In Defence of Liberty

Driven by data; ridden with liberty.

Statistics and Lampposts IX: Independent Schools and Dependent Variables

The BBC headline beamed that “state pupils do better at university, study shows”. The study in question was the Higher Education Funding Council for England’s (Hefce) research entitled Differences in degree outcomes. The Hefce report followed the outcomes of 130,000 students beginning their degrees in 2007, considering their educational background and other factors. The key graph showed the propensity to gain first-class or upper second-class degrees, sorted by A-Level attainment and school type – ‘state’ or ‘independent’ schools. It showed that, given the same A-Level grades, students arriving from state schools achieved a greater percentage of top degrees than their privately-schooled counterparts.

This graph has been widely shared, appearing to show that state-schooled students do better at university than their independently-schooled counterparts. (Photo: Hefce)

This graph has been widely shared, appearing to show that state-schooled students do better at university than their independently-schooled counterparts. (Photo: Hefce)

In the lexicon of statistics, the school type and A-Levels are independent (or explanatory) variables, and the tendency to obtain a top degree is the dependent (or response) variable. However, in this case, the ‘independent’ variables are not statistically independent: one affects the other. Indeed, parents are often rather interested in how dependent educational attainment before university is on their choice of school.

Whilst Times Higher Education’s headline was more accurate – State pupils on same grades as private counterparts ‘get better degrees’ – the BBC headline is misleading. The headline is confusing precisely because the two explanatory variables are non-independent. The graph does not imply that state-schooled students “do better at university”; because it does not show what proportion of state-schooled students achieve each classification of A-Level grades or degrees. As Table 4 in the Hefce report shows, university students coming from independent schools have a median A-Level tariff – that is, a score based upon the grades obtained – of 320, compared to a median tariff of 280 for entrants from every other school type. Of independent-schooled students that made it into university, 67.0% gained either a first-class or upper second-class degree. Alternatively, 58.8% of graduates from sixth form schools or further education colleges gained these top degrees classifications. Thus, as a cohort, university students arriving from sixth forms or standard colleges do worse than their independent school compatriots.

This is the propensity of accepted university entrants to gain 1:1 or 2:1 degrees, by school type. (Photo: Hefce)

This is the propensity of accepted university entrants to gain 1:1 or 2:1 degrees, by school type. (Photo: Hefce)

So what is going on?

School performance rankings are determined by what was achieved in GCSEs or equivalent qualifications. It is often suggested that the top schools only achieve their ranking through surreptitious selection of the best pupils. If it were the case that schooling has no effect on grades, then the expected outcome would be there should be no difference between these propensities for academic achievement after leaving school. However, the opposite is true. The Hefce report demonstrated a similar effect of school performance on degree achievements, when university students with equal A-Level qualifications were compared. Therefore, these schools are enhancing the pupils’ academic abilities, whilst other schools fail to stretch their pupils.

There is a statistically significant difference between propensities of students coming from schools in the top quintile and all others. (Photo: Hefce)

There is a statistically significant difference between propensities of students coming from schools in the top quintile and all others. (Photo: Hefce)

It is concordant through all the graphs displayed in the Hefce report that A-Level grades are a good explanatory variable for success in university degrees. This data suggests that British universities will continue to choose their admissions primarily based upon academic achievements prior to entry. However, other data, such as performance during interviews, is likely to remain important to universities picking their students.

Advertisements

Information

This entry was posted on April 6, 2014 by in Statistics and tagged .
%d bloggers like this: