Abstract
The efficacy of the levels hypothesis (the prediction that ideas residing at superordinate positions in a text hierarchy are most memorable) was evaluated as a suitable descriptor of children's reading comprehension of expository prose. Third-and sixth-grade students read expository passages taken from extant curricular materials and responded to either a paper and pencil comprehension test (an open ended or multiple choice test) or produced an oral free recall. Results revealed that in only one instance (third-graders responding to the multiple choice test) did the children respond in the predicted manner: greater comprehension of main idea, gist information. The findings were interpreted as indicating little support for the levels hypothesis as a singular, accurate predictor of the information children will remember after reading expository prose. Other text-based and reader-based factors were discussed as possible explanations for the results. Implications for future research and educational practice were also discussed.