The Role of Prior Knowledge in Learning From Analogies in Science Texts

Abstract

Two experiments examined whether inconsistent effects of analogies in promoting new content learning from text are related to prior knowledge of the analogy per se. In Experiment 1, college students who demonstrated little understanding of weather systems and different levels of prior knowledge (more vs. less) of an analogous everyday situation read a text about weather systems that included the analogy or a control version that did not. Results indicated that those with more prior knowledge of the analogy performed better on weather system learning measures (sentence verification and number of concepts in essays). Prior knowledge of the analogous domain interacted with presence of the analogy in the text on 1 learning measure: Those with more prior knowledge who read the analogy text had fewer misconceptions in their conceptual models of weather than those who read the control text. Think-aloud protocols collected in Experiment 2 suggested that analogies in the text constrained prior knowledge activation and processing of the weather system content. Whereas previous research has shown that prior knowledge of a to-be-learned target domain positively impacts learning, this research elaborates this effect by showing that prior knowledge of an analogically related domain positively impacts target domain learning.

Notes

¹From the participants' point of view, Phases 2 and 3 were independent experiments: They had different study numbers, different locations, and different researchers conducting the work.

^a n = 27.

^b n = 30.

^c n = 24.

^d n = 25.

²Two outliers were removed from the analysis because their reading times exceeded 3 SDs above the mean.

³Pre- and posttest knowledge scores were not significantly correlated, warranting the use of gain scores.

⁴The significant correlation between number correct on pre- and posttest for this set of 22 items (r = .211, p < .05) necessitated the use of an analysis of covariance.

^aMarginal significance.

^bSignificance.

⁵It should be noted that the proportions do not add to 1, as the denominator was of total self- explanation inferences generated.