Predicting Learning from Text: A Comparison of Two Procedures
Much research focused on learning from text is concerned with the effect of level of information on recall. To determine information levels, hierarchical text analyses are used to parse texts, and a number of these have been shown to correlate with recall. This study compares two such procedures, the Johnson (1970) and Meyer (1975) procedures, to determine their power to predict recall of information from texts. In two experiments, a total of 194 university freshmen read one of two expository passages and wrote a recall protocol. These protocols were then analyzed by independent raters for the presence of information from the passages read. Johnson and Meyer analyses were used to examine recall patterns. The main effect of hierarchical level was significant for both analyses, but the mean percentage of recall increased as a function of hierarchical level as determined by the Johnson analysis and decreased as a function of hierarchical level as determined by the Meyer analysis. Implications derived from the study are discussed.