Help Lite (Aleven, Roll)

Hints during tutored problem solving – the effect of fewer hint levels with greater conceptual content

Vincent Aleven, Ido Roll, Kenneth Koedinger

Abstract

This in vivo experiment compared the effectiveness of two styles of hint sequences during tutored problem solving. The study was carried out in the Geometry LearnLab. Two conditions were compared, each working with its own tutor version. The tutor versions differed only with respect to the content of the hint sequences. A key difference between the hint sequences was that the number of hint levels was reduced from about 7 in a typical hit sequence to 2 or 3. This was achieved by removing hints that merely reminded students of their current goal within the problem, by removing hints that encouraged students to try to address their question by using the Glossary, and by being more concise in explaining how a theorem or definition could be explained. At the same time, conceptual content was added, in the form of explanations of geometry terms.

Glossary

• Help-seeking behavior: Help-seeking related decisions students make (whether implicitly or explicitly) while learning. These decisions affect the type of actions taken by the student (ask for a hint, search the glossary, try to solve, ask the teacher or a friend), and the duration and effort dedicated to these actions (e.g., skip the hint, skim through it, or read it thoroughly).

• Clicking through: The most common type of Help Abuse. When 'clicking through' students ask for more hints repeatedly, without reading the hints themselves, until the most elaborated hint ('bottom out hint' is displayed.

• Hint sequence:

• Hint level:

• Hint content:

• Help Avoidance:

• Help Abuse:

• Bottom out hint:

• Metacognition

• Gaming the system

• Cognitive Tutor:

Research question

How is robust learning affected by shorter hint sequences with richer conceptual content?

Independent Variables

Type of hints:

• Control: Standard Cognitive Tutor hints; Including 7 levels of hints of different types: containing either domain knowledge or learning recommendations (such as 'search the glossary for ...')
• Experimental condition: Included only 2-3 levels of hints, each of which includes only domain knowledge.

Dependent variables

The study uses two levels of dependent measures: - Assessing Help Seeking behavior by analyzing log-files against a model of ideal help-seeking behavior - Assessing learning in a normal post-test. - Assessing performance (correctness) on the hinted step, as well as on the following step using the same skill.

Accelerate future learning: Would be appropriate and interesting to assess whether students use hints more effectively in subsequent tutor units (e.g., reduced clicking through and other hint abuse, perhaps reduced hint avoidance). This assessment was not pursued however, in part because the study took place in the last tutor unit that students worked on during the school year.

Transfer: Data insufficiency items included in the post-test form.

Hypothesis

Students pay more attention to short hint sequences as they feel they are more helpful and easier to understand. Thus, the shorter hint sequences reduce hint abuse, such as students’ clicking through hints until they get the answer, without paying attention to why the answer is what it is. The richer conceptual content helps them to make sense out of the tutor’s hints, reducing implicit learning and also making students more likely to attend to the hints. Thus, there are two reasons why the new hints result in better sense making and less implicit learning.

Explanation

Having informative, relevant and on-time hints provides the student an effective learning trajectory when learning-by-doing becomes to difficult. The original help sequence require the learner for more responsibility - identify relevant hints, search the glossary, etc. These activities require cognitive load. However, the updated hint sequence offer relevant instruction when required, and low on extraneous cognitive load.

Annotated bibliography

Aleven, V., & Koedinger, K.R. (2000) Limitations of student control: Do students know when they need help? in proceedings of 5th International Conference on Intelligent Tutoring Systems, 292-303. Berlin: Springer Verlag. Aleven, V., McLaren, B.M., Roll, I., & Koedinger, K.R. (2004) Toward tutoring help seeking - Applying cognitive modeling to meta-cognitive skills . in proceedings of 7th Int C on Intelligent Tutoring Systems, 227-39. Berlin: Springer-Verlag. Aleven, V., Roll, I., McLaren, B.M., Ryu, E.J., & Koedinger, K.R. (2005) An architecture to combine meta-cognitive and cognitive tutoring: Pilot testing the Help Tutor. in proceedings of 12th Int C on Artificial Intelligence in Education, Amsterdam, The Netherlands: IOS press. Aleven, V., McLaren, B.M., Roll, I., & Koedinger, K.R. (2006). Toward meta-cognitive tutoring: A model of help seeking with a Cognitive Tutor. Int J of Artificial Intelligence in Education(16), 101-30 Roll, I., Aleven, V., & Koedinger, K.R. (2004) Promoting Effective Help-Seeking Behavior through Declarative Instruction. in proceedings of 7th Int C on Intelligent Tutoring Systems, 857-9. Berlin: Springer-Verlag. Roll, I., Baker, R.S., Aleven, V., McLaren, B.M., & Koedinger, K.R. (2005) Modeling Students’ Metacognitive Errors in Two Intelligent Tutoring Systems. in L. Ardissono, (Eds.), in proceedings of User Modeling 2005, 379-88. Berlin: Springer-Verlag. Roll, I., Ryu, E., Sewall, J., Leber, B., McLaren, B.M., Aleven, V., & Koedinger, K.R. (2006) Towards Teaching Metacognition: Supporting Spontaneous Self-Assessment. in proceedings of 8th Int C on Intelligent Tutoring Systems, 738-40. Berlin: Springer Verlag. Roll, I., Aleven, V., McLaren, B.M., Ryu, E., Baker, R.S., & Koedinger, K.R. (2006) The Help Tutor: Does Metacognitive Feedback Improves Students' Help-Seeking Actions, Skills and Learning? in proceedings of 8th Int C on Intelligent Tutoring Systems, 360-9. Berlin: Springer Verlag.