POLAR opposites – LEO and the class effect on graduate earnings

Image: IKON

It may come as no surprise that there is a link between the number of poor students admitted to university and average graduate salaries.

Last year’s detailed IFS/Nuffield study of the first ‘linked data’ on graduate salary outcomes demonstrated that graduates from wealthier backgrounds continue to have a large earnings advantage even after we allow for differences in their university experience.

This week’s LEO release gives us an opportunity to understand how this social class effect on graduate outcomes might vary between different subjects and universities. We’ve made just a small start in this area by picking out a few subjects to examine in more detail. The findings suggest that even though it may be a more level playing field in terms of accessing higher education, social class and its relationship to the type of institution and subject at which you study, mean it is certainly an uneven landscape when leaving it.

For example, Mathematical Sciences sees a mean gap of £12,640 in three-year graduate median earnings between those courses that admit the most POLAR quintile 1 (most disadvantaged) students and those that admit the least.

At five years after graduation, the relationship begins to become even more clear, as we can see in the graph.

Meanwhile, English Studies has a much smaller gap of (only) £2860 in three-year graduate median earnings between those courses that admit the most POLAR quintile 1 (most disadvantaged) students and those that admit the least. We can see from this graph (for data five years after graduation) that the trend line is a little less steep. 

The inclusion of the POLAR data enables us to make a very crude assessment of whether different courses are ‘over’ or ‘underperforming’ in terms of social mobility. Those above the trend lines are producing graduates with slightly higher than expected salaries given the class make-up of their intake. We could (had we more time) produce a graph like this for every subject area.

This is not the kind of sophisticated benchmark that would satisfy HESA, and we hope that stats wonks with far greater skills than ours are busy at work creating proper class benchmarks for LEO outcomes. Doing so is essential to really understand what LEO can tell us about higher education’s impact on social mobility, or perhaps, the real lack thereof in the UK today.

Method note – the usual caveats with POLAR apply. POLAR is a less reliable measure of social advantage for London and has not been used in this release for Scottish institutions. 

Leave a Reply