Prediction Games: Encouraging Engagement with Data.

다국어 초록 (Multilingual Abstract)

Prediction games, data-driven games modeled after fantasy sports, are aimed to motivate people to explore, analyze, and develop their own understanding of large data sets. They revolve around activities where players examine historical data and information resources to make predictions about future events. As a result, they may help improve the players’ domain knowledge and data interpretation skills. But what matters in the design of such games? And, as we envision prediction games created by instructors in an educational environment, what forms of support aid the authoring of prediction activities yet involve very little to no programming? To answer these questions, we first conducted a survey of fantasy sports players which showed that many seek out information including news and data. They analyze this content to make predictions, resulting in them learning more about the sport. Next, we developed Fantasy Forecaster, a prediction game prototype to gather system requirements and user feedback. Lessons from the survey and development of the prototype informed our prediction games framework and its implementation in the climate domain: Fantasy Climate.Fantasy Climate is a prediction game based on weather data where players select a location among a set of choices based on whether their assessment of upcoming weather. In particular, they are asked to select which location will be warmest and coolest compared to their historic norms on an upcoming date. The game also featured communication tools, integrated climate-related news, and historical weather data with visualizations to make sense of them. User studies of Fantasy Climate revealed that social interaction, particularly asynchronous discussions made the game more engaging and helped players gather information for prediction making. Also, the in-game presentation of domain-related news had an effect on engagement and players' performance.From our prediction games framework and the implementation of Fantasy Climate, we identified a set of necessary and valuable prediction activity specifications which led to the development of the Activity Creation Wizard (ACW). The ACW is an environment that guides the author through a series of steps to author their prediction activity. Features of the ACW included a help system that provides the author with explanations, tutorials and examples during the authoring process. Also included were a template component that allows the author to reuse the customizations of a previously created prediction activity, and tools to automate repetitive and tedious tasks such as building the prediction schedule.The evaluation of the ACW showed no background knowledge was required to use the ACW to author a prediction activity. The help system was in general adequate in assisting the participants in their information needs, templates were found useful by many, and automation reduced the time taken for repetitive tasks. Some authors did not want to use templates or automation in order to have more control over the design of their activity. However, the help system, templates, and automation tools of the ACW were not sufficient in helping the participants understand the consequences of their customization on the prediction activity. Reasoning about the effects of their choices on gameplay was noted as the primary challenge during the authoring task by several participants. Additionally, the evaluation identified alternative ways of authoring the prediction activity that challenged our current design of the ACW, including the potential value of co-dependent customizations and collaborative authoring.Finally, the ACW evaluation also involved a task where participants created a prediction game in the domain of their choice. Interviews with participants on their created prediction games revealed two major findings. One finding was that educational, social, and socio-cultural factors play an important role in what makes prediction games engaging. The other finding was authoring resulted in a recognition by the participants of the educational benefits of prediction games which align well with the primary motives of this research work.