Exam boards told Ofqual it was too closely involved with this summer's controversial GCSE grading scandal to investigate the process, with at least one board calling for an independent inquiry, emails seen by TES reveal.
Correspondence also shows the regulator was warned more than five months before results day that there were problems with the system it was using to combat grade inflation. Three boards eventually deemed the statistical approach unreliable and four out of the five exam boards had difficulty producing GCSE English results that satisfied Ofqual's demands.
Nearly 150 pages of letters, emails and minutes released under the Freedom of Information Act suggest that Ofqual struggled to stay on top of the unfolding crisis (see below). Heads claim that many thousands of pupils have missed out on C grades because of a dramatic shift in GCSE English grade boundaries between January and June.
The correspondence shows that by the end of July, exam boards were considering how they could "defend" an unlikely set of results. Despite being more lenient than the statistical predictions Ofqual had ordered, grades were tougher than in 2010 and 2011 - Edexcel grade Cs had fallen by 11.4 percentage points.
As TES went to press, an alliance of teaching unions, local authorities and schools were waiting for responses - expected yesterday from Ofqual and the AQA and Edexcel boards - to threats of legal action over the results.
But school leaders want a full independent inquiry, even if they are granted a regrading. "If we are to avoid making the same mistakes again we need someone independent to get to the bottom of what happened," said Russell Hobby, general secretary of heads' union the NAHT.
TES has learned that the Commons Education Select Committee will not decide for at least another 10 days whether to hold more hearings as part of its so far limited inquiry into the affair.
Ofqual has conducted the only other investigation. But emails show exam boards expressed misgivings about the regulator's suitability for the job on the day it was announced.
A message from Welsh exam board WJEC on 25 August reads: "One question which I think needs to be considered . is whether there are aspects of this investigation which need to be handled by someone independent of the regulators."
Earlier the same day, Pearson - parent company of the Edexcel board - emailed Ofqual, saying: "My concern is that you are in effect looking at whether grade boundaries are correct. The public will understandably see this as an investigation. I see this as surprising given Ofqual was party to their setting and we will say so."
The correspondence confirms that the watchdog was heavily involved and ordered both WJEC and Edexcel to over-rule their examiners' judgements.
It also shows that AQA and OCR both found grading GCSE English "difficult", with OCR warning on 17 July that it could not meet statistical predictions. On the same day, Ofqual told boards it would be "very concerned" if there was "an increase at grade C nationally". The previous week, it said it "could not accept any apparent grade inflation in GCSE English", more than three weeks before Edexcel offered "compelling evidence" to justify a rise.
Problems with Ofqual's "comparable outcomes" approach to reducing grade inflation emerged on 13 March when WJEC told the watchdog it had "some difficulty" with the statistical predictions.
An Ofqual spokesperson said: "The `comparable outcomes' approach was developed in association with exam boards and other leading assessment experts. All of the exam boards and regulators have agreed the approach. The correspondence between Ofqual and the exam boards shows proper regulatory processes and exchanges."
Did Ofqual fully understand what it was dealing with?
On 27 July, Ofqual warned exam boards of "likely" grade inflation in GCSE English, but by 22 August was saying the "most likely reason" for a drop in higher grades was the exodus of 20,000 independent and grammar pupils.
On 20 September, Ofqual's website claimed there were "not big changes" to the GCSE English cohort, which was "similar enough" for comparable outcomes to be valid.
An email from Edexcel on 8 August shows Ofqual thought a large rise in the English grade C boundary would produce results tougher than statistical predictions. The board had to point out they were still more lenient.
A later exchange with Edexcel reveals the regulator tried to explain a mismatch between results and statistical predictions with a withdrawn pilot qualification never used in the predictions.
Original headline: Exam boards told Ofqual it was too close to scandal