Responses pattern to conceptual questions on Newtonian Mechanics: implications for learning

Abstract

The responses pattern to the Force Concept Inventory (FCI) is a key discriminator in the debate between coherence and pieces in the private initial knowledge of a subject: the different hypotheses about the structure of such knowledge lead to verifiable predictions. Previous works have made descriptive studies of those patterns, while another, spotting the occurrence of very short term processes that determine responses, have made use of statistical inference. In order to make a meaningful comparison between those works, we have used the same statistical inference tests in the item analysis of the FCI: for each “misconception” from the taxonomy used in the design of the FCI, contingency tables were done for each couple of items showing options com-patible with it, and Fisher test was performed on each (143 for "misconceptions" and 161 for Newtonian dimensions). It was observed that for the vast majority of pairs the hypothesis of independence could not be rejected, so "misconceptions" are a very weak organizer, unlike the unconscious reflex decisions above mentioned. Some possible consequences of this perspective for teaching are briefly considered.