Features

XX.1 January + February 2013
Page: 53
Digital Citation

Crowd saucing


Authors:
Conor Linehan, Tom Leeman, Christopher Borrowdale, Shaun Lawson

back to top 

The World Health Organization suggests that worldwide rates of obesity have more than doubled since 1980 [1]. Rising rates of obesity present a massive public health problem, as excess weight can lead to a number of debilitating conditions, including cancer, type 2 diabetes, cardiovascular disease, and stroke. The majority of large-scale studies conclude that diet is at least as important as physical exercise in terms of maintaining healthy weight, while there have been suggestions that diet alone can explain the clear trend of rising levels of obesity [2].

In response, health organizations worldwide have published healthy-eating guidelines, attempted to educate people on what constitutes a healthy diet, and promoted the benefits of healthier eating. The 5 A Day for Better Health initiative in the U.S. and the Change4Life program in the U.K. are well-known examples of this approach. There is evidence that these programs have successfully improved people's ability to discriminate healthy from non-healthy foods, and raised awareness of the importance of maintaining a healthier diet. However, there is little evidence that they have led to any improvement in dietary intake [3].

Dietary behavior appears incredibly resistant to change, and there are many possible reasons for this. For example, studies conducted in disciplines such as sociology, economics, political science, and public health suggest there are widespread, systematic problems with how we engage with food, requiring widespread, long-term, systematic solutions. However, from a psychological perspective, individual food-related decisions are made in the moment, and these decisions may be susceptible to carefully targeted interventions. They are affected by both the context (social, emotional, temporal, occasional) in which they are made and the consequences of previous choices. Bringing about changes in the ways in which people understand their context and experience the consequences of eating may, on its own, facilitate change in those in-the-moment decisions. Indeed, by understanding the difficulties people face in making healthier food choices, it should be possible to develop relatively simple technology-led interventions.

One key challenge in developing any intervention on dietary behavior is that much food purchasing and consumption in developed countries is habitual, and determined by convenience rather than nutritional qualities. People often fail to think about what they are eating or to consider alternatives. Simply challenging and provoking people to think more critically about their food choices in the moment may bring about significant and lasting change.

Another problem is the complexity involved in understanding the concept of balance in a diet. Earlier we mentioned that people in developed countries are quite good at recognizing the objective nutritional qualities of an individual meal. A much more abstract and difficult task is understanding how that meal contributes to the overall balance of a diet. It is extremely difficult for us to form an accurate picture of how balanced our diet has been over a period of weeks. Indeed, given that the long-term balance is what determines how healthy a diet really is, any one meal could be considered healthy in one context but unhealthy in another, depending on other meals.

The difficulty in making healthier food choices on a day-to-day basis is also complicated by the temporal gap between eating and any observable health-related consequences. You do not instantly feel healthier or lose weight after eating a salad. Nor do you feel less healthy or noticeably gain weight after eating a chocolate bar. If people did notice observable health problems directly after eating unhealthy foods, as happens, for example, when ingesting something poisonous, they would probably avoid such foods in the future. The lack of any natural short-term feedback mechanism means that it is very difficult to learn from experience.

There is great potential in mobile and social computing technologies to help us make healthier food choices. For example, technology may be used to provoke more critical thought at opportune moments. Keeping track of the nutritional values of individual meals and using this data to construct an objective understanding of the healthiness of a longer-term diet should be quite possible through technology. In addition, the ubiquity of social and mobile technology offers the potential to facilitate a short-term feedback system and to help people make better-informed decisions about their diets.

back to top  Measuring and Modifying Behavior

Many approaches may be taken to provoke deeper, more critical thought in people making dietary decisions. However, the concept of behavior modification appears to underlie most technology-led research projects and commercial applications in this area. This approach involves objectively measuring and recording food-related behavior, analyzing whether change has occurred in that behavior, and presenting useful feedback to users. Indeed, research typically explores the role of technology in improving one or more of these processes, which is unsurprising given the often-cited usefulness of smartphones for personal informatics, the self-measurement of aspects of everyday behavior.

Accurately and objectively measuring and recording dietary behavior in an unobtrusive manner is a significant challenge. It is important to record as much of the target behavior as possible; otherwise, analysis will be unreliable. In dietary research, recording all meals but none of the snacks consumed over a period, for example, would give an invalid measurement of the target behavior. To address this, Mankoff et al., for instance, explored the measurement of dietary behavior through collecting participants' shopping receipts [4].

Advances in analysis of dietary data typically focus on removing the need for an expert dietitian as part of an intervention. For example, Noronha et al. crowd-sourced the analysis of the food photographs they collected using Amazon Mechanical Turk [5]. In a previous study, we at Lincoln also investigated the distributed analysis of food photographs through an unpaid, voluntary, collaborative social media system.

In behavioral interventions, feedback is presented to participants to give them information regarding how closely their current level of performance matches their goals. Feedback should always be delivered in as timely and specific a manner as possible. In dietary research, the delivery of timely feedback has the potential to massively improve people's ability to learn from experience how to eat healthier. Research has explored novel methods for presenting feedback to participants. Social networking applications, location-based mobile applications, and social games have all been used in many projects to deliver feedback in a novel, engaging manner.

Indeed, social technology has been identified as a particularly useful and interesting context in which to facilitate behavioral interventions. Online social networks (OSNs), commonly used for the disclosure and discussion of health, can expose social norms and facilitate social comparison, and have been used previously in interventions for goals including lowering energy consumption, increasing exercise, promoting healthier eating, and reducing food waste. OSNs also offer the technological infrastructure to facilitate the recording and analysis of behavior, and the presentation of feedback.

We now discuss two projects conducted recently at LiSC. The first, Social Receipt, targeted food buying, and the second, Plate and Rate, targeted food consumption. Both projects aimed to provoke users to think more critically when making dietary decisions, by providing individualized, informative feedback delivered through social networking applications.

ins01.gif

back to top  Social Receipt

Social Receipt aimed to provoke people to think more critically and make more informed decisions about food purchasing. It sought to provide feedback to users on the healthiness of their diet. A bespoke social networking system was developed that allowed for the uploading of photographs and text-based interaction between users. Food purchasing was measured through the participants' uploading of shopping receipts to the website. Analysis of receipts was crowdsourced among users, and users rated receipts via the traffic-light system commonly seen on food packaging in the U.K. (see Figure 1). We were interested in whether we could derive valid health information from the uploaded receipts through crowdsourcing the analysis of those receipts among the users of the site. Dietary feedback was delivered to users numerically, through the crowdsourcing task, but also socially, through the interaction of participants using typical social-network commenting systems. A prototype of the application is available to access in a limited capacity at http://socialreceipt.tomleeman.co.uk

Twelve participants were recruited; the study ran for 28 days. During this time, participants were expected to take a photo of their shopping receipt each time they visited a supermarket and to share these receipt images on the website. Uptake was low, as only eight of the 12 participants ever used the application, and engagement with the receipt-sharing aspect was not a success. Participants uploaded an average of only 1.5 supermarket receipts each during the four-week trial. Thus, it is highly unlikely that all food purchases were recorded via the application, confirmed by participant responses to a post-study questionnaire. The small amount of receipt data uploaded also meant that an analysis of any change in dietary behavior was impossible. There was also surprisingly little social interaction recorded. Questionnaire responses suggested that participants understood the benefit of the application but did not find it sufficiently engaging to either use consistently or allow it to affect their food purchasing.

Thus, while Social Receipt appeared to have all of the necessary features of an engaging, useful food intervention, participants found it neither useful nor engaging! Participants suggested that features such as integration with existing social networking sites, such as Facebook, plus the inclusion of some game-like activities, may have helped their engagement. These were some of the issues we examined through our follow-up project, Plate and Rate.

back to top  Plate and Rate

Plate and Rate aimed to provoke people to think more critically and make more informed decisions about food consumption. It also sought to facilitate an understanding of long-term dietary balance, something that people often find difficult. A Web-based application was developed, which, similar to Social Receipt, allowed for the uploading of photographs and text-based interaction between users. Food consumption was measured through the uploading of food photographs, and participants were expected to upload photographs of all their meals. Analysis of these photographs was carried out anonymously by other users of the application, who were asked to rate the plates of food uploaded in terms of how closely they resembled the Eatwell Plate, a visualization of a balanced diet developed by the U.K. National Health Service (see Figure 2). Ratings were averaged across users in order to generate a consensus rating for each plate. Feedback was presented to users via their own personal profiles, on which they could see a number of statistics showing how they differed from the guideline for each meal uploaded, as well as a visualization of the overall balance of their longer-term diet.

We were interested in the effect of two separate design strategies on participant engagement. One version of the software involved integration with existing social media (Facebook). Another version involved some game mechanics designed to promote photograph uploads.

Forty-nine participants, split into four experimental conditions, used the application over a two-week period. Results suggested that both integrating the application with Facebook and using game mechanics significantly improved user engagement. Moreover, engagement with this application was much higher than with Social Receipt, with participants averaging one photo upload and two ratings of others' photos per day. While we cannot claim to have measured all of the participants' food consumption, it was certainly an improvement.

Ratings of food photographs varied widely across participants, suggesting that gaining accurate crowdsourced information via the Eatwell Plate may be problematic. Indeed, the concept of dietary balance proved difficult to visualize for participants and quantify for analysis. Unsurprisingly, given the unreliability of participant ratings, those ratings do not suggest there was any trend in dietary intake either improving or worsening over the course of the study.

back to top  Beyond Behavior Modification

The most significant challenge encountered across both projects was in engaging participants with the intervention. Participants recruited for the study did not seem particularly concerned with improving the healthiness of their food purchasing or consumption. This finding should not be a great surprise, however, since participants were recruited based on convenience, rather than from a pool of people who specifically needed or wanted to improve their diets.

It could be argued that the behavior-modification strategies employed in these studies, and indeed in most technology-led healthier-eating tools, are suitable only for those who have already made a commitment to change their behavior. One often-cited theory of behavior change, the transtheoretical model, suggests that different styles of intervention are appropriate depending on the different levels of commitment demonstrated by participants toward behavior change [6]. This means that we would have seen better results had we recruited a more committed sample of participants. However, it also means the types of food interventions currently being developed are only ever going to be useful for a small section of the population, that gains will only ever be incremental, and that those who could benefit most will never engage sufficiently.

The question remains of how to create technology-led interventions aimed at those people who do not already see the need for change. This is clearly where we can do the most good, but it requires a completely different approach. Current styles of technology-led intervention can help you monitor and continue to engage with a program to which you have already committed. What we need, in order to have a real impact on those who would benefit most from intervention, is transformative technology—something that will encourage the understanding that there is a problem and encourage people to engage with an intervention in the first place. This design space is much less informed by personal informatics and behavior modification and is probably more within the realm of the arts.

Indeed, the solution to this problem may lie in an approach such as critical design [7]. The intention of critical design research is typically to develop technology that provokes critical thought over, and deeper understanding of, complex issues. For example, there are a number of projects that question the often-mundane ways in which computer games treat violence and death [8]. Moving forward, it seems worth investigating whether provocative critical design may be useful as a means of persuading people that dietary change is necessary, to the point where they are willing to engage with, and benefit from, the type of behavior-modification technology currently being developed.

back to top  References

1. WHO. Obesity and overweight fact sheet. 2012; http://www.who.int/mediacentre/factsheets/fs311/en/

2. Swinburn, B. Increased energy intake alone virtually explains all the increase in body weight in the United States from the 1970s to the 2000s. European Congress on Obesity. Amsterdam, The Netherlands, 2009.

3. Jebb, S., Steer, T., and Holmes, C. The 'Healthy Living' Social Marketing Initiative: A review of the evidence. Department of Health, London, 2007.

4. Mankoff, J., Hsieh, Hung, H.C., Lee, S. and Nitao, E. Using low-cost sensing to support nutritional awareness. Proc. of UbiComp 2002. 371–378.

5. Noronha, J., Hysen, E., Zhang, H. and Gajos, K.Z. PlateMate: Crowdsourcing nutritional analysis from food photographs. Proc. of ACM UIST 2011. 1–12.

6. Prochaska, J.O. Decision making in the transtheoretical model of behavior change. Med Decision Making 28, 6 (2008), 845–849.

7. Blythe, M., McCarthy, J., Light, A., Bardzell, S., Wright, P., Bardzell, J., and Blackwell, A. Critical dialogue: Interaction, experience and cultural theory. ACM CHI Extended Abstracts 2010. 4521–4524.

8. Grace, L. Critical gameplay: Software studies in computer gameplay. ACM CHI Extended Abstracts 2010. 3025–3030.

back to top  Authors

Conor Linehan is a lecturer in HCI at the School of Computer Science at the University of Lincoln, U.K. His research interests focus on the design of social mobile, and games technology for the promotion of physical health and emotional well-being.

Tom Leeman recently completed an undergraduate degree at the School of Computer Science at the University of Lincoln, U.K., where he carried out the Social Receipt project as part of his studies.

Christopher Borrowdale recently completed an undergraduate degree at the School of Computer Science at the University of Lincoln, U.K., where he carried out the Plate and Rate project as part of his studies.

Shaun Lawson is a professor of social computing in the School of Computer Science at the University of Lincoln, U.K., where he directs the Lincoln Social Computing (LiSC) Research Centre. His research focuses on the role interactive and social technologies play in people's everyday lives and how we might design new digital experiences that address major societal issues such as sustainability and public health.

back to top  Figures

F1Figure 1. Screenshot of the rating system used in Social Receipt.

F2Figure 2. Example of the feedback system used in Plate and Rate. The inner circle represents the participant's data, the outer ring the guidelines.

back to top 

©2013 ACM  1072-5220/13/01  $15.00

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.

The Digital Library is published by the Association for Computing Machinery. Copyright © 2013 ACM, Inc.

Post Comment


No Comments Found