discourage users them to maintain their privacy [7]. from storing private data on mobile phones. When informa- tion is being disclosed through a smart-phone, nudges may In the context of location sharing applications, providing provide alerts about the recipients, contexts, or type of data feedback to users whose location has been requested by oth- being shared. ers has been shown to have both positive and negative im- plications [8]. It can prevent excessive requests and hence Many different types of nudging interventions are possible. protect people’s privacy. However, unless appropriate notifi- Some simply consist of informing the user — in which case cations are used, feedback receivers could also be annoyed. they relate to privacy research on informed consent. Some In addition, notifications may inhibit users from requesting focus on making systems simpler to use — in which case, others locations and hence affect system usage. privacy nudges fall into the realm of research on privacy usability. However, other nudges aim at countering spe- PRIVACY NUDGES IN INDUSTRY cific cognitive and behavioral biases, such as neutralizing Examples of industry products or solutions that influence de- the detrimental effects of immediate gratification biases in cision making in regards to privacy (either to better protect privacy decision making [1] by altering the individual’s per- the user, or instead to influence her to reveal more informa- ception of the sequence of costs and benefits associated with tion) take various forms, and some have been applied to mo- revealing sensitive information. bile devices. Some of these solutions may be interpreted as soft paternalistic for privacy protection, in the sense that they The literature on soft paternalism applied to privacy deci- nudge towards privacy. They include privacy/security us- sion making is in its infancy, and therefore extremely scarce. ability solutions, simplifications of privacy settings, or tests However, a number of recent studies and products focus on and delays before one can post information. More frequent, mechanisms that may be categorized as nudges. We present however, are the examples of products and solutions that a brief overview of them in the following sections. nudge individuals to give up even more of their privacy, sur- rendering sensitive information. These include privacy de- PRIVACY NUDGES IN THE LITERATURE faults that are open, lack of usability in privacy settings in- Previous research on the drivers of privacy concerns has demon- terfaces, poorly designed warnings, and other rewards for strated that users’ attitudes towards security and privacy are sharing data or encouraging friends to share data. influenced by numerous factors, including information avail- able, personal beliefs, economic valuations, moral reason- Connections in social applications ing, social values, cognitive biases, and so on. Therefore, Some applications provide information about who can see providing adequate information, making privacy tools more your data, who has seen your data, or how many people can evident, or rewarding and punishing users as they make safer see your data. For instance, Flickr.com, a video and image or riskier decisions are all ways of nudging or influencing sharing website, provides information on each user-owned privacy behavior. The privacy literature offers some exam- picture stating who can see it, followed by a link to edit the ples of these approaches. privacy settings for that picture. This may be a nudge to- wards privacy, as users may decide to share certain photos For example, recent experimental research has shown that with friends, and share other photos with everyone. 2 Social networking sites often show the number of connec- doing so. Sophisticated users may choose to employ soft- tions a user has. These connections may be called follow- ware tools to prevent excess disclosure. For example, the So- ers, friends, or ties. In some cases, connections can have cial Media Sobriety Test, socialmediasobrietytest. access to all the user’s information that is on the applica- com, and Mail Goggles on Gmail googlelabs.com both tion. Twitter and Google Buzz are examples of sites that allow the user to set certain hours of the week when they may prominently show the number of connections. In the case of typically embarrass themselves, such as weekend evenings LinkedIn.com, a job searching social network, the user may after trips to the bar. During these hours, social network prefer to add additional connections, even with people they sites or Gmail may be blocked until the user can complete a don’t know well, in order to grow their job-searching net- dexterity or cognitive test. The user has the option to bypass work. However, by opening their information to more con- the test. Alternatively, a user may set up a warning system nections, they may be compromising their privacy. These ap- if a message is likely to be poorly interpreted. ToneCheck plications may nudge users towards increasing their connec- tonecheck.com scans emails written in Outlook to dis- tions and revealing more information. Indeed, several online cover whether the tone is off-putting, and will ask the user to social networks such as Facebook.com and LinkedIn.com confirm before sending it. This may help discourage users periodically encourage users to add new connections by search- from sending or posting regrettable information. ing the user’s email accounts for email contacts. Other tools may discourage users from posting information Connections such as friends in Facebook and followers in by reminding the user who can see it. NetNanny is a tool Twitter do not set the boundaries for information flow. One’s that parents can user to protect their children online. It will connections may be able to share information with other un- show a message every time a child posts on a social network. intended recipients, or even make it available to the pub- This message reminds the child that her parents will see the lic. In Twitter, for example, re-tweets allow connections to post as well netnanny.com. pass on information without the original sender’s control. In Facebook, default privacy settings usually allow sharing of individual’s information with friends of friends. Therefore, ONGOING WORK WITH MOBILE APPLICATIONS the information provided about the number of connections By studying and understanding the specific biases and user may mislead the user about the privacy of their data and de- actions in regards to mobile applications, we hope to sug- crease the likelihood that the user will take an information- gest and test nudges that will help users make decisions that protective stance. improve their satisfaction and well being. We are moving towards that goal by first understanding users’ needs, prefer- Privacy Settings ences, biases, and limitations about privacy, and second by The privacy settings allowed in an application impact the using that information to evaluate the efficacy of techniques user’s ability to control how their information is shared. Both that exploit biases to improve decision making. As an exam- the default settings and the usability of the settings user in- ple, we are currently pursuing foundational studies with two terface create nudges towards and away from privacy [10, applications developed at Carnegie Mellon: a location shar- 12, 13]. ing application called Locaccino [15] and a privacy agent for Twitter. Some websites make privacy options very simple. For exam- ple, Pandora.com, an online music station, explicitly gives Locaccino is a unique location sharing application that al- users two options regarding their profile page: make pri- lows users to control the conditions under which they make vate or keep public. These clear options allow a user to their location visible to others. This includes controlling the choose without understanding complex details or settings. times and days of the week when different groups of people Conversely, the lack of granularity may encourage users to can see the user’s location as well as the specific locations make everything public. where the user is willing to be visible. For instance, a user can specify rules such as “I’m willing to let my colleagues Several tools provide simple ratings of privacy settings. Pri- see my location but only when I am on company premises vacyCheck,1 and ProfileWatch,2 give Facebook settings a and only 9am-5pm on weekdays.” Research conducted by privacy score. Other services provide a user-friendly layer our group has shown that this level of expressiveness is crit- on the Facebook privacy settings, allowing the user to change ical to capturing the location sharing preferences many peo- the settings. For example, Privacy Defender3 provides a slid- ple have when it comes to disclosing their locations to oth- ing color scale that allows the user to set their Facebook op- ers across a broad range of scenarios [4]—in contrast to the tions as more or less private. These software services ac- much narrower set of scenarios supported by location shar- tively encourage stricter privacy settings. ing applications such as Foursquare. As part of our ongoing research, we are interested in bet- Reduction of Information Disclosure ter understanding how different elements of Locaccino func- If an individual expects she may be likely to post information tionality effectively nudge people in different directions. This she may later regret, software exists to discourage her from includes experimenting with new interface designs as well 1 http://rabidgremlin.com/fbprivacy as new ways of leveraging some of the machine learning 2 http://atherionsecurity.com/idpro.html techniques we have been developing, from exposing differ- 3 http://privacydefender.net ent sets of default privacy personas to users [14] to helping 3 them refine their privacy preferences [9]. We are looking at 6. S. Egelman, J. Tsai, L. F. Cranor, and A. Acquisti. the preferences of like-minded users who have been using Timing is everything?: the effects of timing and the system for a while and trying to use their preferences to placement of online privacy indicators. In Proceedings guide new users. This would have the potential of reducing of the 27th international conference on Human factors regret by giving new users the benefit of the experience ac- in computing systems, CHI ’09, pages 319–328, New quired over time by others. We plan to explore to what extent York, NY, USA, 2009. ACM. such an approach can be made to work and to what extent it seems beneficial. 7. G. Hull, H. R. Lipford, and C. Latulipe. Contextual gaps: Privacy issues on facebook. Ethics and The Twitter privacy agent is an application we are building Information Technology, pages 1–14, 2010. to help Twitter users behave in a more privacy protective 8. L. Jedrzejczyk, B. A. Price, A. K. Bandara, and way. We plan to build tools that will provide nudges that B. Nuseibeh. On the impact of real-time feedback on guide users to restrict their tweets to smaller groups of fol- users’ behaviour in mobile location-sharing lowers or discourage them from sending tweets from mobile applications. In Proceedings of the Sixth Symposium on devices that they may later regret. We plan to empirically Usable Privacy and Security, SOUPS ’10, pages test the impact of these nudges on user behavior. We will 14:1–12, New York, NY, USA, 2010. ACM. also examine whether fine-grained privacy controls result in more or less data sharing. 9. P. Kelley, P. Hankes Drielsma, N. Sadeh, and L. Cranor. User-controllable learning of security and privacy We expect our work on nudges in behavioral advertising, so- policies. In Proceedings of the 1st ACM workshop on cial networks, and location sharing to be effective for im- Workshop on AISec, pages 11–18. ACM, 2008. proving privacy decisions on mobile devices. We further 10. Y.-L. Lai and K. L. Hui. Internet opt-in and opt-out: hope our soft-paternalistic approach to have a broader im- investigating the roles of frames, defaults and privacy pact, guiding the development of tools and methods that as- concerns. In Proceedings of the 2006 ACM SIGMIS sist users in privacy and security decision making. CPR conference on computer personnel research, pages 253–263, 2006. ACKNOWLEDGMENTS This material is based upon work supported by the National 11. G. F. Loewenstein and E. C. Haisley. The Foundations Science Foundation under Grant CNS-1012763 (Nudging of Positive and Normative Economics, chapter 9: The Users Towards Privacy), and by Google under a Focused Re- Economist as Therapist: Methodological Ramifications search Award on Privacy Nudges. This work has also been of ‘Light’ Paternalism. Oxford University Press, 2008. supported by NSF grants CNS-0627513, CNS-0905562, and 12. W. Mackay. Triggers and barriers to customizing by CyLab at Carnegie Mellon under grants DAAD19-02-1- software. In Proceedings of the SIGCHI conference on 0389 and W911NF-09-1-0273 from the Army Research Of- Human factors in computing systems: Reaching fice. Additional support has been provided by the IWT SBO through technology, pages 153–160. ACM, 1991. project on Security and Privacy in Online Social Networks (SPION), Nokia, France Telecom, and the CMU/Portugal 13. Ralph Gross and Alessandro Acquisti. Information Information and Communication Technologies Institute. Revelation and Privacy in Online Social Networks. In Workshop on Privacy in the Electronic Society (WPES), REFERENCES pages 71–80, 2005. 1. A. Acquisti. Privacy in electronic commerce and the 14. R. Ravichandran, M. Benisch, P. Kelley, and N. Sadeh. economics of immediate gratification. In Proceedings Capturing social networking privacy preferences: can of the ACM Conference on Electronic Commerce (EC default policies help alleviate tradeoffs between ’04), pages 21–29, 2004. expressiveness and user burden? In Proceedings of the 2. A. Acquisti. Nudging privacy: The behavioral 5th Symposium on Usable Privacy and Security, economics of personal information. Security & Privacy, page 1. ACM, 2009. IEEE, 7(6):82–85, 2009. 15. N. Sadeh, J. Hong, L. Cranor, I. Fette, P. Kelley, 3. A. Acquisti and J. Grossklags. Privacy and rationality M. Prabaker, and J. Rao. Understanding and capturing in individual decision making. Security & Privacy, peoples privacy policies in a mobile social networking IEEE, 3(1):26–33, 2005. application. Personal and Ubiquitous Computing, 13(6):401–412, 2009. 4. M. Benisch, P. Kelley, N. Sadeh, and L. Cranor. Capturing Location-Privacy Preferences: Quantifying 16. R. Thaler and C. Sunstein. Nudge: Improving decisions Accuracy and User-Burden Tradeoffs. Journal of about health, wealth, and happiness. Yale Univ Pr, Personal and Ubiquitous Computing, 2011. 2008. 5. L. Brandimarte, A. Acquisti, and G. Loewenstein. 17. J. Y. Tsai, S. Egelman, L. Cranor, and A. Acquisti. The Misplaced Confidences: Privacy and the Control effect of online privacy information on purchasing Paradox. Technical report, Mimeo, Carnegie Mellon behavior: An experimental study. Information Systems University, 2010. Research, In press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`4a49749Y4:8ZD% =0< & "0/', & .E=03"% & M/0=22K4 & 5 & .'0?/<", & >=2+"& $"0?/'0?&=0<&/<+=2&>=2+"?% \E" & I5J) & (,=*/03 & ?+33"?$? & 0"G & =##,'=.E"? & $' & F"E=>/',& .E=03"% & )/=2"? & ?/[ & ?#"./(/.& ;/'(#"%@0AB#"C+ ?$,=$"3/"?&$E=$&=,"&+?"<&FK&#"'#2"&$'&/0(2+"0."&"=.E&'$E",^& N$E/.?4&>=2+"?4&B=2+"&-"0?/$/>"&!"?/304 &I5J)4&#",?+=?/>"& ?'*"&G',H &/0&#",?+=?/>" &$".E0'2'3K&E=? &<,=G0 &'0&$E"?"& $".E0'2'3K4&*'F/2"&#E'0"? ?$,=$"3/"?&C"%3%4&X98497ZD%,,-$./%&.'*"?&(,'*&$E"&.'0."#$&'(& +01(+% , "2+0(3%+3$2%P & $E=$ & '+, & "0>/,'0*"0$ & ?$,+.$+,"? & $E"& ;:5%:.-++)4)*-')#1%@0AB#"C+ .E'/."? & =>=/2=F2" & $' & +?4 & =0< & *',"'>",4 & $E=$ & $E"," & /? & =0& O7%*%&50(',*=$/'0&/0$",(=."?&=0<&#,"?"0$=$/'0&C"%3%4&O)5DP& /0">/$=F2K&=&<"(=+2$&'#$/'0^&<"?/30",? &.=0&.=,"(+22K&?"2".$& Q/?."22=0"'+?%&R%S%:%&)'*#+$",?&=0<&?'./"$KP&-'./=2&/??+"?% $E=$&<"(=+2$&$'&3"0$2K&0+<3"&$'&$E"&<"?/,"<&F"E=>/',%&\'&$E"& F"?$ & '( & *K & H0'G2"<3"4 & $E/? & /<"= & E=? & 0'$ & F""0 & ?$+0"$+ "[#2/./$2K&/0&$E"&<'*=/0&'(&#",?+=?/>"&$".E0'2'3K4&=2$E'+3E& O+*=0&M=.$',?% <"(=+2$ & '#$/'0? & .=0 & F" & .'0?/<","< & ?+33"?$/'0? & (,'* & $E"& .'*#+$",&X984&#%9:aZ%&M/0=22K4&+1%2+(1)&"0?+,"?&=&#=,$/.+2=,& 89>?E2F:>8E9 F"E=>/',&$E,'+3E&$E,"=$?&=0<&(',."&X:Z%&50&E/?&(,=*/03&'(&$E"& B=2+"? & +0<",2K & #",?+=?/'0% & 5*#2/./$ & /0 & #",?+=?/'0 & /? & $E"& (/"2< & '( & #",?+=?/>" & $".E0'2'3K4 & M'33 & "[#2/./$2K & "[.2+<"<& @+<3*"0$ & $E=$ & '0" & F"E=>/', & ', & '+$.'*" & /? & F"$$", & $E=0& .'",./'0&=?&"&$".E0'2'3K&.'**+0/$K% O","4&5&,"?#'0<&$'&$E"&.=22&(',&.'0$,/F+$/'0?&=F'+$&"$E/.=2& /??+"? & $' & $E" & G',H?E'# & VI5J)P & I",?+=?/'04 & 50(2+"0."4& ,;HF&%=&9=8>8,&%2&=8D9 J+<3"4 & =0< & )'",./'0 & $E,'+3E & *'F/2" & <">/."?W & XYZ% & & QK& B=2+" & -"0?/$/>" & !"?/30 & CB-!D & /? & = & $E"',"$/.=2 & =0<& =##,'=.E&$'&"$E/.?&/?&(,'*&$E"&#",?#".$/>"&'(&E+*=0&!"#$%& *"$E'<'2'3/.=2 & (,=*"G',H & /0$"0<"< & $' & E"2# & <"?/30",?& '#=,$/.+2=,2K & *',=2 & >=2+"? & ?+.E & =? & (=/,0"??4 & =+$'0'*K4& =..'+0$ & (', & E+*=0 & >=2+"? & /0 & = & #,/0./#2"< & =0<& #,/>=.K4 & =0< & E+*=0 & G"2(=,"% & 6($", & " & G=K & $E,'+3E'+$ & $E" & <"?/30 & #,'."?? & X9:Z%& +0<",?$=0=2+"?&'(&*',=2&/*#',$&=0<&$E+?&?#"=H?&$'& B=2+" & -"0?/$/>" & !"?/30 & $E"',K & =0< & *"$E'<'2'3K & X9:Z% & 5& "$E/.=2&.'0.",0? &/0&$".E0'2'3K&<"?/30%&R"K&B-!&("=$+,"?& $E"0&?E'G&E'G&"$E/.=2&3'=2?4&#,/0./#2"?4&=0<&#,'F2"*?&'(& /0.2+<"&/$?&.'*#,"E"0?/>"&=$$"0$/'0&$'&?$=H"E'2<",?&=0<&/$?& I5J)&.=0&F"&(,=*"<&/0&$",*?&'(&>=2+"?&=0<&>=2+"&$"0?/'0?% $,/#=,$/$"&*"$E'<'2'3K&X9:Z% 789:%;92%7&?=F;=8,&%>&:G9EHEDI B-! & <"*=0" & $".E0'2'3K& ?$=H"E'2<",?P & 0'$ & '02K&$E'?" & GE'& +?" &$E" &$".E0'2'3K & F+$& .'**+0/$K% &I",?+=?/>" & $".E0'2'3K & .'0.",0? & $E" & +?" & '(& $E'?" & GE' & =," & =((".$"< & FK & /$? & +?"% & B-! & =2?' & ?+33"?$?& $".E0'2'3/.=22K & 3"0",=$"< & ', & *"+20",=F2"&?$=H"E'2<",?%&50&$E"&.=?"& '(&*'F/2"&#",?+=?/'04&$E/?&*=K&/0.2+<"&$""0?&X749aZ4&;%-%& )'#K,/3E$ & e & :899 & (', & $E" & /0/<+=2 & #=#",? & FK & $E" & #=#",?b & =+$E',?%& c=$/0'?&=0<&F2=.H?&X:SZ4&=0<&6(,/.=0?&X:7Z&GE'&,"2K&E"=>/2K& )'#K/03&&#",*/$$"<&'02K&(',&#,/>=$"&=0<&=.=<"*/.&#+,#'?"?%&\E/?&>'2+*"& '0&$E"/,&*'F/2"&#E'0"?&(',&.'**+0/.=$/'0&=0<&d"F&=.."??4& /?&#+F2/?E"<&=0<&.'#K,/3E$"<&FK&$E"&&"" & F,'="% & -+.E & 3,'+#? & ?E'+2< & F" & 0"/$E", & 0"32".$"< & 0',& =F+?"<&/0&$E"&<"?/30&'(&*'F/2"&I5J)&$".E0'2'3/"?% B-!b?&*"$E'<'2'3K&/0.',#',=$"?&$".E0/.=24&"*#/,/.=24&=0<& !"#!$%&'()*+#,$-&+.(&+"#-/*0$!1#+!()*+#,$-&+.(&+"#-*!"#!$2#* Q#7$2-"# * #"&$-< * R6$3 * =$)+$,$ * &1(& * A!"$2!+"#D * +- * ()3(4-* 1"3 * -4-&$5 * 6$(&'2$- * -'%%"2& * "2 * '#7$25+#$ * %(2&+!')(2* '#_'-&+6+$7< *-+#!$ *+& *-$$5- *&1(& * #"*-"!+$&4*!"')7* 6'#!&+"#* ,()'$-/ * 85%+2+!() * +#,$-&+.(&+"#- * (772$-- * -&(9$1")7$2* 3+&1"'&*-"5$*('&1"2+\$7*'-$-S*ABD/*`(3*.+,$-*.",$2#5$#&-* !"#!$%&+"#- * "6 * ,()'$- * (#7 * &1$ * 1'5(# * 2$-%"#-$ * &" * &1$* (*)+5+&$7*('&1"2+&4*&"*'-$*!"$2!+"#/*H$%)"4+#.*!"5%'&$2-* (2&+6(!&/ * :"2 * $;(5%)$< * (# * $5%+2+!() * -&'74 * 5+.1& * %2$-$#&* 6"2 * )(3 * $#6"2!$5$#& * 1(- * %"&$#&+() * =$#$6+&- * (#7 * !"-&-< * (-* %(2&+!+%(#&-*3+&1*-!$#(2+"-*7$-+.#$7*&"*%'-1*&1$*="'#7(2+$-* !"5%'&$2-*)(!9*&1$*!"#&$;&'()*(3(2$#$--*(#7*_'7.5$#&*"6*(* "6 *!$2&(+# * ,()'$- * +#* &1$* 7$-+.# *"6 *%$2-'(-+,$ * &$!1#")".4<* 1'5(#*=$+#./*^1+)$*&1+-*5+.1&*=$*-$$#*(-*(#*"%%"2&'#+&4*&"* -+5+)(2*&"*>(.$*(#7*?2(4@-*2$!$#&*-&'74*"6*$&1+!()*2$-%"#-$-* '-$*!"5%'&$2-*6"2*6(+2*$#6"2!$5$#&<*'#!)"'7$7*=4*1'5(#* &" * %$2-'(-+,$ * &$!1#")".+$- * ABCD/ * :+#())4< * !"#!$%&'()* =+(-$-<*+&*!(#*=$*-'2%2+-+#.)4*7+66+!')&*&"*%2"7'!$*!"5%'&$2* +#,$-&+.(&+"#-*$;%)"2$*&1$*,()'$-*(&*1(#7*(#7*&1$*&$#-+"#-* -4-&$5- * &1(& * (2$ * 62$$ * "6 * =+(-/ * H$-+.#$2- * !(# * $#!"7$* =$&3$$#*&1$5/*01$*2$5(+#7$2*"6*&1+-*%(%$2*!"5%2+-$-*-'!1* '#!"#-!+"'- * =+(-$- * +# * &1$ * -4-&$5< * '#+#&$#7$7 * =+(-$- * !(#* (*!"#!$%&'()*+#,$-&+.(&+"#/ $5$2.$*+#*'-$<*(#7*#$3*=+(-$-*!(#*(2+-$*(-*(*-4-&$5*+-*'-$7* +#*#$3*!"#&$;&-*ATPD/*:'2&1$2<*()&1"'.1*1'5(#-*"6&$#*=)(5$* 8)-$31$2$<*E*1(,$*(2.'$7*&1(&*FGH*+-*3$))I-'+&$7*&"*(772$--* !"5%'&$2- * 6"2 * =(7 * "'&!"5$-< * !"5%'&$2- * )(!9 * 5"2()* $&1+!() * !"#!$2#- * +# * %$2-'(-+,$ * &$!1#")".4 * AJD/ * 01$ * FGH* (!!"'#&(=+)+&4 * ATC(.$*(#7*?2(4*2$%"2&*&1(&*-&'74*%(2&+!+%(#&-* E*!"#-+7$2*&12$$*!)(--$-*"6*,()'$-*2$)(&$7*&"*5"=+)$*>ELM* 6"'#7*!"$2!+,$*R-1"!9*&(!&+!-S*&"*=$*(!!$%&(=)$*+6*+&*3(-*&1$* &$!1#")".+$-N * ,()'$- * #$!$--(2+)4 * +5%)+!(&$7 * =4 * &1$ * >ELM* %$2-"#@-*"3#*!1"+!$*&"*'-$*&1$*-4-&$5*ABCD/*01+-*2$&'2#-*'-* (%%2"(!1< * ,()'$- * +5%)+!(&$7 * =4 * %(2&+!')(2 * 5$&1"7- * "6* &"*"'2*$(2)+$2 *7$6+#+&+"#*"6*('&"#"54N *R%$"%)$@- *(=+)+&4*&"* %2"5"&+#.*=$1(,+"2*!1(#.$<*(#7*,()'$-*+5%)+!(&$7*=4*&1$* 7$!+7$< *%)(#<*(#7*(!& *+#*3(4- *&1(&*&1$4*=$)+$,$*3+))*1$)%* 7$-+2$7 * $#7-/ * E * ()-" * !"#-+7$2 * ,()'$ * &$#-+"#- * (#7 * 1"3* &1$5*&"*(!1+$,$*&1$+2*."()-S*ATBD/* 7+66$2$#!$- * +# * ,()'$- * 5(4 * #"#$&1$)$-- * )$(7 * &" * &1$ * -(5$* =$1(,+"2/*E*3+))*72(3*%(2&+!')(2)4*'%"#*54*$(2)+$2*(#()4-+-* K"3$,$2<*'-$2-*#$$7*+#6"25(&+"#*&"**(--$--*&1$*-'+&(=+)+&4*"6* AJD * (#7 * O$27+!1$,-94 * (#7 * L$'$#-!13(#7$2@- * $&1+!()* &1$*-4-&$5*&"*&1$+2*."()-/*O$27+!1$,-94*(#7*L$'$#-!1(#7$2* %2+#!+%)$-*6"2*%$2-'(-+,$*&$!1#")".4*APD/ -&(&$*&3"*$&1+!()*%2+#!+%)$-*2$)(&$7*&"*7+-!)"-'2$N* FEV 01$ * !2$(&"2- * "6 * ( * %$2-'(-+,$ * &$!1#")".4 * -1"')7* +,-&./01-2&34&5*'6 7+-!)"-$ * &1$+2 * 5"&+,(&+"#-< * 5$&1"7-< * (#7 * +#&$#7$7* 01$ * >ELM * $#7$(,"2 * +- * +#&+5(&$)4 * &+$7 * &" * &1$ * ,()'$ * "6* "'&!"5$-< * $;!$%& * 31$# * -'!1 * 7+-!)"-'2$ * 3"')7* ('&"#"54 * AJD/ * Q'&"#"54 * R2$6$2- * &" * %$"%)$@- * (=+)+&4 * &"* -+.#+6+!(#&)4*'#7$25+#$*(#*"&1$23+-$*$&1+!()*."()/ 7$!+7$< *%)(#<*(#7*(!& *+#*3(4- *&1(&*&1$4*=$)+$,$*3+))*1$)%* &1$5*&"*(!1+$,$*&1$+2*."()-S*ATBD/*>ELM*&$!1#")".4*&13(2&-* FEEV>$2-'(-+,$*&$!1#")".+$-*5'-&*#"&*5+-+#6"25*+#*"27$2* ('&"#"54*31$#*+&*+-*'-$7*&"*.$&*%$"%)$*&"*7"*&1+#.-*&1(&*(2$* &"*(!1+$,$*&1$+2*%$2-'(-+,$*$#7/*APD (.(+#-& *&1$+2*"3#*."()- *U$/./< *&"*3(-&$*5"#$4*"#*'-$)$--* W#$*-&$%*=$4"#7*&1+-*+-*&1$*,()'$*"6*&+5'(2"3)*'+-"+0 )AJDN* %2"7'!&-V/*O'&*>ELM*!(#*()-"*'%1")7*('&"#"54<*31$#*+&*+-* &1(&*%$"%)$*-1"')7*#"&*"#)4*=$*+#6"25$7<*='&*-1"')7*1(,$*(#* 7$%)"4$7 * +# * -'%%"2& * "6 * (# * +#7+,+7'()@- * ."()- * U$/./< * &"* $;%)+!+&*"%%"2&'#+&4*&"*"66$2*"2*3+&11")7*!"#-$#&/*:2+$75(#<* =$!"5$ * 5"2$ * (!&+,$V/ * E#7$$7< * W+#(-I?'99"#$# * 1(-* K"3$< * (#7* :$)&$# *+7$#&+6+$7* -+;* !"5%"#$#&- * "6*+#6"25$7* %2"%"-$7*&1$*7$,$)"%5$#&*"6*&1$"24*(#7*5$&1"7")".4*6"2* !"#-$#&*ATTDN**'+-"+0*!"5%2+-$-*%'4.+0$(&+"--7)*'2/"0"+*"7 ) !"#$%&'()*#$+,")-.//'(0)-1-0"2-*(-*(#*+5%"2&(#&*7+2$!&+"#* $,(""2"+0<*(#7*2&+&2$4)3&-0($*0&'+<*31+)$*6"2*!"#-$#&*&"*=$* 6"2*&1$*>$2-'(-+,$*0$!1#")".4*!"55'#+&4*ATXD/ &+5'(2"3*2$a'+2$-*#"&*"#)4*3&-*4'-.("<*(-*O$27+!1$,-94*(#7* K"3 * 7" * &1$ * >ELM * (%%2"(!1$- * -&(#7 * 3+&1 * 2$-%$!& * &"* L$'$#-!1(#7$2 * $;1"2&< * ='& * ()-" **'2/("#"+-&'+/ * >ELM* ('&"#"54Y *Q- * :".. * #"&$-< * %$2-'(-+"# * +5%)+$- * ,")'#&(24* &$!1#")".+$-*&1(&*(2$*'#7$25+#$7*=4*+#6"25$7*!"#-$#&b6"2* !1(#.$ * ATC< * %/TZD/ * E#6)'$#!$< * &""< * -'..$-&- * ,")'#&(2+#$--[* $;(5%)$<*?(%&$+#*(#7*8!9)$-@-*%$2-'(-+"#*%2"6+)+#.*ATZDb +#7$$7< *M+()7+#+ *-1"3- *1"3*&"*2$!".#+\$ *(#7*!+2!'5,$#&* 7$-$2,$*1$+.1&$#$7*-!2'&+#4*2$.(27)$--*"6*&1$*(!!$%&(=+)+&4* +#6)'$#!$*(&&$5%&-*A]D/*M$#&2()*&"*&1$*+.3,"*+-*+-*&1$*+7$(*"6* "6*&1$+2*$#7-/*Q-*c+!1()-9+*ATd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b"4&:",,%OLVVMP?% &'(%;"64!-?%C422'-!%)$")%:0%$*5$!-)%;"64!%*-%)$!%-"&!)0%'&% FF?8(*!3:",9%D?9%c'+!9%Y?9%",3%8!6)!,9%Z?%@,&'(:!3%T',-!,)%*,% :0% #$*63(!,? % @& % @ % "3'2) % " % :'/*6! % "226*#")*', % )' % $!62 % :!% )$!%U'\*66"%D('+-!(.%@:26!:!,)*,5%]"64!NC!,-*)*;!%Y!-*5,?%@,% ";'*3%-2!!3*,59%",3%*)%-$'+-%:!%2*#)4(!-%'&%2'6"(%/!"(-9%@% B*-/&&@+,L$.-%.')&.PQ').9"#"++.:,'&*,"'+-,"!.8-,%&*&,/&.-, . +*66%/!%42-!)?%D!#"4-!%*)%#$"66!,5!- %:0%;"64!-9%@%-!!%)$!% ?($'&2.?/+&,/&$%OLVVLP? "226*#")*',%"-%"%)$(!")%)'%:0%"4)',':09%",3%@%!E2!(*!,#!% FL?8(*!3:",9%D?9%b"$,9%K?c?%^(?9%",3%D'(,*,59%7?%]"64!%C!,-*)*;!% 2-0#$'6'5*#"6 % (!"#)",#! % G6!"3*,5 % :! % )' % 3(*;! % !;!,% Y!-*5,%",3%*,&'(:")*',%-0-)!:-.%B$(!!%#"-!%-)43*!-?%@,%K?%d$",5% &"-)!(?%@,-)!"39%@%-$'463%/!%(!:*,3!3%'&%:0%;"64!%'&%-"&!)0? ",3%Y?%e"66!))"%O!3-?PK.%902",R8-2A0'&*.:,'&*"/'+-,.",@. C","L&2&,'.:,%-*2"'+-,.?($'&2$D.S-0,@"'+-,$4%U?%Z?%C$"(2!% @%-!!%)+'%"22('"#$!-%)'%"33(!--*,5%*,3*;*34"6%4-!(-H%;"64!-?% OLVVWP9%M[=NMRL? 8*(-)9%3!-*5,!(-H%;"64!%#'::*):!,)-%-$'463%/!%:"3!%#6!"(% FM?8(*!3:",9%D?9%",3%S*--!,/"4:9%c?%D*"-%*,%#':24)!(%-0-)!:-?% )$('45$%/(",3*,5%",3%)$!%*,&'(:!3%#',-!,)%2('#!--9%-'%)$")% 78C.>E:?%GT9%MOFAAWP9%MMVNM[R? 4-!(-%#",%:"I!%*,&'(:!3%#$'*#!-?%C!#',39%*,)!(&"#!-%-4#$% "-%B'339%J'5!(-9%",3%K"0,!H-%*,&'(:")*;!%5('#!(0%-$'22*,5% F[?^'$,-',9%Ye?9%",3%U46;!09%^?U?%7##'4,)"/*6*)0%",3%#':24)!(% 3!#*-*',%-0-)!:-?%878C.P=K.FL%OFAAQP9%Q=NW[? #"() % % -$'463 % /! % )"*6'("/6!? % B$!0 % -$'463 % 42$'63 % 4-!(% "4)',':0%/0%"66'+*,5%4-!(-%)'%#$''-!%+$*#$%*,&'(:")*',% FQ?b"2)!*,9%U?%",3%Z#I6!-9%Y?%C!6!#)*,5%!&&!#)*;!%:!",-%)'%",0% ":',5 % ;"64!N6"3!, % '2)*',- % O!?5?9 % -4-)"*,"/*6*)09% !,3?%@,%B*-/4.B5H?F7?:I5.;? T"2)'6'50%U!3*"%OLVVRP9%FQNF=? FR?U*#$"6-I*9%^?%Z)$*#"6%3",5!(-%'&%:'/*6!%2!(-4"-*',?%@,%D?^?% !"#!$%&'"# 8'55%",3%Y?%Z#I6!-%O!3-?P9%C-3+!&.B&*$0"$+-,K%C)",&'(3% 7))!,)*', % )' % ;"64!- % :"0 % #',)(*/4)! % ,') % ',60 % )'% T"2)'6'50%U!3*"%OLVVRP9%FMRNF[L?? 4,3!(-)",3*,5%!)$*#"6%*--4!-%'&%:'/*6!%K@ST%)!#$,'6'50G F=?a*,"-Nb4II',!,9%c?%D!$";*'(%#$",5!%-422'()%-0-)!:-.%7% /(*,5*,5 % "))!,)*', % )' % #',#!(,- % /!0',3 % 2(*;"#0 % ",3% (!-!"(#$%:'3!6%",3%"5!,3"?%@,%B*-/4.B5H?F7?:I5.;<<=9%`STC% 3*-#6'-4(!G/4) % "6-' % )' % *,#(!"-*,5 % )$!*( % -#'2! % ",3% QVMM%OLVV=P9%FW[NFRW? !&&!#)*;!,!--G)$!*(%2'+!(%)'%3'%5''3%*,%)$!%+'(63?%84()$!(% +'(I%-$'463%#6"(*&0%('6!%'&%)$!-!%;"64!-%)$('45$%!:2*(*#"6% FA?a*,"-Nb4II',!,9%c?%",3%c"(g4:""9%U?%%7%-0-)!:")*#% &(":!+'(I%&'(%3!-*5,*,5%",3%!;"64")*,5%2!(-4"-*;!%-0-)!:-?%@,% ",3%)!#$,*#"6%*,;!-)*5")*',-%'&%K@ST%)!#$,'6'50? B*-/4.B5H?F7?:I5.;<<=9%`STC%WFMR%OLVFVP9%[NF[? ()*)()#!)& LV?K"5!9%J?Z?%",3%b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c?%",3%C4,-)!*,9%T?J?%U0@L&D.:2A*-1+,L.O&/+$+-,$. M?D!(3*#$!;-I09%Y?%",3%S!4!,-#$+",3!(9%Z?%B'+"(3%",%!)$*#-%'&% "3-0'.9&"!')K.M&"!')K.",@.9"AA+,&$$K%h"6!%f,*;!(-*)0%K(!--% 2!(-4"-*;!%)!#$,'6'50?%878C%[L9%Q%OU"0%FAAAP9%QFNQ=? OLVV=P? [?T*"63*,*9%J?D?%:,%!0&,/&D.>)&.?/+&,/&.-%.B&*$0"$+-,9%T'66*,-9% LM?B'339%K?U?9%J'5!(-9%h?9%",3%K"0,!9%C?^?%S435*,5%)$!%#"()%*,% (!;*-!3%!3?%OFAA=P? )$!%-42!(:"(I!)1%c'+%:4#$%*-%!,'45$%*,&'(:")*',%&'(%&''3% -$'22!(-i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`H. Y10;. '<. +",. ,$,2+-#21$. %,9#2,:. 1))-')-#1+#'0. O;. +,,01L,-:H. /0. 1%%#+#'0. +'. )-'9#%#0L. 2'0+-#O*+#0L. +'. +",. -#:,. #0. %'(,:+#2. ,0,-L;. 1-,. *:,%. 10%. ),-:'01$#:,%. 10%. 1LL-,L1+,. ,0,-L;. *:1L,. %1+1. 10%. :'(,+#(,:. 'P0,%. O;. +,,01L,-:H. ?,:,1-2". #0. +",. 8C. "1:. ,%*21+#'01$. #0<'-(1+#'0. #0. 1. (,10#0L<*$. P1;7. +",. ('O#$,. #0%#21+,%.+"1+.aMb.'<.+,,01L,-:."1%.1.N^7.(*:#2.:;:+,(.'-. +,2"0'$'L#,:. P,. 1-,. %,:#L0#0L. P#$$. 1$$'P. <'-. 1))-')-#1+#'0. )"'0,.#0.+",#-.-''(:7.P#+".+P'.+"#-%:."19#0L.1$$.+"-,,._@c`H. #0. U2''$V. P1;:. +",-,O;. <':+,-#0L. 10. 12+#9,. 2'((*0#+;. '<. K.:,)1-1+,.:+*%;.-,)'-+,%.+"1+.d]].:*-9,;,%.+,,01L,-:.1L,%. +,,01L,-:.P",-,.#+.#:.2''$.+'.(#0#(#:,.,0,-L;.*:,H.. @c.+'@a.P,-,.2'$$,2+#9,$;.P1:+#0L.,0'*L".,0,-L;.+'.)'P,-. d7e]A. :2"''$:. 10%. 1. +"#-%. '<. +",. ,0,-L;. O,#0L. *:,%. P1:. 1. "()*+,!-./0+,12! %#-,2+.2'0:,f*,02,.'<.U:+10%O;V.O,"19#'*-._@`H. B0,-L;.8:,7.N,,01L,-:7.Y'O#$,.N,2"0'$'L#,:.. [*-. P'-I. '0. +",. U+1I#0L. '0. +",. +,,01L,-:V. )-'T,2+:. 4PPPH(1%d0-LH'-L6. P#$$. ,0L1L,. ;'*0L. ),')$,. 41L,%. @A& @a@6. #0. -,%*2#0L. +",#-. 'P0. ),-:'01$. ,0,-L;. *:,. 10%. (1I,. !"#$%&'()*+*,-..*/"%*)(0*&12&3&2456*#5#0%7*8$*)(0*#5#0%79*54)("%7:* !"#$&1'*#0%;&))02*"16$*/"%*#%&35)0*512*5<520;&<*#4%#"707:*=(&7*3"64;0*&7* ............................................................. #486&7(02*512*<"#$%&'()02*8$*)(0*02&)"%7*"/*>?@!,-..:* @ ./+.#:.)-'O$,(1+#2.-*00#0L.:+*%#,:.P#+".+,,01L,-:.<-'(.1L,. . @d&@M. #0. :2"''$:. #0. +",. 8C. P",0. +",;. O,L#0. )-,)1-#0L. <'-. 3!QB. ,W1(#01+#'0:7. P,. +",-,<'-,. 1$:'. +1-L,+. )-,&+,,0:. +'. ,0:*-,. P,. 210. P'-I. +",. :1(,. :*OT,2+:. <'-. +",. c. ;,1-. %*-1+#'0.'<.+",.)-'T,2+H. !"#$%$&'( )*+,-'#( $,( +%%$%./'#( %"0+1/#( ','1-2( .#'( %*+%( 0$33( /'7$,'( %*'( +%%1$6.%'#( "7( )""3( +,/( 6'$,-( N)""3O( $#( "7%',( 3+#%( %*1".-*( +/.3%*""/4( 5*$#( 0$33( 6'( +)*$'&'/( %*1".-*( %*'( 'B%1'8'32($8!"1%+,%4(L1"8(#%./$'#("7('B$#%$,-(3$%'1+%.1'(+,/( )1'+%$",( "7( 8"6$3'( %')*,"3"-$'#( 9)+33'/( :;<=( :+>'( ;( $,$%$+3(/'#$-,(#'##$",#(0'(*+&'($/',%$7$'/(%*1''(8+$,(3'&'3#( <$77'1',)'?(%"('/.)+%'(%'',#(+6".%()*"$)'#(%*'2()+,(8+>'(%"( "7()""3,'##($,(%*'()",%'B%("7(%'',+-'14(5*'(7$1#%A(+,/('+#$'#%( 1'/.)'( ','1-2( .#'( +,/( !1"&$/'( 7''/6+)>( ",( ','1-2( .#+-'4( %"(+)*$'&'A($#(%*'()""3,'##(+##")$+%'/(0$%*(*+&$,-(/'#$1+63'( :"6$3'( %')*,"3"-$'#( 0$33( +3#"( 6'( .#'/( %"( -+%*'1( ','1-2( %*$,-#( %*+%( "%*'1#( +#!$1'( %"( 9#.)*( +#( 3+%'#%( %')*,"3"-2( "1( .#+-'( $,7"18+%$",( %*1".-*( #'37@1'!"1%( +,/( #',#$,-( )3"%*'#?A( %*'( #')",/( $#( )""3,'##( +##")$+%'/( 0$%*( +)%$",#( "1( %')*,"3"-$'#A( 7"1( 'B+8!3'( /'%')%$,-( 0*$)*( %1+,#!"1%( +)%$&$%$'#( %*+%( -+$,( 1')"-,$%$",( 71"8( !''1#4( 5*'( %*$1/A( +,/( 8'%*"/#( +1'( 6'$,-( .#'/A( $,( +//$%$",( %"( 8"1'( .#.+3( ','1-2( 8"#%( )*+33',-$,-( %"( +)*$'&'A( $#( %*+%( "7( *"3$#%$)+332( N6'$,-O( 8",$%"1$,-( $,( %*'( *"8'4( 5*'#'( 8"6$3'( %')*,"3"-$'#( 0$33( )""3( +,/( "7%',( 1'#.3%#( $,( 6'$,-( +/8$1'/( +,/( "7%',( /'7'11'/( 8+>'( !'1#",+3$#'/( +,/( +--1'-+%'/( ','1-2( .#+-'( %"( 62( !''1#4( T%( $#( %*$#( 3+%%'1( )+%'-"12( %*+%( 8+1>'%$,-( $,7"18+%$",( +))'##$63'( $,( 8'+,$,-7.3( 0+2#( %"( ',+63'( )"8!+,$'#( "7%',( #''>( %"( $,73.',)'( +#A( 62( &$1%.'A( !1"/.)%#( )"8!+1$#",( +,/( )"8!'%$%$",( 6'%0'',( !''1#( %"( 7"#%'1( +,( %*'2( +##")$+%'( 0$%*( 6')"8'( )""3( +,/( +1'( %*',( /'#$1+63'( 62( +)%$&'( )"88.,$%2( "7( %'',+-'1#( $,%'1'#%'/( $,( 1'/.)$,-( "%*'1#4(( ','1-2( .#'4( 5*'( 8"6$3'( /'&$)'#( 0$33( +3#"( 8+>'( +&+$3+63'( C""3,'##($#(-','1+332(+()*+33',-$,-(!1"!'1%2(%"(/'#$-,($,%"( #%+%.#( .!/+%'#( +6".%( $,/$&$/.+3( ','1-2( .#'( +,/( !1"&$/'( +( !1"/.)%( +,/( 0$%*$,( %*'( %*1''( %2!'#( "7( )""3( 0'( *+&'( %+1-'%'/('/.)+%$",+3(8+%'1$+3(%"(.#'1#4( $/',%$72(%*'(#')",/(9+##")$+%'/(0$%*(+)%$",#("1(+)%$&$%$'#(%*+%( 5*'( !"!.3+1$%2( "7( *"8'( ','1-2( 8",$%"1#( 71"8( -+$,(1')"-,$%$",(71"8(!''1#?(+#(8"#%(3$>'32(%"(6'(+)*$'&+63'( 8+,.7+)%.1'1#( #.)*( +#( ;3'1%:'( 90004+3'1%8'4)"8?( +,/( 0$%*$,(%*'(!1"U')%4(5*$#(0$33(6'()".!3'/A(%"(+(3'##'1('B%',%A( C.11',%( C"#%( 90004).11',%)"#%4)"8?A( +,/( #'1&$)'#( #.)*( +#( 0$%*( %*'( 7$1#%( %2!'( 9*+&$,-( /'#$1+63'( %*$,-#?4( 5*'( 8"6$3'( D""-3'( E"0'1:'%'1( 90004-""-3'4)"8F!"0'18'%'1?( 8'+,( +!!3$)+%$",#( 0'( +1'( /'#$-,$,-( 0$33( ',+63'( %'',( .#'1#( %"( %*+%( 8",$%"1$,-( '3')%1$)$%2( .#'( $,( %*'( *"8'( $#( $,'B!',#$&'( 8",$%"1( %*'$1( ','1-2( .#+-'( 9)+!%.1'/( .#$,-( 3"0( )"#%( +,/( .,)"8!3$)+%'/4( ( G"0'&'1A( '&',( 0*'1'( %*'( ','1-2( 8",$%"1$,-(%')*,"3"-2($,(%*'(*"8'(+,/(8"6$3'(/'&$)'#?(+,/( )",#.8!%$",( $,7"18+%$",( $#( !1"&$/'/A( %*'( &$#.+3$H+%$",( "7( /'&$#'( %*'$1( "0,( 0+2#( "7( 1'!1'#',%$,-A( #*+1$,-( +,/( %*$#( $,7"18+%$",( "7%',( )+,,"%( '+#$32( )"11'3+%'/( 0$%*( )"8!+1$,-( %*'( $,7"18+%$",4( 5*1".-*( ".1( !+1%$)$!+%"12( )",#.8!%$",(6'*+&$".1(IJK4(5*$#($#('$%*'1(6')+.#'(%*'(.,$%#( +!!1"+)*( 0'( 0$33( )1'+%'( +!!3$)+%$",#( %*+%( #.!!"1%( "7( 8'+#.1'8',%( +1'( 1'3+%$&'32( 8'+,$,-3'##( %"( .#'1#( "1( %*'( ).#%"8$#+%$",( +,/( 'B!1'##$&$%2( #.)*( %*+%( %*'2( )+,( 6'( $,7"18+%$",( $#( $11'3'&+,%( %"( %*'$1( $,%'1'#%#( 9'4-4( )"#%( +!!1"!1$+%'/( $,( +( N)""3O( 0+2#( 62( %'',+-'1#4( V,'( #)',+1$"( $,7"18+%$",(8+2(,"%(8'+,(8.)*( 7"1(%'',+-'1#(0*"(/"(,"%( 0'( ',&$#+-'( $#( %'',+-'1#( 6'$,-( +63'( %"( )"8!+1'( ','1-2( !+2( %*'( 6$33#?4( L.1%*'18"1'( %*'1'( $#( +( )"88",( 3+)>( "7( .#+-'(.#$,-(+(8'%1$)(+,/(&$#.+3$H+%$",(%*'2(%*'8#'3&'#(*+&'( +0+1','##(+6".%(%*'(+8".,%("7(','1-2()",#.8'/(62(/'&$)'( /'&$#'/( +,/F"1( +/"!%'/( 90*$)*( $#( +##.8'/( %"( 6'( )""3( $,( #$,(%*'(*"8'(+,/(','1-2@#+&$,-("!%$",#(IMK4( %*'$1(!''1(-1".!?(%"(/'%'18$,'(0*"($#(%*'(0$,,'1(9.#$,-(%*'( 3'+#%(','1-2?(+,/(0*"($#(%*'(3"#'1(9.#$,-(%*'(8"#%(','1-2?4( 5*'( >'2( )*+33',-'( "7( %*$#( 0"1>( $#( ,"%( ",32( /'#$-,( *$-*32( T,(%*$#('B+8!3'($%($#(3$>'32(%*+%()"8!'%$%$",(0$33(',)".1+-'( .#+63'( 8"6$3'( %')*,"3"-$'#( %"( !1"&$/'( +))'##( %"( ','1-2( )",#$/'1+%$",( +,/( 1'/.)%$",( "7( ','1-2( .#'4( ;3#"A( %*'#'( .#+-'( $,7"18+%$",( !1'#',%'/( $,( +( 8'+,$,-7.3( 0+2( 7"1( 8"%$&+%$",#( 0$33( 6'( ,.1%.1'/( +,/( #.!!"1%'/( %*1".-*( .#'( "7( %'',+-'1#A(6.%(+3#"(%"(',#.1'(%*+%(%*'#'(%')*,"3"-$'#(%*+%(+1'( %*'( :;<( %')*,"3"-$'#( %*+%( 0$33( !1"&$/'( $,7"18+%$",( +,/( #.77$)$',%32( N)""3O( %*+%( %*'2( +1'( /'#$1+63'( +,/( #")$+332( +/&$)'( ",( 8'+#.1'#( %"( 1'/.)'( ','1-2( .#'A( $,( +//$%$",( %"( +))'!%+63'4( G+&$,-( )1'+%'/( /'&$)'#( %*+%( +1'( .#'/( +,/( !1"&$/$,-($,7"18+%$",(",().11',%(','1-2(.#+-'4(((( .,/'1#%""/( 62( %*'( %'',#A( 0'( 0$33( %*',( .#'( +( 1+,-'( "7( +!!1"+)*'#( %"( 3'+/( %"( 3",-@%'18( 6'*+&$".1+3( 8"/$7$)+%$",4( %"&().()$/#"0.*12$$ 5"( +)*$'&'( %*'#'( -"+3#( +( !+1%$)$!+%"12( +!!1"+)*( $#( .#'/( 5*$#( 0"1>( 0$33( 6.$3/( .!",( %*'( 55:( 8"/'3( "7( 6'*+&$".1( 0*$)*( $,&"3&'#( 0"1>$,-( /$1')%32( 0$%*( 2".,-( !'"!3'( $,( )*+,-'( IQQK( +,/( 0$33( 3$,>( $,%"( 8"1'( 1')',%( 0"1>( ",( #)*""3#(%"()+112(".%(/'#$-,(+,/('&+3.+%$",(#%./$'#4(( '8"%$",+3(',-+-'8',%(7"1(6'*+&$".1()*+,-'(9ISKAIWK?(0*$)*( /'8",#%1+%'#( %*+%( 6'*+&$".1( )*+,-'( $#( 8"1'( '77')%$&'A( P'( ,"0( /$#).##( %*'( )*+33',-'( "7( N)""3O( +,/( ".1( $,$%$+3( ',-+-$,-( +,/( !1"/.)%$&'( $7( %*'1'( $#( +,( '8"%$",+3( 7$,/$,-#A( %*'( !'1#.+#$&'( +#!')%#( "7( %*'( %')*,"3"-$'#( 0'( +1'( ',-+-'8',%(6'%0'',(%*'(%')*,"3"-2(+,/(%*'(.#'14(5*.#(".1( )1'+%$,-A(+,/(%*'(%')*,"3"-$)+3()*+33',-'#(0'(7+)'4(P'(-$&'( 8"6$3'( %')*,"3"-$'#( *+&'( %"( #.!!"1%( +!!1"!1$+%$",( $,( )""3( +,("&'1&$'0("7(1'3+%'/(0"1>(+,/(7$,$#*(0$%*(+(/$#).##$",("7( 0+2#( 6.%( +3#"( *+&'( %"( 6'( /'#$-,'/( $,( +( 0+2( %*+%( %'',+-'1#( >'2($##.'#(+,/(7.%.1'(0"1>4( )+,('+#$32(1'3+%'(%"(%*'8(+,/($,(+(8+,,'1(%*+%($#(1')'!%$&'(%"( !"#$%"&''#()#$*+$,%**'-$ '8"%$",+3($,%'1!1'%+%$",(9$4'4(%*'(#2#%'8#(%*'8#'3&'#(/"(,"%( P*$3'( %*'( 8'+,$,-( "7( N)""3O( *+#( 6'',( )",#$/'1'/A( 7"1( ,')'##+1$32(*+&'(%"(6'('8"%$",+3A(%*'2(U.#%(*+&'(%"(6'(+63'( 'B+8!3'( IQRK( IQSKA( %*'1'( $#( ,"%( +( #$,-3'( .,$&'1#+332( %"( +!!'+1( '8"%$",+3( '&',( $7( %*+%( '77')%( $#( !1"U')%'/( 62( %*'( +!!3$)+63'( /'7$,$%$",4( C""3( 8+2( 6'( +,%$@#")$+3( "1( $33$)$%A( $%( .#'1?4( 8+2(6'('B!',#$&'(+,/(*$-*32(/'#$1+63'A("1($%(8+2(1'!1'#',%( 5*$#(!1"U')%(0"1>#(71"8(%*'(+##.8!%$",(%*+%(%'',+-'1#(*+&'( $,,"&+%$",( 9+,/( %*'#'( +1'( ,"%( 8.%.+332( 'B)3.#$&'( %*'( !"%',%$+3( %"( 8+>'( #$-,$7$)+,%( )*+,-'#( %"( ','1-2( .#+-'4( )+%'-"1$'#?4( T,( %*'( )+#'( "7( %'',+-'1#A( !''1( -1".!#( "7%',( X"%(",32()+,()*+,-$,-(%'',+-'(6'*+&$".1(+77')%(%*'$1(3",-@ !"#$%&"#'()*+%,'"-%.,!%!/"0%*#"%*+'(%1)%*%&('1!1()%2/"#"%!/"0% B!/"%),$."#%7"&")71)9%()%!/"%3('!%(6%!/"%7"813"C-%.,!%*6!"#% 3*)%,'"%4&"'!"#%&(2"#5%!(%*66"3!%!/"%*!!1!,7"'%*)7%."/*81(,#'% 1)1!1*+% !#1*+'% 2"% 21'/% !(% (&")% !/"% '0'!"$% ,&% !(% *'% 217"#% (6% !/"1#% &*#")!'-% '1.+1)9'% *)7% 6#1")7':% ;'% $*)0% !"")*9"#'% &*#!131&*!1()%*'%&(''1.+":%%% /*8"% *% 9#"*!"#% *$(,)!% (6% +"1',#"% !1$"% !/*)% *7,+!'-% !/1'% 3*)% ?)% (#7"#% !(% ")*.+"% ")"#90% $()1!(#1)9% 1)% !/"% /($"% !/"% #"',+!%1)%!/"%,'"%(6%$*)0%/19/%")"#90%!"3/)(+(91"'%',3/%*'% &#(="3!% 21++% ,'"% +(2% 3('!% XY"% 1!% *8*1+*.+"% 1!'% 21++% ."% &#(817"7% 6(#% *% !"")*9"#'5% &*!!"#)'% (6% ")"#90% ,'"% 1)% (#7"#% !(% ,)7"#'!*)7% ),$."#%(6%,'"#%43/*$&1()'5%*)7%1'%/(&"7%!/*!%!/"0%21++%/"+&% !/"1#% ."/*81(,#'% *)7% $(!18*!1()'% $(#"-% *)7% /(2% !/"0% $*0% ")3(,#*9"% (!/"#% !"")'% !(% &"'!"#% &*#")!'% 1)!(% .,01)9% !/"% >1!% ."%1)6+,")3"7:% '(%!/"0%3*)%!*>"%&*#!:%% % 1)% !(% ;%&*#!131&*!(#0%7"'19)%&#(3"''%!(%3#"*!"%D;EF%*)7%D;EG% 1)6+,")3"% #"7,3!1()% 1)% "+"3!#13*+% *)7% !#*)'&(#!% ")"#90% ,'":% 21++% ."% ()9(1)9% 21!/% '"8"#*+% 7166"#")!% 9#(,&'% 1)% !"")*9"#'% ?)1!1*++0-% '!(#1"'% (6% ")"#90% ,'*9"% 21++% ."% 3(++"3!"7% 6#($% 1)%'3/((+'%*3#(''%!/"%QR%1)8(+8"7:%%L"%"@&"3!%!(%7"8"+(&%*% !"")*9"#'% 1)% !/"% '3/((+'% 21!/% 2/13/% 2"% *#"% 2(#>1)9:% "0%$(.1+"%&#(7,3!'-%()"%6(#% ,'*9"%/*.1!'%*'%!/"%&#(="3!%&#(9#"''"':%%%%% FHIFJ% 0"*#% (+7'% BD;EFC% *)7% *)(!/"#% 6(#% FJIFK% 0"*#% (+7'% +"(,!"-$.'+/$ BD;EGC:%;%>"0%1'',"%1'%'"+"3!1)9%2/13/%$(.1+"%&+*!6(#$B'C% ;% '$*++% ),$."#% (6% 2"*#*.+"Z$(.1+"% #"'"*#3/% &#(!(!0&"'% !(% !*#9"!% *)7% !/1'% 1'% +1>"+0% !(% ."% !/"% !#*7"I(66% 1)% !"#$'% (6% *)7% &#(7,3!'% /*8"% ."")% 7"8"+(&"7% 6(#% 3()8"01)9% ")"#90% 6"*!,#"'% &#(817"7% *)7% 7"813"% &(&,+*#1!0:% L/1+"% '$*#!% ,'*9"% 1)6(#$*!1():% Q.1[#"")% \T]% ,'"7% $(.1+"% &/()"'% *'% &/()"'% ',3/% *'% M+*3>."##0% 7"813"'-% 1N/()"'-% ;)7#(17% *$.1")!% 71'&+*0'% !(% 918"% ,'"#% 6""7.*3>% ()% !/"1#% 7"813"'-% *)7% O(>1*% P0$.1*)% /*)7'"!'% *#"% 1)3#"*'1)9+0% !#*)'&(#!*!1()%."/*81(,#':%?!%#"+1"7%()%!/"%2"*#*.+"%'")'1)9% &#"8*+")!% *$()9% *7,+!'% 1)% !/"% QR-% !/"1#% /19/% 3('!% (6!")% ,)1!-%[PD%3"++%'19)*+'%*)7%!/"%&*#!131&*)!'5%$*),*+%1)&,!%!(% $*>"'% !/"$% 1)*33"''1.+"% !(% 0(,)9"#% !"")*9"#'% 21!/% +1!!+"% 7"!"3!% !#*)'&(#!*!1()% $(7":% ^3(I6#1")7+0% !#*)'&(#!*!1()% '&")71)9%&(2"#%*)7%#"'!#13!"7%!(%4&*0% *'%0(,%9(5%B3()!#*3!% ."/*81(,#'-%',3/%*'%3*#&((+1)9-%!*>1)9%.,'%*)7%303+1)9%"!3:% 6#""C% 3*++% &+*)':% S#($% (,#% 3,##")!% '!,71"'% 1)% '3/((+'% 21!/% 2"#"% ")3(,#*9"7% *)7% '/(2)% *'% #"2*#7'% ()% !/"% *$.1")!% 0"*#%T'%B*9"%FUIFFC%*)7%0"*#%FU'%B*9"%FHIFVC%1!%1'%*&&*#")!% 71'&+*0:%^)"#90_16"%\H]%1'%*%&"#8*'18"%'")'1)9%*)7%6""7.*3>% !/*!% !/"% &/()"'% !/"0% (2)% *#"% .*'13% 7"813"'% (6!")% /*)7"7% '0'!"$:% ;% '"#8"#% 2*'% 3())"3!"7% 21#"+"''+0% 81*% *% .*'"% 7(2)%6#($%*)%(+7"#%'1.+1)9%(#%&*#")!:%L"%/*8"%*+'(%6(,)7% '!*!1()%!(%")"#90%'")'(#'%2/13/%#"&(#!"7%!/"1#%")"#90%,'*9"% !/*!% !/"% 3/1+7#")% 1)% (,#% 1)1!1*+% '!,71"'% /*8"% +1!!+"% 1)!"#"'!% 1)% "8"#0% 3(,&+"% $1),!"':% "%3*++':%P"8"#*+%0"*#% ,&()%,'"#`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eB) Q!""#>) Q7) 3'#) T"31">) P7>) M'(3(&'() MC<"!&"'%"+) 4&,-) ,-"/) 31&(') 4&,-) ,-") /.!") -.1&+,&%) 3'#) 1.'(5,"!/) "'"!(*) M/.,&.'31) f&!,$31) =-"!3<&+,+7) &') 8',"!'3,&.'31) H"+&(') !"#$%,&.')3&/+).2),-&+)4.!67)) 3'#)M'(3(30&1&,*)Q.'2"!"'%")g)U.!#&QE8>)?+1.>)JVVW7) !"#$%##"&'( @]B) H"<3!,/"',) .2) M'"!(*) h) Q1&/3,") Q-3'(") RJVVZS) 8')%!"3,&'()"'"!(*)+39&'()#"9&%"+)2.!),""'3("!+)4")%.'+&#"!) M'"!(*)Q.'+$/<,&.')&'),-")c'&,"#)i&'(#./7)P",!&"9"#) ,-3,) :%..1;) &+) 3) <.4"!2$1) 23%,.!) &') /.,&93,&'() 3#.<,&.') 3'#) 2!./) 3<) 4") '""#) -,,) <"!+.'31) (.31) 3,&.'+Y"%$6)Y"%$673+) 3'#) /36") (..#) $+") .2) "'"!(*) 3') @WB) [3/"+) N&"!%">) H&3'") [7) O%-&3'.>) 3'#) M!&%) N3$1.+>) &',"(!31) <3!,) .2) ,-") ("'"!31) #&+%.$!+") 0",4""') ,""'+7) ?$!) aE./">) -30&,+>) 3'#) "'"!(*X) "C3/&'&'() #./"+,&%) 3<) &',"!3%,&.'+) 3'#) "'"!(*) %.'+$/<,&.'>b) &') QE8dIVX) 0$,) ,.) %!"3,") ,"%-'.1.(&"+) ,-3,) %3') 0") <"!+.'31&+"#) 3'#) N!.%""#&'(+) .2) ,-") JA,-) &',"!'3,&.'31) %.'2"!"'%") .') 3<) U"4) j.!6>) Uj>) =-") ",-&%+) .2) <"!+$3+&.') @AB) &') ,-") %.',"C,) .2) ,-&+) )<<7)IZA]5IZZe>)GQF>)JVIV7) 3!")+./"4-3,)/&,&(3,"#)&'),-3,),-")!"#$%,&.').2)"'"!(*)$+") @\B) [.') k!."-1&%->) =343''3) H&113-$',>) N!"#!3() i13+'D3>) &+) 3%%"<,"#) ,.) 0") 3) '"%"++3!*) /.9"7) E.4"9"!>) ,-&+) 4&11) 0") ["''&2"!) F3'6.>) O$''*) Q.'+.19.>) T"9"!1*) E3!!&+.'>) 3') 3!"3) 4") "C<1.!") &') #",3&1) 3+) ,-") #"9"1.)ac0&(!""'X)&'9"+,&(3,&'()3)/.0&1") FGHI) 3'#) FGHJ) b) &') N!.%""#&'(+) .2) ,-") J\,-) &',"!'3,&.'31) "C<1.!") &'&,&31) +%"'3!&.+) 3'#) !"K$&!"/"',+) 2.!) FGHI) 3'#) %.'2"!"'%") .') E$/3') 23%,.!+) &') %./<$,&'() +*+,"/+>) 2$!,-"!) $'#"!+,3'#) ,-") 6"*) %-3!3%,"!&+,&%+) .2) %..1) 2.!) U"4) j.!6>) Uj>) cOG>) QE8) dVZ>) <<7IVe^5IV]J>) GQF>) #&22"!"',) (!.$<+7) 8') <3!311"1) 4") 3!") 31+.) "C<1.!&'() 3'#) JVVZ7) ,"%-'&%31) <.++&0&1&,&"+) 3'#) "C<"!,) ,"%-'&%31) #"+&('+>) 3'#) @AB) H3'&"1) T"!#&%-"9+6*) 3'#) M!&6) U"$"'+%-43'#"!7) IZZZ7) &'9"+,&(3,&'() ,""'3("!+;) %$!!"',) 3,,&,$#"+) ,.43!#+) "'"!(*) =.43!#) 3') ",-&%+) .2) <"!+$3+&9") ,"%-'.1.(*7) Q.//$'7) $+"7)L")-.<"),-3,),-")'.9"1),-"/"+)&'),-&+)4.!6)4&11)0").2) GQF)eJ>)])RF3*S>)]I5]A>)IZZZ7)) &',"!"+,) ,.) ,-") <3!,&%&<3',+) .2) ,-") 4.!6+-.<) 3'#) ) H39&#) =!3'>) 3'#) P"0"%%3) M7) _!&',"!>) a_",,&'() ,.) (!""'X) $'#"!+,3'#&'() !"+.$!%") %.'+$/<,&.') )$*'&+,-!./-'0#( &') ,-") -./">b) &') N!.%""#&'(+) .2) ,-") IV,-) &',"!'3,&.'31) L") 4.$1#) 1&6") ,.) ,-3'6) MNOPQ) R$'#"!) ,-") H&(&,31) %.'2"!"'%") .') c0&K$&,.$+) %./<$,&'(>) U"4) j.!6>) Uj>) M%.'./*)3'#)M'"!(*)N!.(!3//"S)2.!)2$'#&'(),-&+)4.!67)))! cOG>))c0&Q./<)dVA>)<<7)JeJ5J]I>)GQF>)JVVA7) 1-2-1-'$-#( @IVB) ?dH.''"11>) i7G7) 3'#) H7`7) L3!#1.4>) G) ,-".!*) .2) ,-") @IB) TTQ) U"4+) RJVVWS) =""'3("!+) 3!") :+,3'#0*) 9&113&'+;7) .!&(&'+) .2) %..1'"++>) G#93'%"+) &') Q.'+$/"!) P"+"3!%->) P",!&"9"#) 2!./) J\>))<3("+7)I^)5)IA>)JVVV7) -,,) [7?7>) U.!%!.++>) [7Q7) 3'#) H&%1"/"',">) Q7Q7) Q-3'(&'()k.!)_..#7)G9.')T..6+>)U"4)j.!6>)IZZe7) @JB) T"31">) P7) 3'#) Q!""#>) Q7) G22"%,&9") 8',"!3%,&.'X) E.4) "/.,&.'31) 3("',+) 322"%,) $+"!+7) 8',"!'3,&.'31) [.$!'31) .2) @IJB) =3<<>) G7) 3'#) O7) T&!#>) O.%&31) /3!6",&'() 3'#) ,-") E$/3'5Q./<$,"!)O,$#&"+>)W\)RZS7)\]]5\\W>)JVVZ7) /"3'&'() .2) %..1>) O.%&31) F3!6",&'() l$3!,"!1*>) Ie>) I) <3("+7)IA)5)JZ>)JVVA7) @^B))Q-!&+,."!) G7) TD.!6+6.(>) _&$1&.) [3%$%%&>) `$%&3'.) _3/0"!&'&>) =3,$) U&"/&'"'>=.<&) F&66.13>) Q3!&') @I^B)=-")O1""<)Q.$'%&1)RJVV\S)[$'6)O1"") 3'#) F3++&/.) T"!,.'%&'&>) aM'"!(*1&2"X) =-!"3,) ,.) =""'3("!+7) P",!&"9"#) 2!./) <"!93+&9") "'"!(*) 343!"'"++) 2.!) -.$+"-.1#+>b) &') -,,) c0&%./<) dIV>) JVIIS7) <<7)^WI5^WJ>)JVIV7) ) ) ) ) ) ) Nudging People at Work and Other Third-Party Locations Max L. Wilson1, Derek Foster2, Shaun Lawson2, Simon Eddison1 1 Future Interaction Technology Lab Lincoln Social Computing Research Centre College of Science School of Computer Science Swansea University, UK University of Lincoln, UK m.l.wilson@swansea.ac.uk, simon.d.eddison@gmail.com {defoster,slawson}@lincoln.ac.uk ABSTRACT A recent report [1] has indicated that if the 17 million UK Nudging people towards positive behaviour change is an workers, who regularly use a desktop PC, powered it off at important issue recognised by academia, individuals, and night this would reduce CO2 emissions by 1.3 million tons even governments. Although much research has been - the equivalent of removing 245,000 cars from the road. published in this area, little has focused on non-domestic Similarly, if a UK business with 10,000 computers leaves environments such as the workplace. It is widely reported them on all night for one year, it will cost £168,000 that changing individual behaviour of employees can make ($220,000) and emit 828 tonnes of CO2. The same report, a significant contribution to sustainable resource however, suggested that at least three in ten workers in the consumption. This position paper focuses on the unique UK do not always power off their PC overnight. Further, aspects that make nudging consumption behaviour in third- many more machines are in use or provide services 24 party environments like the workplace a very different hours a day, all year round. problem to that of nudging in people’s domestic and private lives. Several studies are discussed to provide context as As an example in our own context, Figure 1 compares the well as evidence towards our position. electricity usage at the University of Lincoln campus for the first week in December in 2009 and 2010. There are two Author Keywords compelling features of Figure 1 that characterise the typical Persuasion, Nudge, Work, Ownership, Sustainability, energy consumption of a workplace. First, the graph clearly Behaviour Change. shows how little energy the university uses at the weekend. Second, this period in 2010 coincided with severe weather ACM Classification Keywords that meant that many staff members were unable to travel to H5.m. Information interfaces and presentation (e.g., HCI): the campus. The dramatic reduction in energy consumption Miscellaneous. can be clearly seen in the first 3 days of the graph and highlights that people can have a significant impact on General Terms consumption at work, as well as in their own personal Theory, Human Factors, Design. environments. INTRODUCTION The HCI community has recently shown a great deal of interest in the development of interactive systems that facilitate behaviour change for sustainability. Much of this research has exploited ideas recently re-popularised by Thaler and Sunstein [10] in that individuals can be ‘nudged’ to make better lifestyle decisions, given the right information and the environment in which to do so. Much of this work has focused on how individuals might improve their own private and domestic lifestyle, behaviour, and sustainable resource consumption; however such work has Figure 1 Campus electricity usage December 2009/10 rarely taken account of the fact that people spend a significant amount of their waking hours at work where Despite environmental concerns now playing an established they also contribute towards resource consumption. role in the public sector, as well as the corporate and business agenda, there is still much to gain by exploring new ways of persuading people to adopt positive energy usage behaviour. The first and obvious research question is: Copyright © 2011 for the individual papers by the papers' authors. Do domestic PINC (Persuasion, Influence, Nudge & Copying permitted only for private and academic purposes. This volume is Coercion) methods simply translate to workplace and other published and copyrighted by the editors of PINC2011. third-party environments? In this position statement we review initial evidence that they do not, and discuss the environmentally friendly is, as Stebbins called it, a Serious reasons why. We propose a framework for thinking about Leisure, where people work hard at achieving their goals. Nudge methods in different contexts, and discuss our future Installing home technology is often a temporary project, work in this area. and can be seen as Project Leisure, where people take behaviour change to be a new task. The aim of much RELATED WORK nudging research, however, is to be embedded in people’s Thaler and Sunstein [10] have recently re-popularised the Casual Leisure, so that good consumption is encouraged interest in the idea of Nudge, where the right environments simply and unobtrusively within our lives. These forms of and the right information delivered at the right time can leisure, however, are very different from our work lives, encourage people to adapt and improve their behaviours. which are goal-oriented, formalised, and externally driven. Much research has focused on directly improving one’s own behaviour, whether it be reminders to exercise, or to EARLY EXPERIMENTAL FINDINGS notably reduce energy consumption. Research into simple Study 1 – Water Consumption in the Work Place home energy monitors [3], for example, suggests that pay- One early finding in this space was from Kuznetsov and as-you-go meters typically reduce consumption by only 3%, Paulos [7] who anecdotally saw unexpected results in a while those that focus on reducing their payments often work environment, and so proceeded to focus on domestic reduce their consumption by 0-10%. Having an in-house scenarios. Their anecdotal findings saw consumption monitor that provides instant feedback has been shown to increase – double in fact. reduce consumption by between 5 and 15%. Other prototype systems, such as Kuznetsov and Paulos’s One of our recent studies in Swansea University, UK, domestic ambient light display [7] successfully encouraged focused directly on this surprising issue. We created a series people to reduce their water consumption, by visualising of feedback installations, and installed them in a shared better or worse consumption to their previous average use. work-place kitchen. Like the work by Kuznetsov and Paulos, the installations used a Phidget microphone to track Other research typically provides anonymous averages from water flow through the pipes. The installations were a group or community to a user, so that the user can see supported by informational posters, which included a link their own behaviour or consumption in the context of to a website to provide feedback. Otherwise, we remained others. In previous work [5], we reduced domestic energy as un-intrusive as possible in order to record normal usage consumption through a carefully designed mixture of online as closely as possible. After recording baseline average social media and home energy monitors. Our findings readings, we first recreated the ambient light display suggested that the use of energy feedback delivered in a provided used by Kuznetsov and Paulos, which: glowed social context significantly reduced consumption when green with less-than-average consumption; glowed yellow compared to energy feedback without a social context. We 10% either side of the mean; and glowed red thereafter. have also shown similar results in a personal fitness/activity domain [4]. A related approach involves facilitating ‘friendly’ competitive behaviour; for instance it has already been shown that the work environment affords powerful opportunities for facilitating such behaviour – for instance Siero et al [8] demonstrated that when a group of employees received information not only about their own energy usage, but also about that of a ‘competing’ group of employees from the same company but a different workplace, they significantly altered their energy usage behaviour compared to a situation in which they only Figure 2: Our ambient-light installation received information about their own usage. Three further displays were installed in subsequent weeks. Despite the success of the work by Siero et al some thirteen The first used similar measures, in respect to average years ago, little research since has explored energy consumption, to create competitive gaming-style text- behaviour interventions based on competition between oriented messages on an LED display, such as: “You’re employees. Therefore, a key question for Nudge researchers beating most people” and “Sorry, you lost”. The second going forward is how do differences between the work and display converted the light system into a series of audible domestic leisurely sides of life affect the potential of beeps. The final display tried a different tack altogether, by behaviour change interventions? Also, what theoretical simply providing environmental information relating to grounding can we draw upon to begin to explore any their water consumption, such as the average amount of differences? Stebbins [9] introduced a seminal framework water available to people in the third world on a daily basis. for understanding people’s leisure time. For some, being Initially, as per the prior anecdotal evidence, the ambient To mitigate the absence of financial motivation in light display did double the average consumption of water employees and to develop workplace energy metaphors, we during the 2 weeks it was displayed. In comparing studying intend to run a series of focus groups and participatory the additional displays, we saw all but the audio condition design workshops to engage and empower the employee in increase the consumption. While the increase shown by developing an understanding of both the economic and these alternatives was significantly less than the ambient environmental impact of their working practices. The light display in particular, none were significant. Although participatory design workshops will provide an opportunity the audio feedback did marginally reduce consumption, we for employees to be directly involved in designing the UX also recorded a significant number of opt-out button presses element of Electro-Magnates therefore helping to address in the audio condition, indicating that people disliked this ethical concerns over privacy and appropriate disclosure of particular installation. Qualitative comments from an energy data. optional online survey confirmed this. Given the surprising Early work to date includes prototyping a high-impact increase created by the ambient light display, we concluded energy interface for overall energy usage in Figure 3, page the study by reinstalling the ambient light display for a final viewed on 09/01/2011, as well as a competitive league table week. Although not quite double the average consumption, for buildings. Both prototypes are designed for large we again saw a significant increase in energy consumption. situated displays and are abstracted presentations of what is In the end, none of the displays managed to significantly possible with raw energy sensor data which in itself is decrease consumption of water. It is promising, however, intangible and difficult to interpret. that not all the displays increased consumption significantly. This means that such displays do not simply have the opposite effect in work environments. Instead, the results suggest that people simply do not care for the consumption of the company as a whole, and potentially do not mind entertaining themselves with the resources of the company by using additional resources. The fact that significantly more users opted out of the audio display, which was the only one to reduce average consumption, further indicates that people do not mind avoiding resources in this area; that they do not feel personally motivated to accept the nudging technology. Study 2 – Energy Consumption in the Work Place Our recently commenced Electro-Magnates study [6] aims to reduce energy usage in the workplace by utilising a suite of social persuasive applications to encourage pro- environmental behaviours. Personal desktop applications (social widgets) and situated displays will be used to deliver energy feedback to individuals, groups and communities Figure 3 High-impact visualisation of overall energy usage about their own – and others’ – energy usage to foster exchange of performance and to support constructive DISCUSSION competition to reduce consumption. The workplace in the The workplace, as an example of a non-domestic, non- context of this study is educational and public sector work- personal environment, creates many unique issues for the place environments in the county of Lincolnshire, UK. ideas behind nudging behaviour. Consequently, we have identified three initial dimensions that differentiate In previous work [5], we reduced domestic energy domestic and workplace environments that might be used as consumption through social norms and social technology. a formative framework for thinking about applying nudging However, designing a similar system for the workplace technology in different environments:- presents greater challenges across a range of design, ethical and technical issues. From our study focus groups in the Expression of Self. First, the workplace may be termed a domestic environment we discovered that for some people special environment in that there are usually constraints and cost was the primary motivating reason to reduce their rules in how employees can interact and carry out activities energy use. In the workplace employees are not typically in the workplace compared to their less inhibited personal responsible for paying energy costs, neither are they life. This is particularly important when considering directly responsible for meeting any governmental carbon employee consumption of resources with emphasis on policies in place that could lead to institutional ‘carbon’ ownership, freedom of choice and sustainable behaviour. fines. Ironically, an individual may be committed to pro- environmental behaviour when at home but is forced to engage in negative practices at work such as using ACKNOWLEDGMENTS inefficient energy-intensive equipment or sitting in an over- We’d like to thank the NIMD2010 participants for heated environment. discussions. The Electro-Magnates project is funded by the HEFCE LGM fund. Sense of Responsibility. Second, prior research typically assumes that individuals are trying to change their REFERENCES behaviour, or reduce their consumption, but for many the [1] 1E PC Energy Report 2009. online at workplace is not their own and not their responsibility. http://www.1e.com/Energycampaign/Index.aspx Consequently, not only is the environment and technology controlled for them, people have a diminished sense of [2] Arroyo, E., Bonanni, L. and Selker T., Waterbot: responsibility for the energy costs and environmental Exploring Feedback and Persuasive Techniques at the sink, impact. In Proc CHI 2005, ACM Press(2005) [3] Darby, S. The Effectiveness of Feedback on Energy External Constraints. Third, the workplace or type of work Consumption. Enironmental Change Institute, The has its own requirements – they may need to maintain 24-7- University of Oxford, 2006. 365 server support. It may be normal for some businesses to have 3 or more machines running per individual, but [4] Foster, D., Linehan, C., Lawson, S., and Kirman, B. unusual for others to have a computer at all. This kind of Motivating physical activity at work: using persuasive top-down requirement might make individuals feel out of social media for competitive step counting. In Proc. control of the environment and its consumption, leading to Mindtrek 2010, ACM Press(2010) lack of motivation. [5] Foster, D., Lawson, S., Blythe, M., and Cairns, P. Given these limiting and influential factors, it is hard to Wattsup?: Motivating reductions in domestic energy consider how we can utilise the same nudging technology consumption using social networks. In Proc. NordiCHI that we typically apply in domestic contexts. The few 2010.ACM Press(2010) successful workplace nudging installations have typically [6] HEFCE 2010. online at been dependent on a driven community. The CleanSink http://www.hefce.ac.uk/lgm/build/lgmfund/projects/show.as project [2] saw some positive influence in hospitals, where p?id=195&cat=15 cleanliness is both required and important for care. Our on- [7] Kuznetsov, S. and Paulos, E. UpStream: Motivating going study on energy consumption in Lincoln, is focusing Water Conservation with Low-Cost Water Flow Sensing on driving community motivation, which may encourage and Persuasive Displays. In Proc. CHI 2010, ACM Press expression of self and increase sense of ownership, whilst (2010). working within the external constraints of the workplace. [8] Siero, F.W., A.B. Bakker, G.B. Dekker, and M.T.C. van CONCLUSIONS AND FUTURE WORK den Burg. . Changing organizational energy consumption Much of the prior research on Nudge, and other PINC behavior through comparative feedback. In Journal of issues, has assumed that individuals are focusing on their Environmental Psychology 16: 235-246. (1996) environments, behaviours, consumption, and other things [9] Stebbins R. (2007) Serious Leisure: A Perspective For that they are in some control over. How does Nudge fare in Our Time. Transaction: New Brunswick. environments, like the workplace, that are typically outside [10] Thaler, R and Sunstein, C. (2008) Nudge: Improving of an individual’s control? Such questions are important for Decisions About Health, Wealth, and Happiness: Yale larger organisations who want to improve their collective University Press. behaviour, whether it is a business trying to reduce its own consumption or meet it’s quota of carbon credits, or a government trying to reduce the nation’s consumption. In our future work, we are focusing on this issue in two ways. First, our funded research is focusing further on encouraging community-driven nudges for reducing business and employee consumption. Second, we are planning future studies that specifically investigate the nudge of groups and communities rather than of individuals, as to meet the UN’s Millennium Goals1, we need to nudge the behaviour of the global community and not just that of individuals. 1 http://www.un.org/millenniumgoals/