Abstract
This paper studies pupil performance on several scientific ‘variable‐handling’ questions presented to 15 Year 6 classes by way of a simple investigation. A descriptive analysis of the data is given; showing performance in relation to variables such as school, sex, and the context of the investigation. Some anecdotal evidence is offered to account for those schools which performed well above or below the mean for the whole sample. However, there is also a very marked ‘context’ effect on pupil performance; attributable to the type of investigation. Pupils’ appear to interact with question context in such a way that it will prove difficult to assess reliably a pupil's variable‐handling ability. Whilst it has proved possible to suggest that there may be some basic variable‐handling skills which pupils need to learn initially, there is a subtle, perhaps idiosyncratic, interplay between pupil and task which makes the interpretation of performance evidence far from straightforward. This, it is argued, has significant implications for both teaching and assessment strategies; particularly in relation to the National Curriculum for England and Wales.