Trends in Persuasive Systems Design: Techniques, Methods and Application Domain

Transcription

1 Trends in Persuasive Systems Design: Techniques, Methods and Application Domain Isaac Wiafe School of Technology GIMPA Accra-Ghana Keiichi Nakata IRC, Henley Business School University of Reading UK ABSTRACT There are a number of claims that persuasive technologies are effective in changing or shaping attitude, behavior or both. However, few empirical investigations provide support to this claim. In this paper, we present findings of our study on peer-reviewed papers published in the International Conference on Persuasive Technology from 2006 and Our study focused on the success of existing systems, most dominant technique used, dominant domain of application and most used framework or theory among others. We found that out of the 44 systems reviewed, 23 were reported to be successful, 2 to be unsuccessful and 19 did not specify whether or not it was successful. 56 different techniques were mentioned and it was observed that most designers use ad hoc definitions of techniques. Hence we proposed that there is a need for research to establish unambiguous definitions of techniques and methods. We also observed that the persuasive systems design (PSD) model which is currently the only comprehensive framework for designing persuasive technologies was not reported to have been used in any of the reviewed papers. Keywords: Theories and methods, systems evaluation, specification, system features INTRODUCTION The application of interactive technology for changing human behavior continues to attract attention both in research and the industries. Since its introduction, it has sturdily made progress from simple applications on PCs and mobile devices to more complex and intelligent applications. Applications such as the SmartKitchen [1], Ubifit [2] and icat [3] demonstrates the potentials of persuasive technology. However, research on evaluation of successes, dominant techniques, theories and application domains are few. Hence there is inadequate empirical knowledge on the domination of techniques, methods and application domain. In line with this, we seek to understand how persuasive technology designers have applied theories, methods and techniques across the various domains and also on success reported. Specifically, we reviewed persuasive technology applications which have been published in the International Conference on Persuasive Technology conference from 2006 to To our knowledge, although there are some reviewed papers on persuasive technology, there has been no review that considered techniques, methods and theories without using a predefined framework for evaluation. We expect that this study will provide researchers and designers information on how existing persuasive technologies are designed, their performance and thus provide clues to issues that need attention based on empirical evidence. In the remainder of this paper we present discussions on the motivation of research followed by the methodology used for review. Section 4 presents our findings followed by discussions and conclusion. Isaac Wiafe et.al. 84

2 RESEARCH MOTIVATION Similar to this study, some researchers have reviewed existing work in persuasive technology. One of such research is by Torning et al [4]. In their study, they evaluated all full papers in the persuasive conference proceedings from 2006 to 2008 using the Persuasive Systems Design (PSD) Model [5, 6]. They established that majority of research papers published were experimental, and 84.4% addressed behavior change instead of attitude change. Tailoring, tunneling, reduction and social comparison were identified to be the most studied methods for persuasion. As tailoring and social comparison were discussed in 11 out of the 51 papers reviewed whereas tunneling and reduction were mentioned in 10. Competition and verification were not discussed at all in any of the papers reviewed. They added that ethical considerations were not sufficiently addressed in most papers and also many of the research papers seem to describe the investigated persuasive systems in a relatively vague manner. As such they proposed the need for research into to prescribe predictable persuasive designs and methods for a clearer measurement of successful designs. Recently, another similar study has been conducted by Lehto and Oinas-Kukkonen [7].They evaluate behavior and attitude change in web-based applications with the PSD model and observed that most of the reviewed applications focused on behavior change rather than attitude change. Reduction, self-monitoring, simulation, and personalization were the most used features for accomplishing user s primary task support, whereas tailoring was less used. Apart from those mentioned above most of the related works tends to focus on single systems and report on either how successful a particular technique has been or how the application was adopted. For instance Harjumaa et al. [8] showed that goal-setting is very effective in behavior change interventions when he performed a field trial on the Polar FT60 (a persuasive system designed to promote physical activity). Again, his evaluation was based on the PSD model. We thus observed that most of the existing research on evaluation of persuasive technologies have either focused on a single system or used the PSD model as a basis for evaluation. Consequently, authors tend to consider only techniques defined in PSD model. Hence, less detailed information on other methods and techniques used by researchers in the field of persuasive technology is available. As such we seek to evaluate existing persuasive applications and identify theories, methods and techniques that have been used for designing them. METHOD In order to have a coherent approach to the review, we focused our attention on proceedings of the Persuasive Conference series from 2006 to 2010 (the proceedings for persuasive 2011 was not officially published at the time of this research). Ideally it would have been more appropriate to have a collection of all existing persuasive technologies and access their trends and performance. However, due to the fact that there are a vast number of research publications on persuasive technology, it would be virtually impossible to identify all of these systems with any keyword. Also, in order to ensure that the systems reviewed are of an appreciable standard and quality, we used the Persuasive series since it is the only comprehensive conference on persuasive technology. As with any research of this nature, the main challenge is the selection criteria and the ability to capture the intension of the design. As such we first identified all papers which described any form of a persuasive application. Fifty three papers were identified to have discussed a form of a persuasive application in proceedings from 2006 to However, the number was trimmed to Isaac Wiafe et.al. 85

3 44. Our selection criterion was based on the definition of persuasive technology as an interactive technology which is designed with the intention to shape or change attitude or behavior [9, 10]. As such systems which did not demonstrate enough functionality with an intention to change or shape any form of behavior or attitude [11-14] were discarded. In addition, some systems were identified to have appeared more than once (Percues [15, 16] and icat [3, 17] ) and thus were considered as a single application. Each of the 44 papers were reviewed and a catalogue was developed with the following attributes: the name of the system, the type of interface used, the domain of application, the type of change, the theory or framework used, the list of techniques applied, system success and the year of publication. In identifying the change type, we observed the intention of the designer to either change attitude towards target behavior (ATTB) or current behavior (CB) as defined in [18]. Though one may argue that attitude towards target behavior influences a user s behavior [19], our aim was to identify instances in which the designer aims to change a user s attitude so as to lead to a behavior change later and those that the designer aims at impacting a behavior change without necessarily changing attitude. Our findings and discussion in the next section are based on the selected systems. FINDINGS Out of the 44 systems reviewed, 21 were intended to change attitude towards behavior whereas 23 were for behavior change (see Table 1). We observed that this is contrary to the findings of Torning et al [4]. His research showed that from 2006 and 2008, 84.4% of the 51 peered-reviewed full papers addressed behavior change. However our study found that 52.3% focused on changing behavior. This disparity may be attributed to the fact that our study consider attitude towards the target behavior and current behavior whereas they consider attitude and behavior, hence there is a possibility that they considered systems designed for attitude towards target behavior as changing behavior. We also found that most of the papers failed to provide a clear definition of the reviewed system and this confirmed the argument by Torning et al [4], Lehto and Oinas-Kukkonen [7] that most designers do not clearly define their design approach clear. Table 1. Frequency of persuasive applications, the type of change and reported success from 2006 to 2010 Success/Type of ATTB CB Change Successful Unsuccessful Unspecified Due to the inadequate provision of information by authors, a significant amount of the papers did not provide information on the success of the system. It was observed that out of the 44 reviewed, 23 (52.3%) were reported to be successful and 2 (4.6%) unsuccessful. 19 (43.2%) papers did not specify whether or not the system was successful. Out of the 23 successful applications 10 aimed at changing a user s attitude towards behavior whilst 13 were on behavior change. 10 of the systems designed to change attitude towards behavior did not report on the success of the system as compared to 9 aiming at behavior change. Isaac Wiafe et.al. 86

4 International Journal of Societal Applications of Computer Science Vol 1 Issue 2 December 2012 Distribution of persuasive applications according to domain The distributions of systems among various domains showed that majority of the systems are designed for health promotion. See Fig.1 and table 2. Incidentally, all the Persuasive conference proceedings since 2006 were dominated by themes related to health promotion. Distribution of Persuasive Technology application domain from 2006 to 2010 Leisure 2% security 2% unspecified 2% 9% 7% 53% 25% Fig.1.Distribution of persuasive technology applications according to domain Table 2.Frequency of persuasive applications, their domain of application and type of change Type of Change/Domainn of application Leisure Security Unspecified ATTB CB Also, our findings confirms claims by [20] that persuasive technology has a potential of providing effective solutions to preventive healthcare in the near future. 11 out the 44 were design to prevent the degradation of the environment whereas 4 were on and 3 on education. The number of applications designed for security and leisure recorded one each (see Fig. 1 and Table 2). Interestingly 1 paper did not specify the domain in which the systems is to be applied [21]. In terms of success, 9 out of the 23 systems designed for healthcare related issues reported success while 1 was unsuccessful. 13 did not specify whether or not the system was successful. Table 3 is the distribution of applications according to domain and reported successs for persuasive applications reviewed between 2006 and Isaac Wiafe et.al. 87

5 Table 3.Frequency of persuasive applications, their domain of application and reported success from 2006 to 2010 Success/ Domain of application Leisure security unspecified Successful Unspecified Unsuccessful Dominating Technique In order not to redefine terms and meanings of techniques used, a list of all words used for defining techniques, methods or system features in the reviewed papers was created and tabulated. A total of 56 techniques were listed and we observed that the most frequently used technique or method was feedback. Table 4 shows the frequency of the first 10 techniques or methods identified from the review.19 out of the 44 (43.2%) mentioned feedback as a technique or method for changing behavior or attitude towards the target behavior. This is followed by self-monitoring and suggestion which recorded 16 each, representing 36.4%. Social role was used in 11 of the applications whereas tailoring and tunneling were used in 10 applications. Reminder was mentioned in 9, and rewards and reduction in 7. 8 out of 21 applications designed to change attitude towards target behavior used simulation or feedback and 6 used self-monitoring. For applications designed to change behavior, self-monitoring recorded 11 out of 23 whereas 9 used reminders. Suggestion was also used in 8 applications designed for health as compare to4 in environmental applications. Feedback was mostly applied in applications designed for health or environmental issues with a record of 9 each. It was also observed that, language [22, 23], warning and advice [24] were used even though they all recorded frequencies less than 3. Table 4.Ten most used persuasive techniques in applications from 2006 to 2010 Technique/ Domain of application Leisure security Feedback Self-Monitoring Suggestion Social role Simulation Tailoring Isaac Wiafe et.al. 88

6 Tunneling Reminders Reduction Reward Dominating Theory A total of 15 theories, frameworks or models were record in the 44 reviewed papers. Out of the 44, 24 did not specify the fundamental theory or framework which was used for designing the system; this represents 54.6%. 11of the systems which did not specify the type of theory were designed to change attitude towards target behavior whereas 13 were for behavior change. The most applied framework was the Fogg s functional triad [9]. Six out of 44 mentioned this as a form of guide or design principle. This was followed by the trans theoretical model [25] which was used in 4 applications designed to promote healthcare. The cognitive dissonance theory [26] and the goalsetting theory [27] among other theories recorded a frequency of 1 each. Interestingly, the PSD model [6] which is the only comprehensive model designed specifically for design persuasive technologies was not identified with any of the reviewed papers though it has been used a number of times to evaluate persuasive applications [4, 7, 8, 28]. Table 5.List of theories or frameworks which were used to design applications more than once from 2006 to 2010 Theory or Framework/ Domain of application Fogg s Functional Triad Transtheoretical Model 4 4 Social learning Theory DISCUSSION AND PRACTICAL IMPLICATIONS From the results, we observed that one of the most compelling and challenging issues in persuasive technology design continues to be the application of systematic methods for design. It can be recalled that such concerns have been raised by a number of researchers [4, 10, 18, 29, 30] however, there is little adherence to this call even though some researchers have proposed methods and approaches for designing persuasive technology [6, 18, 30]. Also, it can be observed that some designers continue to treat the functional triads proposed by Fogg [9] as a framework for designing persuasive technology. However Wiafe et al [18], argued that it is not appropriate to treat the techniques as a framework since it does not provide any guide to design; rather it provides features which can be identified or used to attain persuasion. The PSD model [6] which is currently the only comprehensive framework for designing and evaluating persuasive technologies was not reported to have been applied in any of the applications reviewed even though it was introduced in 2008 and 23 out 44 applications were published between 2009 and This might be due to the fact that the PSD model is relatively new as compared to other identified model. Also due to the interdisciplinary nature of the field, most designers may tend to apply models or frameworks within their specific domain which have Isaac Wiafe et.al. 89

7 been proven to be success in human-human persuasion (observe that the trans theoretical model was used for developing health related applications). Nonetheless, there is a need for investigations to ascertain why systems designers shy away from the PSD model. The definition of distinct techniques or methods is another crucial issue that needs attention. 56 different techniques or methods were used as reported. Though one may argue that these 56 can be classified into a smaller number as in the case of the PSD model, there is the need for further research to establish a clear understanding of terms and vocabulary that is used within the domain. The ambiguous definition of terms and definitions inherited from social psychology makes it almost impossible to establish distinct definitions of techniques or a method. However it is worth appreciating that the application of technology for persuasion mainly involves automation of techniques; hence the need for distinct definitions of techniques or methods will improve design methodology. The current ad hoc adoption of terms for techniques or methods may further inhibit the use of a well-structured procedure or framework for designing persuasive technologies. For instance, [31] claimed that the persuasive technique employed in persuading users to conserve water was feedback. However, the provision of an ambient display suggests a technique which can be also be considered as self-monitoring. Some researchers also consider the provision of just-intime responses as a form of technique; however a system can provide suggestions, reminders, rewards, praise and punishment as just-in-time. Physical clues and languages were also considered in some cases as persuasive techniques [22, 23]. Even though Fogg [9] argued that these are forms of techniques or features considered to leverage social cues, there is need to establish when language can be considered as a technique since majority of persuasive systems uses language in one way or the other. Most designers and researchers uses method, technique and approach interchangeably, it would be preferable if researchers can make distinctions between them. This is because, in a number of cases, a designer s definition of a technique or method engulfs a number of other techniques or methods hence making it challenging to identify distinct methods or techniques without ambiguity. These findings among others suggest that research work should focus on unambiguous definitions of persuasive techniques. In particular we propose that there is the need for the creation of a Persuasion Techniques Catalogue which will guide researchers and designers on distinct definitions of terms. This is important because, persuasive technology has expended the number of techniques which were used in the tradition human-human persuasion. Now, designers are capable of persuading using likable virtual animated objects, such as animals, bodies of water, pet plants, etc. which were not possible in human-human persuasion. CONCLUSION AND FUTURE WORK In this paper we have presented findings and discussions on selected persuasive technology applications which were published in the first 5 proceedings of the International Persuasive Technology conference. We observed that out of the 44 systems reviewed, 23 were reported to be successful, 2 to be unsuccessful and 19 did not specify whether or not it was successful. 56 different techniques were mentioned as compared to the 28 systems features proposed in the PSD Model. The most dominant domain of application was healthcare in which 4 of them used the Transtheoretical model. Also, 6 of the reviewed applications used the functional triad proposed by Fogg as a guide to design. Additionally, we observed that persuasive technology designers continue to follow ad hoc approaches in spite of calls for them to follow a systematic approach. It was identified that designers are not using the PSD model though it provides a stepwise approach for designing persuasive applications. As such we propose that further investigations are needed to ascertain why such an approach is not gaining popularity. Isaac Wiafe et.al. 90

8 Also, due to the lack of comprehensive definitions of terms and techniques most designers tend to define techniques in an ad hoc manner which usually creates ambiguity in meanings. Hence we proposed the need for further studies to provide unambiguous definitions of terms in persuasive technology. In particular we emphasize that the systems features in the PSD model can be expanded with well-defined distinctions between persuasive methods and techniques. REFERENCE 1. Chi, P.Y., J.H. Chen, H.H. Chu, and J.L. Lo. Enabling calorie-aware cooking in a smart kitchen. in Persuasive Technology Consolvo, S., P. Klasnja, D.W. Mcdonald, and J.A. Landay. Goal-setting considerations for persuasive technologies that encourage physical activity. in Persuasive Midden, C. and J. Ham. Using negative and positive social feedback from a robotic agent to save energy ACM. 4. Torning, K. and H. Oinas-Kukkonen. Persuasive system design: State of the art and future directions. in Persusive ACM New York, NY, USA. 5. Oinas-Kukkonen, H. and M. Harjumaa. A systematic framework for designing and evaluating persuasive systems. in Lecture Notes in Computer Science Oinas-Kukkonen, H. and M. Harjumaa, Persuasive systems design: Key issues, process model, and system features. Communications of the Association for Information Systems Lehto, T. and H. Oinas-Kukkonen, Persuasive features in web-based alcohol and smoking interventions: A systematic review of the literature. J Med Internet Res, (3): p. e Harjumaa, M., K. Segerstâhl, and H. Oinas-Kukkonen. Understanding persuasive software functionality in practice: A field trial of polar ft ACM. 9. Fogg, B.J., Persuasive technology: Using computers to change what we think and do2002: Morgan Kaufmann. 10. Harjumaa, M. and H. Oinas-Kukkonen. Persuasion theories and it design. in Lecture Notes in Computer Science Knipscheer, K., J. Nieuwesteeg, and J. Oste. Persuasive story table: Promoting exchange of life history stories among elderly in institutions. in Persuasive Technology Mccreadie, C., J. Raper, A. Gunesh, J. Wood, K. Carey, H. Petrie, L. Wood, O. Survey, S. Tyler, and S. Biggs. Persuasive technology for leisure and health: Development of a personal navigation tool. in Persuasive Technology Duncan, J., L.J. Camp, and W.R. Hazelwood. The portal monitor: A privacy-enhanced eventdriven system for elder care ACM. 14. Gram-Hansen, L.B. Geocaching in a persuasive perspective ACM. 15. Reitberger, W., B. Ploderer, C. Obermair, and M. Tscheligi. The percues framework and its application for sustainable mobility. in Persuasive Technology Khaled, R., R. Fischer, J. Noble, and R. Biddle. A qualitative study of culture and persuasion in a smoking cessation game. in Persuasive Technology Vossen, S., J. Ham, and C. Midden. Social influence of a persuasive agent: The role of agent embodiment and evaluative feedback ACM. 18. Wiafe, I., K. Nakata, S. Moran, and S. Gulliver. Considering user attitude and behaviour in persuasive systems design. in European Conference on Information Systems Helsinki, Finland. 19. Ajzen, I., The theory of planned behavior. Organizational behavior and human decision processes, (2): p Ijsselsteijn, W., Y. De Kort, C. Midden, B. Eggen, and E. Van Den Hoven. Persuasive technology for human well-being: Setting the scene. in Persuasive Technology Isaac Wiafe et.al. 91

9 21. Ju, W. and D. Sirkin. Animate objects: How physical motion encourages public interaction. in Persuasive Technology Parmar, V., D. Keyson, and C. Debont. Persuasive technology for shaping social beliefs of rural women in india: An approach based on the theory of planned behaviour Springer- Verlag. 23. Forget, A., S. Chiasson, P. Van Oorschot, and R. Biddle. Persuasion for stronger passwords: Motivation and pilot study. in Persuasive Technology Cavazza, M., C. Smith, D. Charlton, N. Crook, J. Boye, S. Pulman, K. Moilanen, D. Pizzi, R. De La Camara, and M. Turunen. Persuasive dialogue based on a narrative theory: An eca implementation. in Persuasive Technology Prochaska, J., C. Redding, and K. Evers, The transtheoretical model and stages of change. behavior and health education: Theory, research, and practice, : p Festiger, L., A theory of cognitive dissonance. Row, Peterson, New York, Locke, E.A., Motivation through conscious goal setting. Applied and Preventive Psychology, (2): p Räisänen, T., T. Lehto, and H. Oinas-Kukkonen. Practical findings from applying the psd model for evaluating software design specifications. in Persuasive Technology Oinas-Kukkonen, H. and M. Harjumaa. Towards deeper understanding of persuasion in software and information systems. in Persusive Fogg, B. and J. Hreha. Behavior wizard: A method for matching target behaviors with solutions. in Persuasive Technology Kappel, K. and T. Grechenig. Show-me: Water consumption at a glance to promote water conservation in the shower ACM. Isaac Wiafe et.al. 92