| One aspect of school success is children's ability to make inferences from reading material. Most research has focused on making inferences from small sets of sentences or from fictional prose (e.g., Paris & Lindauer, 1976; Thorndyke, 1976). However, only one study (Pearson, Hansen, & Gordon, 1977) has utilized nonfictional prose to assess how subjects' prior knowledge of material affected their ability to make inferences from it. This study extended Pearson et al. by comparing subjects at different grades with different types of nonfictional content.;Analyses of variance were conducted on two dependent variables: (a) the percentage of correct responses to the explicit and inference questions, and (b) the logarithms of the response times (RTs) to the questions. As expected, the percent correct increased with age. In addition, the procedural articles were easier than the science articles, which were easier than the social science. While high knowledge articles were easier than low knowledge in both the science and social science conditions, no such difference was found for the procedural articles. In general, the greater the percent correct, the faster the response times to the questions. Contrary to prediction, there was no difference in percent correct between the explicit and inference questions; however, the latter had significantly longer RTs than the former questions.;These results support various schema theories of knowledge (e.g., Rumelhart & Ortony, 1977). In general, questions on topics for which subjects had a schema (high knowledge) were answered more quickly and accurately. Some schemas (e.g., the procedural-cooking) seemed general enough for subjects to use for answering questions about both familiar and unfamiliar topics. In addition, the longer RTs to inference questions indicate that such inferences require more processing. The educational implications of these data were also discussed.;Twenty-four females from each of three grades (third, sixth, and eighth) were evenly divided among three content areas (procedural, science, and social science). Each subject read two 400-word articles on topics related to her content area. One article was on a topic the subject knew well (as evidenced by a multiple-choice pretest), while the other was on a topic not known well. After reading each article, the subjects were orally presented with six questions assessing material explicitly presented in the article and six questions requiring subjects to make an inference to answer. The information required to answer each explicit question was also necessary to answer one of the inference questions. Both passage order and question order were controlled by counterbalancing. |
