Effects of levels of instructional assistance on learning and mental effort in an intelligent tutoring system: Proportional reasoning and middle school student

People need varying levels of assistance - specifically, solution-oriented information provided during practice - when learning how to solve problems. This study examined the effects of different levels of problem-solving assistance on learning, mental effort, and time on task when learning proportional reasoning at the middle school level. Proportional reasoning is an important skill that is both foundational to higher-level mathematics and used in everyday life.;To study assistance in this context, two different types of practice problems were presented to learners: worked examples and tutored problems. The worked examples provided the steps and the answers to the proportional reasoning problems for the learner to review. The tutored problems required the learner to solve each step of the problem. While doing so, the learner received immediate, step-by-step feedback from an example-tracing intelligent tutoring system. The learner could also receive a hint if requested. The study had three conditions: worked examples, tutored problems, and a combination of worked examples and tutored problems.;One hundred forty-three middle school students participated in the study. Participants were randomly assigned to one of three treatment conditions: tutored problems (low assistance), alternating worked examples and tutored problems (mid-level assistance), and worked examples (high assistance). Participants in the tutored problems condition needed to solve eight proportional reasoning story problems. Participants in the worked examples condition were presented with eight proportional reasoning story problems that were already solved. Participants in the alternating worked examples and tutored problems condition were first presented with a worked example, followed by a tutored problem. In total, there were eight problems, four of each type. The problems consisted of four isomorphic problem pairs, meaning the story problems and values were different, but the structure in which a learner solved the problem was the same.;The results showed a significant difference between the three treatment conditions on learning. Learners in the worked example condition scored higher on the post-test than learners in the two other conditions. There was not a significant difference on overall mental effort across conditions. However, when analyzing the isomorphic problem pairs, there was a significant difference in the change of mental effort expended between problem 1 and problem 2 in the pair for pairs 2 and 3, but not for pairs 1 and 4. There was also a significant difference on time on task. Learners in the worked example condition took less time to complete the treatment problems than learners in the two other conditions. Upon further examination in terms of time on task, there was also a significant main effect of treatment condition on time, a significant main effect on problem in the isomorphic problem pair, and a significant interaction effect. In the tutored problem condition, time spent on the second problem in the pair was significantly less than the first problem in the pair. In the alternating worked example and tutored problem condition, the second problem in the pair took significantly more time than the first problem in the pair. In the worked example condition, time spent on the second problem was significantly less than the time spent on the first problem in the pair.;The limitations of the study focus on the duration of the study, the alignment of learning materials and assessments, the authentic learning environment, and the intelligent tutoring system environment, and may provide helpful guidance for other researchers who wish to conduct research in this area and for teachers who are interested in using an intelligent tutoring system in their classrooms.;The findings of this study have several implications for further research and classroom instruction. Future studies might be situated in an authentic classroom setting, with the treatment embedded in the mathematics curriculum. Additionally, the hint use data might be mined to learn more about how learners use the available assistance in the tutored condition. One key implication for instruction is the usefulness of worked examples in situations with a compressed timeframe and novice learners. Research considering these limitations and implications might provide educators with a better understanding of how to achieve the optimal balance of assistance in mathematics education for middle school students.