Effects of diagram complexity on comprehension processes and learning outcomes

Two experiments investigated the effects of diagrams on comprehension processes and whether the effectiveness of diagrams is influenced by diagram complexity. In Experiments 1 and 2, students learned about the human heart and circulatory system using either text only, text with simple diagrams, or text with more detailed diagrams. Students' mental models and general knowledge of the domain were tested before and after learning. Post-learning measures also included a memory test of text details and inference questions related to the text. In Experiment 1, students studied the learning materials spontaneously. Students in both diagram conditions tended to improve their mental models more than students who used only text materials, but students who used simple diagrams with text were most likely to form the correct mental model of the domain. Simple diagrams were most beneficial to students for the general knowledge and memory measures; students using detailed diagrams performed only about as well as students who received only text. In Experiment 2, students self-explained (Chi, 2000) the materials during learning. In addition to learning measures, self-explanations generated by students were analyzed. Again, students in both diagram conditions improved their mental models more than students who received only text materials. Students using simple diagrams outperformed the other groups in tests of general knowledge and memory. Notably, students in both diagram conditions generated significantly more inferences than students who received only text during learning. Two types of self-explanation inferences, inferences related to the path of blood through the heart and circulatory system and inferences that indicated the integration of previous and current information, were tied to formation of correct mental models of the domain. Students in both diagram conditions were more likely to generate inferences related to the path of blood than students in the text condition, but only students using simple diagrams generated more integration inferences during learning. The results indicate that the addition of diagrams to text does influence comprehension processes but that diagrams of different complexity are not equally effective. The effectiveness of diagrams appears to be related to their efficiency in guiding important cognitive processes during learning.