Integrating gamification features in ITSs has become a popular theme in ITSs research. This work focuses on gamification of shared student/system control over problem selection in a linear equation tutor, where the system adaptively selects the problem type while the students select the individual problems. In a 2x2+1+1 classroom experiment with 267 middle school students, we studied the effect, on learning and enjoyment, of two ways of gamifying shared problem selection: performance-based rewards and the possibility to re-do completed problems, both common design patterns in games. We also in-cluded two ecological control conditions: a standard ITS and a popular algebra game, DragonBox 12+. A novel finding was that of the students who had the freedom to re-practice problems, those who were not given rewards performed significantly better on the post-tests than their counterparts who received re-wards. Also, we found that the students who used the tutors learned significant-ly more than students who used DragonBox 12+. In fact, the latter students did not improve significantly from pre- to post-tests on solving linear equations. Thus, in this study the ITS was more effective than a commercial educational game, even one with great popular acclaim. The results suggest that encourag-ing re-practice of previously solved problems through rewards is detrimental to student learning, compared to solving new problems. It also produces design recommendations for incorporating gamification features in ITSs.

In the current work, we investigate the effects of gamifying shared student/system control in our linear equation tutor, Lynnette. We investigated two gamification fea-tures: giving students the freedom to re-practice previously completed problems (not allowed e.g., in standard Cognitive Tutors) and rewards (stars) for each problem based on students’ performance. These features are similar to Angry Birds’ or Drag-onBox’ problem selection and rewards systems. We hypothesize that 1) the possibility to re-practice problems, added to shared control over problem selection will enhance students’ learning and engagement; 2) rewards based on students’ performance on individual problems will also lead to better learning and engagement. Consequently, we created four experimental versions of Lynnette to evaluate the effects of the two gamification features. Moreover, we included two ecological control conditions in the study: a standard ITS and a commercial algebra game. The standard ITS is a control version of Lynnette without any gamification features and with full system control over problem selection (as is common in e.g. Cognitive Tutors). The algebra game is DragonBox, which has attracted substantial public attention for allegedly helping young children learn algebra in a very short period of time [8, 12]. Although DragonBox has been the subject of at least one research study [1], we are not aware of any studies that empirically investigated its effectiveness in teaching algebra. Given the publicity surrounding the game, it would be good to know how educationally effec-tive and engaging it is, compared to technology proven to be effective in helping stu-dents learn (i.e., an ITS). We conducted a classroom experiment with 267 middle school students to investigate our hypotheses.