The effect of question choice on performance in a GCSE English Literat
ure examination was investigated using a multilevel model. Although th
ere were statistically significant question effects, these were small.
For the majority of candidates in the study, the difference was less
than +/-1 mark. There was no evidence that candidates were seriously d
isadvantaged by their choice of texts or questions. This was expected
because the underlying criterion-referenced mark scheme used was indep
endent of the questions.