Assessing pupils’ ability to generalize∗

Abstract

The paper presents findings from the British national science monitoring programme operated by the Assessment of Performance Unit. It focuses on 15‐year‐old pupils’ interpretation of presented data and self‐generated observations and, in particular, on their ability to arrive at generalized conclusions when requested so to do. The responses of a national sample of pupils are analysed for the effects of gender, question type, question and ‘ability’ (using the surrogate of likely examination entry). Significant differences are found between responses based on practical situations, as opposed to pencil‐and‐paper questions, and some interaction with ‘ability’ is noted. Data are presented in relation to individual questions, in order to explore the variations observed within questions of the same type. The least able pupils were found to have difficulty with the basic phenomenal reference of even simple data. The difficulties of able pupils tended rather to be associated with the elimination of potential relationships from complex multivariable data. Questions based on material resources are found to present particular problems of validity and reliability. The paper concludes with a discussion of some implications of these findings.