Exploring and Elevating Healthy Behaviors with Social Technologies (AAHB Conference), March 2015
•Download as PPT, PDF•
1 like•826 views
Talk given at the American Academy of Health Behavior conference in San Antonio Texas on exploring and elevating healthy behaviors with social technologies.
Recommended
More Related Content
What's hot
What's hot (20)
Viewers also liked
Viewers also liked (20)
Similar to Exploring and Elevating Healthy Behaviors with Social Technologies (AAHB Conference), March 2015
Similar to Exploring and Elevating Healthy Behaviors with Social Technologies (AAHB Conference), March 2015 (20)
More from Derek Hansen
More from Derek Hansen (18)
Recently uploaded
Recently uploaded (20)
Exploring and Elevating Healthy Behaviors with Social Technologies (AAHB Conference), March 2015
- 1. Exploring and Elevating Healthy Behaviors with Social Technologies AAHB Conference, March 2015 Derek L. Hansen Abell Professor of Innovation School of Technology, BYU dlhansen@byu.edu @shakmatt
- 3. Technology-mediated social participation (TMSP) “The goal is to create new architectures for the online public spaces that energize the population to contribute to vital community and national priorities” - IEEE Computer, Nov. 2010
- 4. Part I: Exploring Health Behaviors
- 5. What can we learn about health behavior now that would not have been possible before?
- 6. From Milgram to Facebook
- 7. Data Sources Direct Streams Free Solutions SocialScientistsComputerScientists APIs Scrapers Software Libraries Aggregated Data Paid Solutions
- 8. Problems with Social Media Data
- 9. Need for Data Scientists
- 10. Research Methods & Skills • Social Network Analysis • Data Visualization • Data Mining • Machine Learning (e.g., classification) • Natural Language Processing (e.g., sentiment analysis) • Predictive Statistics (e.g., regression) • Online Experiments
- 11. Adderall Abuse by College Students Combine data sources & look for patterns
- 12. Predicting Personality via Facebook Predict, don’t just describe Predicting personality from Facebook data Predicting depression from Twitter data
- 13. Look for structure, not just content
- 14. Stigmatized Illness & Social Media Mix in different data as controls
- 15. Historical note: how methodology impacts findings WWW WWW WWW WWW WWW WWW WWW W W W W W W W W W WWW WWW WWW WWW WWW WWW W W W W W W W W W Focus on Content Focus on Search Process, Usability, & Success
- 16. Part II: Elevating Healthy Behaviors
- 17. What can we design to promote healthy behavior now That would not have been possible before?
- 19. Socio-Technical Systems Cognitive-Technical System Social-Technical System How a Cockpit Remembers its Speed Hutchins, Edwin
- 20. Veiled Viral Marketing Leverage unique possibilities, don’t replicate the real world
- 21. Community Wiki Repositories Look outside of health domain for inspiration
- 22. NodeXL SNA Tool Develop new analysis tools & methods, not just interventions
- 23. Mobile, Social Games Wed social science constructs with emergent phenomena
- 24. Relevant Technology Trends • Merging of online and offline – Internet of things, wearables, augmented reality, context-aware devices, beacons • Gaming – Mobile games, social games, virtual reality • Data mining and analysis across channels • Other – Mobile Payments, crowdfunding, tools for casual programmers
- 25. Questions & Discussion Derek L. Hansen Abell Professor of Innovation School of Technology, BYU dlhansen@byu.edu @shakmatt
Editor's Notes
- Introduction. Received PhD from University of Michigan’s School of Information. Am currently at Brigham Young University’s School of Technology’s Information Technology program. See https://scholar.google.com/citations?user=IpLkSvEAAAAJ&hl=en for publications. I am also affiliated with the Computational Health Science group at BYU: see http://dml.cs.byu.edu/chs/people.php My research focuses on the analysis and design of novel social technologies aimed at improving some domain of interest. My home research community is Human Computer Interaction & CSCW, though my work often relates to consumer health informatics, citizen science, and educational games.
- See special issue of IEEE Computer, Nov. 2010 focused on Technology-Mediated Social Participation. The chapter authored by Brad Hesse, Derek Hansen, Thomas Finholt, Sean Munson, Wendy Kellogg, and John Thomas on health 2.0 is particularly relevant: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3046402/
- The use of social media is staggering and completely unprecedented in the history of the world. For example, more than 1.39 Billion monthly active Facebook users and 1.9 billion mobile active users. 20 minutes average time spent on Facebook visit. See https://zephoria.com/social-media/top-15-valuable-facebook-statistics/ for more stats. All this has happened in just a decade! Given this world we live in, what is possible? We need to rethink our research methodologies and strategies. Image credits: http://laalex.deviantart.com/art/Vector-Social-Icons-on-Caps-258636799
- An good example of how things have changed is the difference between the early studies (conducted in the 1960s) by Stanley Milgram and the more recent studies conducted by Facebook researchers that attempt to measure the average social distance between people (i.e., the “degrees of separation) based on hundreds of millions of people. It turns out we are even more connected than we originally imagined. See https://www.facebook.com/notes/facebook-data-team/anatomy-of-facebook/10150388519243859 (where image is from) for details. Notice that Milgram’s methods were appropriate – and the best he could do – in his time. Now, new methods are needed that can leverage the data that is collected.
- Currently the best options for collecting data include: Working directly with those who hold the data (e.g., sabbatical at facebook research) – this can be hard to gain access though. Those who can code themselves can work with data source APIs directly, or they can work with 3rd party companies who aggregate data (see top row). Aggregators can be expensive though (and are targeting a business market, not researchers) and apis of social media sources change frequently and not all companies have apis (or apis for the data you want). Those who can’t code themselves can use existing social media monitoring software (though they may cost money and they are designed for business users, not researchers) or free research-oriented software (but these often lack funding resulting in less polished products)
- Data is performance. Only a small percentage contribute, and they are not typically like the other participants. The ethics of studying users online behavior is still murky and fraught with danger Participants of most social media sites are not representative of the larger population More generally, there is a risk of overgeneralization of findings. Image credits: - http://commons.wikimedia.org/wiki/File:Bend_and_Snap,_contemporary_dance_performance_at_Nazareth_College_Arts_Center,_Rochester,_New_York_-_20090925.jpg http://go-digital.net/blog/2012/05/90-9-1-or-1-9-90-rule-of-social-media-participation/ http://www.beevolve.com/twitter-statistics/ http://upload.wikimedia.org/wikipedia/commons/7/77/Lightmatter_lab_mice.jpg
- There is a growing need for “data scientists” who are conversant in the various analytic and visualization techniques that help make sense of large datasets. Image: http://upload.wikimedia.org/wikipedia/commons/4/44/DataScienceDisciplines.png
- Note that there is great power when these things methods are used together.
- See http://www.jmir.org/2013/4/e62/ for full paper. Tweaking and Tweeting: Exploring Twitter for Nonmedical use of psychostimulant Drug (Adderall) Among College Students. Carl Hanson, Scott Burton, Christophe Giraud-Carrier, Josh West, Michael Barnes, Bret Hansen
- 1) Image based on Jen Golbeck’s paper: http://dl.acm.org/citation.cfm?id=1979614 Personality values were predicted to ~90% accuracy based on Facebook data. The data included structural information (e.g., number of friends), textual data (e.g., swearwords used), and other usage statistics. 2) Table based on Munmun De Choudhury and colleagues at Microsoft Research paper http://dl.acm.org/citation.cfm?id=2466447 that predicts depression based on Twitter activity up to just over 70% accuracy.
- This work by Marc Smith, Lee Rainie, Itai Himelboim, and Ben Shneiderman for the Pew Internet and American Life project. See http://www.pewinternet.org/2014/02/20/mapping-twitter-topic-networks-from-polarized-crowds-to-community-clusters/
- From forthcoming article in Journal of Health Communication titled “Stigma's Effect on Social Interaction and Social Media Activity” by Vanessa Boudewyns 1, Itai Himelboim, Derek L. Hansen, and Brian G. Southwell. The paper found that stigmatized illnesses were discussed less on Twitter than non-stigmatized illnesses when controlling for their popularity (as estimated by number of google searches) and funding (based on NIH funding). Such methods could be used to help identify health topics that are stigmatized and measure how the stigmatization of an illness changes over time (e.g., HIV is an interesting case study – though unique in that it has had so much funding).
- Original studies conducted primarily in medical schools used approximately the following methodology (left-hand side): (1) Choose health topic, (2) select subset of webpages on topic (e.g., ones that show up highly in search engine results), (3) have experts review sites for accuracy and completeness – giving each site a + or – score (with inter-rater reliability reported), (4) report findings/conclusions (“look how much bad content there is!”) Other studies conducted more recently from an “information seeking” perspective (e.g., JASIST) use the following methodology (right-hand side): (1) choose health topic and create common search tasks on topic, (2) choose subset of people (e.g., adolescents, older adults…), (3) systematically observe them searching (with “think aloud” protocol”), (4) code search process, content viewed, and success of searches, (5) report findings/conclusions (“people are more critical of content than we assumed, but some of them sure don’t know how to create good search terms or use browsers effectively”) The first type of study made information provision on the Web seemed hopelessly flawed, yet studies of the 2nd type showed that many people used strategies to compare content from different sites (e.g., to find inaccuracies) or verified with doctors so that information seeking was more successful. Also, importantly, studies of the 2nd type showed what problems and contexts users were engaged in when seeking for information, which led to useful ideas for interventions and training – as opposed to studies of the first type which seemed to always suggest more accurate content on the Web. See http://www.jmir.org/2003/4/e25/ for an example of the 2nd type focused on adolescents. For one focused on older adults see http://dl.acm.org/citation.cfm?id=2132176.2132220
- This is a picture of Douglas Englebart, an early technology pioneer. In a time when many people thought of computers as replacements for humans (i.e., they were focused on automating tasks), Englebart thought of computers as tools that could augment the human intellect.
- The use of social media is staggering and completely unprecedented in the history of the world. For example, more than 1.39 Billion monthly active Facebook users and 1.9 billion mobile active users. 20 minutes average time spent on Facebook visit. See https://zephoria.com/social-media/top-15-valuable-facebook-statistics/ for more stats. All this has happened in just a decade! Given this world we live in, what is possible? We need to rethink our research methodologies and strategies. Image credits: http://laalex.deviantart.com/art/Vector-Social-Icons-on-Caps-258636799
- See Lawrence Lessig’s book “Code and Other Laws of Cyberspace” – version 2.0: http://codev2.cc/download+remix/Lessig-Codev2.pdf The “Architecture” disk is the key here – it’s the one that is so dramatically different when so much of our communications and interactions are mediated by technology. While we can design all of these things, it is the design of the architecture that is so new.
- Edwin Hutchin’s classic paper explores the idea of treating a cockpit as a unit of analysis from a cognitive psychology standpoint – one that includes both human and technological components to perform computation and memory tasks. However, as most of cognitive psychology work, it focuses on one individual and not emergent properties on a social level. What would/does a socio-technical social system look like and how can we analyze them? Lostpedia provides one example of a socio-technical system engaged in “sensemaking by the masses”.
- Derek Hansen and Christianne Johnson, “Veiled Viral Marketing: disseminating information on stigmatized illnesses via social networking sites. See http://dl.acm.org/citation.cfm?id=2110393 for details. One other lesson to take away is to focus on unique constructs that can be applied across different platforms, not things that may be outdated within a few years.
- Thought dated now, my dissertation work examined the symbiotic relationship between threaded conversation (in the form of email lists or discussion forums) and wiki repositories (i.e., collaboratively authored documents). At the time no medical support communities used them, but some tech-savvy communities did such as css-discuss. After understanding what made it work for technical support, I translated the design principles into the medical support community context and developed a custom wiki for them to use to augment their community repositories. 10 years later they are still using it in valuable ways.
- See http://nodexl.codeplex.com/ for NodeXL – a free Social Network Analysis tool. Book on NodeXL by Derek Hansen, Ben Shneiderman, and Marc Smith is available at: http://www.amazon.com/Analyzing-Social-Media-Networks-NodeXL/dp/0123822297 In addition to the tool, we have developed novel visualization techniques (http://hcil2.cs.umd.edu/trs/2011-24/2011-24.pdf) and methods for analyzing data ( https://www.cs.umd.edu/~ben/papers/Hansen2011EventGraphs.pdf or presentation here http://www.slideshare.net/shakmatt/eventgraphs-talk-at-hcil2011 )
- We need mid-level design-based theory that translates theory developed in lab settings and translates it into specific products. A sort of “translational” research – but for design. I am currently working with a graduate student to develop a set of mobile, social games that will leverage step counting devices like Fitbit (or the built-in Android or iPhone step counters). Image credit: https://c2.staticflickr.com/4/3540/3301550694_d2615042cb.jpg See the following articles reviewing mobile apps and the types of behavior change principles they apply (most suggest that they don’t apply very many): http://mhealth.jmir.org/2015/1/e20/ http://www.ajpmonline.org/article/S0749-3797(14)00614-X/abstract http://www.ijbnpa.org/content/11/1/97 http://www.ajpmonline.org/article/S0749-3797(14)00040-3/abstract http://www.tandfonline.com/doi/abs/10.1080/15398285.2013.756343#.VQQteRDF8nQ