Abstract
This article reports on a study carried out into the context dependency and construct validity of practical skill assessment in science. The skills concerned are observing, manipulating, planning, interpreting, reporting and self‐reliance. The literature relating to context dependency of practical skill assessment is reviewed as is that relevant to the construct validity of practical skills. Pupil performance in observation and reporting skills appeared to be context‐dependent, while, in contrast, performance on interpretation and self‐reliance were generalisable. The study further compares pupil performance on a range of practical tasks with external examination grades awarded in biology and chemistry. The findings indicate that performance on few of the skills correlates highly with the external examination grades awarded to pupils. This further evidence of the construct validity of practical skills supports the view that practical work should be given a prominent and discrete place in the assessment of science subjects.