While it is clear from a theoretical point of view that only accurate task interpretation can lead to successful self-regulated learning, empirical studies show insufficient metacognitive monitoring, especially for ill-defined and complex tasks. Based on the framework of the COPES-model of self-regulated learning (Winne & Hadwin, 1998) we devised a series of studies that investigate the issues of calibration to task complexity as well as the impact of epistemological beliefs in detail. The first study focused on the preparatory stages of task definition, goal setting and planning and revealed that students do calibrate their judgments to task complexity and that these metacognitive calibration processes are influenced by epistemological beliefs. The results of the second qualitative study focusing on the enactment stage will be presented here. It is an open issue, if students actually do what they plan to do in the preparatory stages. More specifically, we investigated, if students differentiate between tasks of different complexity, use adequate strategies (calibration) and if these calibration processes are impacted by epistemological beliefs. This more qualitative study consisted of two sessions: during the first session students filled in online-questionnaires about learner variables such as epistemological beliefs. During the second session all students (14 biology students and 21 humanities students) were familiarized with our hypertext about genetic fingerprinting (107 nodes) and subsequently had to solve learning tasks of different complexity according to Bloom’s revised taxonomy (Anderson et al., 2002). During task completion, logfiles were collected and additionally students were prompted to elaborate on their concurrent thoughts. Retrospective interviews were conducted to get more detailed insights. Preliminary results indicate that students adapt their learning strategies in a meaningful way to tasks of different complexity. Analyses with regard to epistemological beliefs are still pending.

Students are confronted with a variety of learning task. Only accurate task interpretation can lead to successful self-regulated learning. Opposed to this claim, empirical studies show insufficient monitoring. For example, calibration between metacognitive judgments and objective indicators of learning is suboptimal, especially in ill-defined and complex tasks (e.g. Wiley, Griffin, & Thiede, 2005). Furthermore, based on the COPES-model (Winne & Hadwin, 1998) it can be assumed that task interpretations as well as subsequent self-regulatory activities are also impacted by internal conditions that serve as standards for metacognitive monitoring and control. With regard to epistemological beliefs, i.e. learners’ beliefs on the nature of knowledge and knowing, it is generally assumed that more “sophisticated” beliefs in complex, uncertain and variable knowledge are associated with better learning processes and higher learning outcomes, presumable because they determine how flexible students are in adapting their learning to task demands. Within a comprehensive series of three studies we investigate these issues of calibration to task complexity and the impact of epistemological beliefs in more detail. All our studies tap different stages of self-regulated learning as specified in the COPES-model. The first study focused on the preparatory stages of task definition and goal setting and planning. When learners were presented with tasks of different complexity and had to answer multiple questions for each task, their judgments revealed discrimination, calibration to task complexity as well as a significant impact of epistemological beliefs (Stahl, Pieschl, & Bromme, 2006). For example, learners set more superficial goals and planned the use of fewer and more superficial learning strategies for simple tasks, whereas they set more elaborate goals and planned the use more deep-processing learning strategies for more complex tasks. The second study that will be presented here presents a qualitative analysis of the whole learning process with a strong focus on the enactment stage. As empirical studies, e.g. concerned with the validity of questionnaires on metacognition (Winne & Jamieson-Noel, 2002), consistently show a low fit between what students say they do and what they actually do, it is an open issue if the good calibration that we found within the preparatory stages can also be detected in students’ enactments. More specifically, we pose the following research questions: (1) Do students discriminate between tasks of different complexity, i.e. by using different strategies for different tasks? (2) Do students calibrate their learning process to task complexity, i.e. do the enacted strategies fit the objective complexity of the task? (3) Are these metacognitive discrimination and calibration processes impacted by students’ epistemological beliefs?

Method:

This qualitatively focused study consisted of two sessions: during the first session all students (14 biology students, 21 humanities students) filled in online-questionnaires: epistemological beliefs were measured with three domain-dependent instruments that were answered with regard to knowledge in (molecular) genetics, the CAEB (Stahl & Bromme, submitted), the DEBQ (Hofer, 2000), and the GCBS (Trautwein, personal communication, June 12^th, 2006). Based on their responses, students were sorted into two matched sub-samples. Within the second phase, one sub-sample received a neutral instruction that encompassed a factual introduction to genetic fingerprinting; the other sub-sample received a sophisticated instruction that was enriched with comments about the epistemological nature of selected facts. As a treatment-check the epistemological beliefs questionnaires were re-administered. Subsequently, students were introduced to navigational options and the structure of our hierarchical hypertext about genetic fingerprinting (107 nodes, see Appendix). Sub-chapters describe three methods for DNA analysis: STR (short tandem repeat) analysis, Y-STR analysis (on the Y chromosome) and mtDNA analysis. Nodes are arranged on five hierarchical levels. While nodes on higher levels are designed for novices, deeper-level nodes are comparable to scientific articles. Besides the hierarchical structure, thematic links lead to further elaborations on the biological background, further examples or elaborations on problems or uncertainties. In the main part of the study students were presented with 3 – 5 learning tasks (depending on speed of task completion) that had to be solved with the hypertext. Logfiles were collected automatically. The tasks represent different categories of Bloom’s revised taxonomy (i.e. in ascending complexity: remember, understand, apply, analyze, evaluate and create; Anderson et al., 2002). Two multiple-choice tasks that represent the category remember were followed by an evaluate task that required students to make independent judgments on the adequacy of the Y-STR-analysis for paternity testing and required them to suggest the best DNA analysis method for that purpose. Subsequent tasks were facultative. During task completion students were prompted every two minutes (“What are you currently thinking about?”) to capture their ongoing cognitive and metacognitive processes. After 1 hour, retrospective stimulated recall interviews were conducted for two of the learning tasks, i.e. one remember task and the evaluate task. First, students were asked in detail about their self-regulated learning process for the respective task, i.e. what was their understanding of the task (task definition according to the COPES-model), what were their goals and plans (goal setting and planning), what had they actually done for task completion (enactment). Second, they were asked about the epistemological nature of the underlying knowledge.

Results

We just finished data collection, thus only preliminary results can be reported: students discriminated between tasks (question 1) and showed calibration to task complexity (question 2). For example, students spent more time on the complex task, accessed more hypertext nodes to solve it and used different strategies for task completion. More specifically, for solving the multiple-choice remember task students often used the search function to quickly find appropriate information; subsequently they checked the answer that best matched the text. For the complex evaluation task on the other hand, they often systematically explored the hypertext (e.g. by navigating left-to-right and top-to-bottom through the hypertext chapters) to gain at least a superficial understanding of the presented DNA methods. Nonetheless, contradictory results were also found: some students considered the evaluate task simpler than the remember task because they applied different criteria. More detailed analyses as well as analyses about the impact of epistemological beliefs are still pending but thorough results will be presented at the conference.