| The performance of seven formal and seven concrete operational high school biology students were video taped during a three-phase learning sequence in water pollution. Subjects explored the effects of five independent variables on two dependent variables over time using a computer simulation program. Predictions were made concerning the effects of the independent variables upon dependent variables through time. Subjects were identified according to initial knowledge of the subject matter and success at solving three selected prediction problems.;The primary goal of this study was to examine the science process skill of prediction, in detail, using naturalistic research methodology and information-processing theory. The think-aloud interview technique, modeled after Ericsson and Simon, led to the identification of 63 behaviors related to program exploration and prediction. Systematic comparative analyses were performed on the behaviors of different comparison groups (e.g. concrete versus formal). Procedural models, depicting cognitive pathways, branch points, feedback loops, and behaviors of thought associated with prediction were developed from sequentially encoded transcripts.;Successful prediction behaviors involved systematic manipulation of the independent variables, note taking, identification and use of correct independent-dependent variable relationships, high interest and motivation, and in general, higher level thinking skills. Behaviors characteristic of unsuccessful predictors were non-systematic manipulation of independent variables, lack of motivation and persistence, misconceptions, and the identification and use of incorrect independent-dependent variable relationships. Successful predictors generally had high initial knowledge of the subject matter and were formal. Unsuccessful predictors generally had low initial knowledge of the subject matter and were concrete. High initial knowledge seemed to be more important to predictive success than stage of Piagetian cognitive development.;The results of this study have implications for problem-solving theory, cognitive development, and science teaching. The procedural models, which trace the behavioral patterns of prediction, may prove useful to computer modeling of the prediction process. |
