ignore or misrepresent the data. They tried to make sense of the outcome by acting as a theorist who conjectures about the causal mechanisms, boundary conditions, or other ad hoc explanations (e.g., shape) to account for the results of an experiment. In Chinn and Malhotra’s (2002) study of students’ evaluation of observed evidence (e.g., watching two objects fall simultaneously), the process of noticing was found to be an important mediator of conceptual change.
Echevarria (2003) examined seventh graders’ reactions to anomalous data in the domain of genetics and whether they served as a catalyst for knowledge construction during the course of self-directed experimentation. Students in the study completed a 3-week unit on genetics that involved genetics simulation software and observing plant growth. In both the software and the plants, students investigated or observed the transmission of one trait. Anomalies in the data were defined as outcomes that were not readily explainable on the basis of the appearance of the parents.
In general, the number of hypotheses generated, the number of tests conducted, and the number of explanations generated were a function of students’ ability to encounter, notice, and take seriously an anomalous finding. The majority of students (80 percent) developed some explanation for the pattern of anomalous data. For those who were unable to generate an explanation, it was suggested that the initial knowledge was insufficient and therefore could not undergo change as a result of the encounter with “anomalous” evidence. Analogous to case studies in the history of science (e.g., Simon, 2001), these students’ ability to notice and explore anomalies was related to their level of domain-specific knowledge (as suggested by Pasteur’s oft quoted maxim “serendipity favors the prepared mind”). Surprising findings were associated with an increase in hypotheses and experiments to test these potential explanations, but without the domain knowledge to “notice,” anomalies could not be exploited.
There is some evidence that, with instruction, students’ ability to evaluate anomalous data improves (Chinn and Malhotra, 2002). In a study of fourth, fifth, and sixth graders, one group of students was instructed to predict the outcomes of three experiments that produce counterintuitive but unambiguous data (e.g., reaction temperature). A second group answered questions that were designed to promote unbiased observations and interpretations by reflecting on the data. A third group was provided with an explanation of what scientists expected to find and why. All students reported their prediction of the outcome, what they observed, and their interpretation of the experiment. They were then tested for generalizations, and a retention test followed 9-10 days later. Fifth and sixth graders performed better than did fourth graders. Students who heard an explanation of what scientists expected to find and why did best. Further analyses suggest that the explanation-based intervention worked by influencing students’ initial