Early entries and statistical predictions
In the past, students sat all of the assessments making up their GCSE at the end of a two-year course. Grade boundaries were set for each of the assessment components simultaneously using a combination of examiners’ expert judgement and statistics. This process allowed consistent qualification standards to be awarded year-on-year by taking account of the difficulty of the exams. The judgemental evidence has always been collected by scrutinising students’ work and the statistical evidence is presented in the form of a prediction of the percentage of students expected to exceed each grade threshold.
The prediction takes account of how the students performed in earlier assessments, and so reflects any changes in the ability of student cohorts between years. It is based on the concept of ‘value added’ – a measure of the progress students make between different stages of education. This approach assumes the value added is constant no matter what group of students within the cohort we’re looking at; the progress from their earlier educational assessments through to GCSE is assumed to be the broadly the same.
More recently the structure of GCSEs changed and most are now offered in a modular format. While the expert judgement used in standard setting is largely unchanged, modularity poses considerable challenges in the provision of statistical evidence, particularly when some modules are offered considerably earlier than others and a picture of the cohort's performance across the whole subject is still unknown. Module standards must be established with a view to maintaining standards across an entire subject in the future.
This short document considers a number of questions relating to early student entries for GCSE modules and the effect on statistical predictions.