When middle school students learn science content withgraphs, the graphing and science knowledge may be mutuallyreinforcing: understanding the science content may help stu-dents interpret a related graph, and information from a graphmay illustrate a scientific concept. We examine this relation-ship between graphing and science by studying how studentslearn from interactive computer models with accompanyingdata graphs. The computer models provide an animated simu-lation that illustrates an unobservable phenomenon, while thedata graph tracks one or more quantities over time. This or-dering study, on middle school students learning about photo-synthesis, indicates that engaging with novel graph conceptshelped students interpret their data as they experimented withthe computer model. The study also provided some supportfor the opposite direction: experimenting with the model firsthelped students make sense of the graphs.

People’s assessment of their ability varies in whether it is mea-sured once following a task or sequentially via confidencejudgments recorded throughout. Multiple models have beendeveloped to predict one-off judgments of performance, whichhave often distinguished between peoples’ biases about theirgeneral ability in a domain and their sensitivity to correctness.We propose a rational model of sequential self-assessmentwhich allows us to make predictions about each individualseparately—unlike in the one-off case which looks exclusivelyat the population level—and to identify, in addition to bias andsensitivity, the extent to which individuals’ beliefs are respon-sive to their most recent evidence over the course of a task. Wefit our model to data where participants solve algebraic equa-tions and show that bias, sensitivity, and responsiveness varymeaningfully across participants.

Self-assessment, or the evaluation of one’s ability on a task, is widely perceived as a fundamental skill, yet in most studies, people are found to be poorly calibrated to their own abilities. Some results seem to show poorer calibration for low perform- ers than for high performers. This effect has been explained in multiple ways: it could indicate worse metacognitive abil- ity among the low performers (the “Dunning-Kruger” effect), or simply regression to the mean. To tease apart these expla- nations we develop a Bayesian model of self-assessment and evaluate its predictions in two experiments. Our results suggest that poor self-assessment is caused by the influence of prior be- liefs and imperfect skill at determining whether a problem was solved correctly or not, and offer only weak support for of a relationship between metacognitive ability and performance.

Appraising one’s own performance after a task, known as self-assessment, has been studied from a cognitive science perspec-tive in domains such as humor, trivia, and logic. Previous stud-ies have found that participants are systematically poor at judg-ing their own performance, though sometimes self-assessmentvaries based on actual performance. We explored calibrationof self-assessment on algebra problems, a domain where peo-ple have typically received explicit instruction. In this domain,we found that people do not behave as they do in other do-mains previously studied: they are generally well-calibrated injudging their algebra performance. This suggests that in thecourse of learning to solve algebra problems, people have alsolearned to accurately judge their performance, both absolutelyand relative to others.

Three experiments investigate how self-generated explanation influences children's causal learning. Five-year-olds (N = 114) observed data consistent with two hypotheses and were prompted to explain or to report each observation. In Study 1, when making novel generalizations, explainers were more likely to favor the hypothesis that accounted for more observations. In Study 2, explainers favored a hypothesis that was consistent with prior knowledge. Study 3 pitted a hypothesis that accounted for more observations against a hypothesis consistent with prior knowledge. Explainers were more likely to base generalizations on prior knowledge. Findings suggest that attempts to explain drive children to evaluate hypotheses using features of "good" explanations, or those supporting generalizations with broad scope, as informed by children's prior knowledge and observations.

In 4 classroom experiments we investigated uses for technologies that automatically score student generated essays, concept diagrams, and drawings in inquiry curricula. We used the automatic scores to assign typical and research-based guidance and studied the impact of the guidance on student progress. Seven teachers and their 897 students participated. We documented the impact of guidance using pretests, embedded assessments, posttests, logged computer interaction data, and student and teacher interviews. We compared guidance designed to promote knowledge integration to 3 alternatives typically used in middle school classrooms. The knowledge integration guidance was more effective than generic guidance and specific guidance, and as effective as guidance designed by experienced teachers who also participated in professional development that emphasized knowledge integration. Results suggest that using automatic scores to assign knowledge integration guidance can provide an inquiry teaching partner: this guidance helps students use evidence to sort out ideas and can free teachers to support students who need extra help.