Why are SQA results askew?

Calculated grades were always a risk, but the SQA results show a troubling pattern. David Kernohan plots it.

The first Tuesday in August marks the start of the Clearing season.

In most years, the release of Higher and Advanced Higher results by the Scottish Qualifications Authority sees applicants finally able to make decisions about the next three or four years of their life. This, of course, is not “most years”. This is 2020. So there’s evidence of a huge socio-economic skew in mark moderation, and there will be many appeals – leaving a lot of 2020 university applications up in the air.

Higher and advanced

As widely trailed, results overall are up year on year in Scotland – for A-C grades results are better by four percentage points for Highers and five percentage points for Advanced Highers. It’s a peculiar situation for an applicant to be in. In pure terms, most have done better than preceding cohorts – but this is tempered by an under-performance against teacher estimates. Results have been lowered dramatically from these initial grades, by an average 10 percentage points for Highers and 8 for Advanced Highers, though as we will see these figures hide some startling splits.

In a regular year this under-performance could be put down to an actual performance – in the pressures of the exam hall – in 2020 the effect has been simulated statistically. Unlike in England, Scottish students disappointed by their grade can – and likely will – make an appeal via their school or college where their final grade differs from the estimate.

Skewed by SIMD

To say these results have been controversial is to understate things. Here’s a scatter graph comparing the percentage point difference in entrants with a given characteristic estimated and awarded a particular letter grade, using data taken from the equality impact assessment.

[Full screen]

You can see plainly that students from the least advantaged areas (SIMD4 quintile 1) are disproportionately more likely to have been moved down, especially in the lower reaches of the grades (we don’t get the fine grain, alas, just the letters). There’s a general uplift towards the higher achieving end (more As are awarded than predicted in all groups) with students being moved away from Bs and Cs in both directions. The NA (award not passed) differences are particularly worrying, more “NAs” were awarded than estimated in SIMD4 Q1 and Q2, and less in Q3, Q4, and Q5.

Though estimates were only adjusted where necessary, 26.2 per cent of grades were adjusted in some way – with around 9 out of 10 of these adjustments bringing the awarded grade down. Moderation took place at centre level – with comparison to recent years of results informing the final grades. But for new schools and schools presenting courses for the first time, the estimates were just directly converted into grades!

Older readers may recall that measures of social advantage do predict lower actual and predicted grades. Looking at A levels a couple of years back I found a reliable but static difference between actual and predicted grades that skewed near identically to lower grades for students with disadvantaged backgrounds. Here’s a version of that plot for the SQA results.

[Full screen]

The Scottish Government response has been bullish, with John Swinney highlighting that:

Without moderation, pass rates at grades A-C compared to last year would have increased by 10.4 percentage points for National 5, by 14 percentage points for Higher and by 13.4 percentage points for Advanced Higher – annual change never been seen in Scottish exam results. I know teachers and lecturers will always want the best for their pupils but I believe that teachers have acted professionally

For me, I think we are seeing a schools inequality effect. Less advantaged pupils will be likely to live in areas where schools have historically performed less well in national exams – I’ll leave whether such exams are in fact a great way of testing how middle class you are as an exercise for the reader (though the data does point to this as a possible explanation).  No matter how well pupils would have done this year (or how much a school might have improved), historic underperformance seems to have dragged calculated grades down.

Bonus graph

For completists, here’s the 2020 and 2019 SQA results and entry rates by subject.

[Full screen]

What about going to university?

With such an enormous number of likely appeals there’s a whole new variable in understanding university this year. On top of all the other whole new variables, like the global pandemic, and suchlike. This makes reading the first iteration of the UCAS Clearing update very difficult.

The good news for the sector is that there are slightly more Scottish applicants in the system than last year (47,770 – up around 1 per cent). However, substantially more of these students appear to be holding an offer (7,360 – up 15 per cent) rather than holding a place. The definitions here are critical: if you are “holding an offer” on the UCAS system you have not yet had your results verified against the conditions attached to that offer.

In normal, non-2020, years you would expect that on the release of the results that these accepted conditional offers would be confirmed straight away. This time round, providers are likely to be spending longer in making application decisions (bearing in mind concerns over calculated grades), and many non-SQA qualifications – for example professional registration examinations, or HNCs – will have delayed results.

SIMD4 quintile 5 (most advantaged) applicants are more likely to be placed – 39 per cent of 18 year olds have been placed, compared with 11.9 per cent of quintile 1. The direction of travel is good (there’s a one per cent year-on-year increase for Q1), but there’s still clearly a lot to do – and this year’s SQA results will not help.

In terms of subject, placed applicants studying subjects allied to medicine are up 5 percent in Scotland on this time last year, with placed students studying Education up 11 per cent. And Scottish domiciled students are more likely both to be placed at a provider in Scotland (up 1 per cent) and to hold an offer at a provider in Scotland (up 16 per cent).

Despite what was expected, there has been a slight (8 per cent) drop in applicants free to be placed in Clearing. Again, this could be explained by uncertainty over marks pending appeal. But a decline in applicants placed at their insurance offer suggests that the ongoing restrictions on offers with unconditional components are having an impact.

What does this tell us about next week?

Unlike Scotland, there are no appeals against calculated grades in England. Students would have the option to resit in the autumn, but this would render them unable to take up a place in September or October. A similarly acute and obvious issue with A level grades would require political intervention to resolve.

The presence of this anomaly in Scotland means we can learn much less about the way the 2020 cycle will play out in the rest of the UK. It’s fair to assume that the resit option could play out in England as a rise in applicants removing themselves from the UCAS system, but we don’t get direct data on that until the cycle is complete.

We see one interesting piece of data about England that is worth bearing in mind as we await the UK’s largest cohort of 18 year olds. Proportionally less cycle entrants are placed at their firm or insurance offer at this stage of the cycle, an artifact of the ban on most unconditional offers.

Leave a Reply