Craig observing

From LearnLab
Revision as of 16:53, 16 February 2007 by Koedinger (Talk | contribs) (Dependent variables)

Jump to: navigation, search

--Scotty 12:53, 19 September 2006 (EDT)

Learning from Problem Solving while Observing Worked Examples

Scotty Craig, Soniya Gadgil, Kurt VanLehn, and Micki Chi


This research project investigated why students learn from collaboratively observing examples. Previous laboratory research has shown that learners who watch a video of a problem solving tutoring session while collaboratively solving the same problems with a partner learn significantly more than learners that observed the tutoring session and solves the problems alone (Chi, Hausmann, & Roy, under revision). The robustness of this effect was tested by seeing if it transfers into the Physics learnlab.

Students either collaboratively or individually observed videos on the principles of rotational kinematics. The videos presented either a tutoring session or an example. The tutoring videos showed an expert human tutor working with undergraduates taking an introductory physics course. The worked example videos consisted of the expert tutor solving the rotational kinematics problems while orally describing the steps and reasoning. The Andes system is used throughtout the experiment both as the backdrop for the two sets of videos and by the students who solved Andes problems both during training and as transfer assesments.


See Craig Observing tutoring Glossary

Research question

How is robust learning affected by collaboratively versus individually observing different types of worked examples?

Independent variables

The current study varied both number of observers and type of video observed. The multiple-observer variable consisted of two participants observing a video while problem solving or an individual participant watching a video while problem solving. Information presentation format was used to manipulate the example type variable. Participants either watched an expert worked example of Andes problems that provided steps to solve the Andes problems and information on why the steps where needed or they watched a tutoring session where a human tutor works with a tutee to help with solve the problems. Since this study will be conduced in the learnlab, the condition where an individual observed the tutoring session was eliminated because previous lab studies have not shown this contrast to be effective.


A dialogue hypothesis for collaboratively observing while problem solving from worked examples would be that viewing the expert tutoring session would produce more learning (normal or robust) than viewing a content equivalent condition of expert problem solving. However, an alternative (Content equivalency hypothesis) would be that since the expert tutoring session and the expert worked example both provide good learning conditions with the same content they should both produce mastery of the material (Klahr & Nigam, 2004).

Dependent variables

  • Transfer, immediate: After exposure to the treatment, students will complete three transfer problems in Andes. These problems will test the same concepts from training in new situations that require implementation of the problems in new ways.
  • Normal post-test: Students will be given a 12 item multiple choice pretest and posttest that taps into their ability to apply the principles of rotational kinematics to new situations. This will serve as a measure of immediate learning for the study.
  • Homework by normal post-test and transfer items: After training, students did their regular homework problems using Andes. Students can do them whenever they want, but most normally complete them just before the exam. The homework problems were divided based on similarity to the training problems. Homework for both similar (near transfer) and dissimilar (far transfer) problems will be analyzed.
  • Accelerated future learning: The training was on Rotational kinematics, and it was followed in the course by a unit on Rotational Dynamics. Log data from the torque homework will be analyzed as a measure of acceleration of future learning.


Some preliminary analyses were conducted on our transfer multiple choice data. So far, an analysis of the data has yielded significant learning gains between pretest to posttest, F (1,65) = 14.987, p <.001 with a proportional M =.56 and M = .66 respectively. However, so far there no significant differences have been found among conditions on the questionnaires.


This study is part of the Interactive Communication cluster, and its hypothesis is a specialization of the IC cluster’s central hypothesis. The IC cluster’s hypothesis is that robust learning occurs when two conditions are met:

  • The learning event space should have paths that are mostly learning-by-doing along with alternative paths were a second agent does most of the work. In this study, the collaboration conditions could comprise the learning-by-doing paths where learners can work together to complete the Andes problems or the paired learners could reply on the video as their information providing agent and simply copy the steps. Alternatively the participants in the solo condition would have to rely exclusively on the video for information and thus reply on more direct copying of steps thus allowing another agent (the video) to do most of the work. In this case, both learning conditions offer the alternate copying path. However, copying could differ in frequency and be more likely to be discouraged in the collaborative condition due to the more social nature of the task.
  • The student takes the learning-by-doing path unless it becomes too difficult. This study attempts to control the student’s path choice by presenting them with the tutorial dialogue that could encourage communication or an expert worked example that gives a walk through of the problem without the dialogue interaction. So, when students take the learning-by-doing path, they learn more than when they take the alternative path.

Annotated bibliography


  • Chi, M. T. H., Roy, M., & Hausmann, R. G. M. (in press). Learning from observing tutoring collaboratively: Insights about tutoring effectiveness from vicarious learning. Cognitive Science.
  • Craig, S. D., Driscoll, D., & Gholson, B. (2004). Constructing knowledge from dialog in an intelligent tutoring system: Interactive learning, vicarious learning, and pedagogical agents. Journal of Educational Multimedia and Hypermedia, 13, 163-183. [1]
  • Gholson, B. & Craig, S. D. (2006). Promoting constructive activities that support vicarious learning during computer-based instruction. Educational Psychology Review, 18, 119-139. [2]
  • Klahr, D. & Nigam, M. (2004). The equivalence of learning paths in early science instruction: Effects of direct instruction and discovery learning. Psychological Science, 15, 661-667.