To ask the Secretary of State for Education, what assessment his Department has made of the adequacy of published data collection on achieved grades against predicted grades for (a) GCSEs and (b) A-Levels by demographic group; and what steps his Department is taking to improve such data collection increase transparency to combat unconscious bias.
To ask the Secretary of State for Education, what recent discussions his Department has had with UCAS on the (a) role of unconscious bias in predicted A-Level grades and (b) potential effect of unconscious bias in those grades on trends in admissions statistics in relation to (i) socio-economic background, (ii) race, (iii) gender and (iv) disability.
The data that has been collected by UCAS on the relationship between predicted and achieved grades relates only to those who apply to higher education using UCAS.
There is no published data on predictions for GCSEs.
UCAS published a report on the factors that are associated with the differences in predicted and achieved A level attainment in 2016 examining the predictions and results for 600,000 English 18 year old applicants between 2010 and 2015 with three or more A levels.
The report is available here: www.ucas.com/file/71796/download?token=D4uuSzur.
Black applicants were proportionally 19% more likely to be overpredicted compared with White applicants. Disadvantaged applicants (measured using POLAR) were 5% more likely to be overpredicted compared with the most advantaged applicants.
We have not had discussions with UCAS about historic data tables of achieved grades against predicted grades by demographic group. UCAS have published data on the differences in predicted and achieved A level points for 18 year old UK applicants with at least three predicted A level grades. The data is available from 2010 and by gender, domicile, disadvantage and ethnicity.
In the 2019 application cycle, of UK 18 year old applicants with at least three predicted A levels who were accepted onto a place, 79% of predicted grades were overpredicted and 8% underpredicted.
We have had no recent discussions with UCAS about the potential role of unconscious bias in predicted A level grades, as this would not be a matter they could control. We have had extensive discussions with Ofqual about this matter and Ofqual publications set out how it has been taken into account in the development of the calculated grades being awarded this summer.