Rule induction on word problems as a function of learner characteristics and task variables
Date
1987
Authors
Steacy, Nancy
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The rule induction rate and ability of Grade 6 students on word problems were assessed. Two classes (N = 57) in a suburban elementary school were administered 144 word problems in three sets. The problems were all isomorphic to each other and, if recognized as such, could be solved using the same procedural rule. The problems varied in whether or not they contained extraneous information and, when present, in the location of that information within the problem context. They also varied in the complexity of the sets involved and in the language used to establish that complexity. Lastly, the problems contained all real words, nonsense nouns, or nonsense nouns and verbs as a way of manipulating the level of language familiarity.
The subjects were grouped according to their performance on measures of three learner characteristics, academic self-concept, test anxiety, and memory span, and according to gender for one comparison. Subjects performed differentially on the total problem set in terms of both rule induction rate and ability. Individual performances over the three problem sets were observed to follow one of three different patterns wherein performance either was maintained at an uniformly high level, improved dramatically, or was uniformly low. The influence of learner characteristics seems to have over-powered any effects of the task variable manipulations. Two separate performance groups were differentiated by the memory span measure. These groups remained distinct regardless of changes in the task. The word problem performance of subjects with low academic self-concept was almost always differentiated from that of all other subjects. The performances of high and low test anxiety groups were sometimes distinct, though neither were ever significantly different from a middle group. Finally, memory span was a reasonable predictor of rule induction rate, and consequently, its predictive power decreased over the three problem sets. Academic self-concept was a moderate predictor of rule induction ability over the three trial tasks. Level of test anxiety was essentially unrelated to rule induction ability or rate on the isomorphic word problems.