Hausmann Study2

The Effects of Interaction on Robust Learning

Robert Hausmann and Kurt VanLehn

Abstract

It is widely assumed that an interactive learning resource is more effective in producing learning gains than non-interactive sources. It turns out, however, that this assumption may not be completely accurate. For instance, research on human tutoring suggests that human tutoring (i.e., interactive) is just as effective as reading a textbook (i.e., non-interactive) under very particular circumstances (VanLehn et al., 2007). This rises the question, under which conditions should we expect to observe strong learning gains from interactive learning situations?

The current project seeks to address this question by contrasting interactive learning (i.e., jointly constructing explanations) with non-interactive learning (i.e., individually constructing explanations). Students were prompted to either self-explain in the singleton condition or to jointly construct explanations in the dyad condition.

Background and Significance

Several studies on collaborative learning have shown that it is more effective in producing learning gains than learning the same material alone. This finding has been replicated in many different configurations of students and across several different domains. Once the effect was established, the field moved into a more interesting phase, which was to accurately describe the interactions themselves and their impact on student learning (Dillenbourg, 1999). One of the hot topics in collaborative research is on the "co-construction" of new knowledge. Co-construction has been defined in many different ways. Therefore, the present study limits the scope of co-constructed ideas to jointly constructed explanations.

Evidence supporting jointly constructed explanations is sparse, but can be found in a study by McGregor and Chi (2002). They found that collaborative peers are able to not only jointly constructed ideas, but they will also reuse the ideas in a later problem-solving session. One of the limitations of their study was that it did not measure the impact of jointly constructed ideas on robust learning. In a related study, Hausmann, Chi, and Roy (2004) found correlational evidence for learning from co-construction. To provide more stringent evidence for the impact of jointly constructed explanations, the present study will manipulate the types of conversations dyads have by prompting for jointly constructed explanations and measuring the effect on robust learning.

Glossary

Jointly constructed explanation Prompting

See Hausmann_Study2 Glossary

Research question

How is robust learning affected by self-explanation vs. jointly constructed explanations?

Independent variables

Two variables were crossed:

• Interaction: singleton vs. dyad
• Engagement: natural vs. prompted

The interaction variable manipulates whether the participants works individually (i.e., singleton) or with a collaborative peer (i.e., dyad).

Crossed with the first variable, the engagement variable manipulates if the participants are prompted to keep working together (i.e., natural) or reminds them to jointly construct an explanation of the step they just observed (i.e., prompted). Prompting for natural speech is to control for Hawthorn effects, which is the tendency of individuals modify their behavior when observed by an experimenter. Prompting for jointly construct explanations increases the probability that the dyad will traverse that particular learning event path.

Hypothesis

The Interactive Hypothesis: collaborative peers will learn more than the individual learners because they benefit from the process of negotiating meaning with a peer, of appropriating part of the peers’ perspective, of building and maintaining common ground, and of articulating their knowledge and clarifying it when the peer misunderstands. In terms of the Interactive Communication cluster, the hypothesis states that, even when controlling for the amount of knowledge components covered, the dyads will learn more than the individuals.

The Coverage Hypothesis: if both peers and singletons cover the same knowledge components, then they will learn the same amount.

Dependent variables

• Near transfer, immediate: electrodynamics problems solved in Andes during the laboratory period.

• Near transfer, retention: homework performance on electrodynamics problems that are isomorphic to the problems solved during the laboratory period.

• Far transfer, retention: homework performance on electrodynamics problems that are not isomorphic to the problems solved during the laboratory period.

• Acceleration of future learning: homework performance on magnetism problems.

Results

In vivo Experiment

The in vivo data for this study have not been collected (target date = January, 2008). Our predicted results are:

Main Effect (Engagement): The coverage hypothesis (CI) makes the predictions shown in the Table below. First, there should be a main effect for interaction with dyads performing better than the singletons because, as the earlier studies on collaboration suggest, peer learning increases the probability of covering the set of knowledge applications compared to solo learning. The interaction hypothesis (IH) makes a similar prediction, but for a different reason. The interaction hypothesis states that the dyads will outperform the singletons because of the interaction itself.

Main Effect (Prompting): The coverage hypothesis predicts a main effect for prompting such that the prompted condition will cover more knowledge components and thus learn more than the natural condition. Evidence for this prediction can be found in studies examining individual and collaborative learning. In terms of individual learning, Chi et al. (1994) successfully increased the rate of self-explanations through the use of content-free prompts, which then led to greater amounts of learning (especially on more difficult items). For collaborative learning, Coleman (1998) was able to prompt groups of students to produce explanations. Therefore, we expect that the prompting will raise the singletons up to the level of the dyads. The interaction hypothesis does not make a clear prediction for the effect of engagement. However, it would probably predict a null effect because the main effect of prompting combines or collapses over the beneficial effects of interaction; thus, washing out any effects of the prompting procedure.

Interaction (Prompting x Engagement): The coverage hypothesis predicts that the prompted-explanation dyads should learn the same amount as the prompted-explanation singletons. The interaction hypothesis would predict that the dyads should learn more than the singletons. This assumes that both groups cover the same knowledge applications, as intended by the prompting manipulation. However, if the solos have less prior knowledge at their disposal than the dyads, then they may not be able to cover as many knowledge applications as the dyads. In this case, both the coverage hypothesis and the interaction hypothesis predict that dyads will gain more. The manipulation check mentioned earlier will be used to determine whether the solos or dyads cover roughly the same number of knowledge components. The materials will be designed to make this as likely as possible.

		Interaction
		Singletons	Dyads
Prompting	Natural	CH: Low Gain IH: Low Gain	CH: High Gain IH: High Gain
	Explain	CH: High Gain IH: Low Gain	CH: High Gain IH: High Gain

In vitro Experiment

We have, however, conducted an in vitro version of the study during the Spring 2007 semester. Unfortunately, we were unable to collect data from all four experimental conditions because our subject pool was extremely limited. Therefore, we limited the experimental conditions to the main treatment of interest: jointly vs. individually constructed explanations.

As in our first experiment, we used normalize assistance scores. Normalize assistance scores were defined as the sum of all the errors and requests for help on that problem divided by the number of entries made in solving that problem. Thus, lower assistance scores indicate that the student derived a solution while making fewer mistakes and getting less help, and thus demonstrating better performance and understanding.

The results from the laboratory study were as follows:

• Differences between conditions

The jointly constructed explanation (JCE) condition (M = .45, SD = .14) demonstrated lower assistance scores than the individually constructed explanation (ICE) condition (M = 1.00, SD = .15). The difference between experimental conditions was statistically reliable and of high practical significance, F(1, 23) = 7.33, p = .01, η_p² = .24.

• Problem by condition

The pattern observed at the level of conditions replicated at the level of problem. That is, when the problem was used as a repeated factor in a multivariate analysis of variance (MANOVA), the JCE condition demonstrated lower normalized assistance scores for all of the problems, except for the first, warm-up problem (see Table).

	ICE (n = 9)	JCE (n = 14)	p	ηp2
Prob1	0.75	0.63	.483	.024
Prob2	1.09	0.32	.003	.341
Prob3	1.08	0.51	.059	.160
Prob4	0.67	0.29	.034	.196

In addition to providing higher quality solutions, the jointly constructed explanation condition (M = 985.71, SD = 45.60) also solved their problems more quickly than the individually constructed explanation condition (M = 1097.75, SD = 51.45). Although the omnibus difference between experimental conditions was not statistically reliable, F(1, 23) = 2.66, p = .12, η_p² = .10, the differences in solution times for the second and third problem were reliably lower for the JCE condition. This finding is particularly interesting because the experiment was capped at two hours; therefore, the dyads were able to complete the problem set more often than the individuals. However, this result only approached significance, χ² = 22.91, p = .15.

• Knowledge component (KC) by condition

Because not all of the individuals were able to complete the entire problem set, their data could not be included in an analysis of the knowledge components. A MANOVA assumes that each individual participates in all of the measures. However, this is not the case when the individuals did not complete the last problem. Therefore, this fine-grained analysis of learning will need to wait until the study can be replicated in the classroom, with a larger sample size.

Explanation

This study is part of the Interactive Communication cluster, and it hypothesizes that the Prompting manipulation should increase the probability that the students follow the deeper knowledge construction monolog or dialogs, whereas prompting for "natural" monologs or dialogs will most likely result in shallow statements about the learning material. Deeper knowledge construction monologs/dialogs will include integrative statements that connect information with prior knowledge, connect information with or previously stated material, or infer new knowledge; whereas shallow monologs merely restate the presented material.

The Engagement manipulation should increase the likelihood that certain knowledge components are covered. That is, interactive dialogs (i.e., the jointly constructed explanation) should be more likely to cover knowledge components than non-interactive monologs (i.e., individual self-explanation).

Annotated bibliography

• Presented at a PSLC lunch: June 12, 2006 [1]

References

1. Dillenbourg, P. (1999). What do you mean "collaborative learning"? In P. Dillenbourg (Ed.), Collaborative learning: Cognitive and computational approaches (pp. 1-19). Oxford: Elsevier. [2]
2. Hausmann, R. G. M., & Chi, M. T. H. (2002). Can a computer interface support self-explaining? Cognitive Technology, 7(1), 4-14. [3]
3. Hausmann, R. G. M., Chi, M. T. H., & Roy, M. (2004). Learning from collaborative problem solving: An analysis of three hypothesized mechanisms. In K. D. Forbus, D. Gentner & T. Regier (Eds.), 26nd Annual Conference of the Cognitive Science Society (pp. 547-552). Mahwah, NJ: Lawrence Erlbaum. [4]
4. McGregor, M., & Chi, M. T. H. (2002). Collaborative interactions: The process of joint production and individual reuse of novel ideas. In W. D. Gray & C. D. Schunn (Eds.), 24nd Annual Conference of the Cognitive Science Society. Mahwah, NJ: Lawerence Erlbaum. [5]
5. VanLehn, K., Graesser, A. C., Jackson, G. T., Jordan, P., Olney, A., & Rose, C. P. (2007). When are tutorial dialogues more effective than reading? Cognitive Science. 31(1), 3-62. [6]

Connections

<insert here>