Research:Directed diabetes info-seeking behavior in Wikipedia
In this study, we sought to explore information seeking behavior in the context of Wikipedia. We instrumented participants browsers to log their scroll positions and then asked them to look up information about diabetes mellitus. Through visualizing the scrolling behavior of these readers, we can see different types of scrolling patterns.
- Identify how effective the presence of hyperlinks are to readers' ability to process information.
- Identify patterns in browsing behavior of experts and non-experts.
- Correlate where users look for links with the location of links in Wikipedia.
Participants in our experiment are Wikipedia readers, not editors. Participants will be recruited for a study on health information seeking behavior. Participants will complete pre- and post-study surveys on opinions about trust in health information online, and health expertise. No identifying or private health information will be collected. The majority of the study will be conducted while the participant is logged in to a special Wikipedia account which has had our study "gadget" pre-installed. Each user will be issued with a unique special account, to maintain privacy and to allow for users to complete the survey in parallel. The gadget will guide users through a survey requiring them to find specific health information (for example, "What oral medication is commonly used to treat Type 2 Diabetes?").
See en:User:Equalx/wiki_isb.js for the gadget code.
The original goals of the study were not met. However we did obtain some preliminary data from mechanical turkers who participated in our study. The following graphs represent the browsing traces of users who participated. Note that all traces start on the Main page and overlapping colored blocks represent multiple tabs/windows open at the same time. User 08 is dropped from the set because they failed to complete the task.
Wikimedia Policies, Ethics, and Human Subjects Protection
This study is a collaborative effort between researchers at the University of Minnesota and Drexel University. It has IRB approval from both institutions (University of Minnesota, 612-626-5654 - #324798347; Drexel University, 215-255-7857 - #732647236). In this study we collect data on user behavior as they navigate around and between pages on English Wikipedia. We collect this data through the wiki_isb user gadget, which must be installed on a user's account to administer data and collect surveys. To maintain user privacy, several steps are taken:
- Participants are given a Wikipedia account which has the gadget pre-installed.
- When the survey is complete, the gadget disables it's data collecting functionality and instructs them to log out.
- The data collection server will be disabled at the completion of the experiment, to prevent unintentional data collection.
Benefits for the Wikimedia community
The project is primarily oriented at scholarly publication, but the models we intend to build should have applications in informing the structure and usability Wikipedia's articles.
Data collection in July/August of 2013, publication and dissemination of results sometime in 2014.
- grouplens.org - Our research lab's website
- Ed H. Chi, Peter Pirolli, Kim Chen, and James Pitkow. 2001. Using information scent to model user information needs and actions and the Web. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI '01). ACM, New York, NY, USA, 490-497. DOI=10.1145/365024.365325
- Xiaojun Yuan and Ryen White. 2012. Building the trail best traveled: effects of domain knowledge on web search trailblazing. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI '12). ACM, New York, NY, USA, 1795-1804. DOI=10.1145/2207676.2208312