How can worked examples be designed to promote self-regulated problem-based learning? In two experiments students of the 9th grade attended in pairs (experiments 1 & 2) or alone (experiment 2) to physics and chemistry problems. In both experiments learning was supported by either continuous or segmented worked examples, the latter being obtained by segmenting the continuous examples into solution steps. In order to further promote the students’ elaboration on the worked examples each step in the segmented condition was preceded by a prompt that suggested auxiliary learning activities (e.g. strategic considerations and/or communication). Both prompts and solution steps (“feedback”) were given on separate sheets. Students self-regulated when to attend to the next prompt or solution step. Experiment 1 (N=60) revealed that segmented worked examples designed in this fashion were superior to more common continuous worked examples in measures of learning outcomes (retention, transfer) and the quality of communication within pairs. Communication quality, however, did not mediate learning outcomes. In order to explore if peer interaction contributes to the positive influence of segmentation with prompting anyway, we further varied in experiment 2 (N=146) if the segmented or continuous worked examples were attended in pairs or alone. The results confirmed the superiority of segmented over continuous worked examples while pair learners did not gain higher learning outcome scores than singles.

The goal of the research program in which the presented studies are settled is to advance the applicability of problem-based learning within science education. In particular we aim to design complex problem solving tasks to be more easily accessible for students. Within the research on learning from examples a set of design principles has proved to be effective for problem-based learning, e.g. structuring examples to emphasize a problem’s solution steps or subgoals and fostering task-relevant self-regulated activities (Atkinson, Derry, Renkl, & Wortham, 2000). The instructional material consisted of two problems in the domains of physics and chemistry education. The instructional goal was to support students in acquiring and elaborating scientific concepts and to connect these concepts to concrete applications in order to promote a more elaborated understanding. Each problem addressed a main concept (density, solubility) embedded into a tangible context (Is a 5 cent coin made of pure cupper? How can you determine the solubility of salts?). For each problem worked examples were created. The continuous worked examples were further divided into solution steps. The division into steps was done to emphasize a problem’s initial and goal states, subgoals and goal-directed operations. In order to instruct strategic learning behavior, each solution step was preceded by a prompt that suggested auxiliary learning activities (e.g. “Express the task in your own words” or “Remember: What is the formula for ‘density’?”). After having followed the instruction the students could request “feedback”, i.e. they received the next solution step of the problem (e.g. a re-formulation of the goal state, the formula for density, etc.). Prompts and feedbacks were written on sheets which were put in successively numbered envelopes. Students decided autonomously if and when to open the envelope with the next support.

Experiment 1 examined if this way of segmenting a worked example influences learning experience and learning outcomes positively. 60 students of the 9th grade participated in the experiment. Students worked in pairs which were tested separately and filmed during the problem solving phase. Half the pairs attended to a problem (“Cupper coin”) with a segmented worked example, the other half attended to the continuous worked example version of the problem. Aside from the prompts, continuous and segmented worked examples were equivalent in content. Learning outcomes for each student were measured by a retention test of the solution of the problem and a transfer test (concerning the concept of density). Learning experience was measured with questionnaires asking for the students’ motivation, feeling of competence and autonomy, attributions, etc. (Hänze & Berger, in press). Prior to the experiment we measured scientific knowledge, text comprehension ability, intelligence, scientific self-concept and goal orientation. Learning behavior (learning time, quality of communication) was derived from the video tapes. The two groups did not differ in any of the pre-experimental measures. Subjective learning experience as well as objective learning outcomes revealed significant differences. With segmented worked examples students reported a higher feeling of competence (t(58) = 3.85, p < .001, d = 0.98) and higher intrinsic motivation (t(58) = 2.86, p < .01, d = 0.73). They rated their own learning success higher (t(58) = 2.02, p < .05, d = 0.52) and the task as easier (t(58) = 2.42, p < .05, d = 0.56) compared to students who worked with the continuous example. Students attending to the segmented worked example also produced a more thorough retention of the solution (t(56) = 2.44, p < .05, d = 0.64) and reached higher scores in applying the concept of density in near transfer tasks (t(58) = 2.02, p < .05, d = 0.49). Although students in the segmented example condition attempted to explain the solution steps more often than in the continuous example condition (t(54) = 2.61, p < .05, d = .69), this verbal behavior did not mediate the effect on performance.

Experiment 2 was designed to explore if peer interaction contributes to the positive influence of segmentation with prompting anyway. In a 2x2 factorial design we varied the worked examples (segmented vs. continuous) and the group setting (single learners vs. pairs of learners). 146 students of the 9th grade participated in the experiment. For each of two problem-solving tasks each student was assigned to one of four experimental groups. Results confirmed the superiority of segmented over continuous worked examples for both problems. Students learning with segmented examples produced more thorough retentions of the solutions (F(1,138) = 6.25, p < .05, eta^2 = .04 for the cupper coin and F(1,142) = 4.27, p < .05, eta^2 = .03 for the solubility of salts task) and reached marginally higher scores in applying the concepts in near transfer tasks (F(1,138) = 3.84, .10 > p > .05, eta^2 = .03 for the cupper coin and F(1,142) = 3.26, .10 > p > .05, eta^2 = .02 for the solubility of salts task). There was no main effect of group setting and no interaction which fits to the lack of a mediator effect of communication on learning performance found in experiment 1.

Taken together, structuring worked examples by dividing its elaboration into steps and prompting strategic learning behavior before attending to the next solution step helps understanding the solution of a given problem and applying the underlying concept in near transfer tasks. The superiority of segmented over continuous worked examples appears to be mainly due to a better accessibility of the information given by the example. However, it cannot be ruled out that prompting interactions more explicitly would have caused more explanations and, thus, an even stronger effect. Further research is necessary to evaluate the contribution of each of the design features (segmentation, prompting) alone.

References

Atkinson, R. K., Derry, S. J., Renkl, A., & Wortham, D. W. (2000). Learning from examples: instructional principles from the worked examples research. Review of Educational Research, 70, 181–214.

Hänze, M. & Berger, R. (in press). Cooperative learning, motivational effects and student characteristics: An experimental study comparing cooperative learning and direct instruction in 12th grade physics classes. Learning and Instruction.