Primary teachers have "an almost blind faith" in the accuracy of exam grades in contrast to secondary colleagues who believe they can be unpredictable and unreliable, according to research published by Ofqual.
As part of a study commissioned by the exams regulator into public perceptions of reliability, two focus groups of teachers from a primary and a secondary in Greater Manchester were asked their opinions of the assessment system.
"I (did my exams with) the Joint Matriculation Board and they're this big power and you just respect the authority. I didn't doubt my results at all," one primary teacher told the researchers from the AQA exam board.
They also asked the opinions of employers, students, NHS workers and jobseekers, and concluded: "The non-teaching participants and, to a lesser extent, the primary school teachers, tended to have an almost blind faith in the ability of the assessment process to award accurate grades," the report says.
But secondary teachers had a different perspective and used words such as "battles" and "mad examiner" to describe their relationships with exam boards.
"We've had some run-ins with exam boards in the past, haven't we?" one said. "We've gone in with the jackboots on, really, because we've felt the kids have been unfairly done to.
"We've had results go both ways, but we've had a couple of major ones go in our favour where the work just wasn't examined effectively. I'd got this funny feeling that the kids actually knew more about what they were doing than the person marking it."
While generally satisfied with the reliability of A-level grades, they suggested GCSEs "could be more variable, unpredictable and, to an extent, unreliable" the researchers found.
A secondary teacher told them: "We've got all those wonderful AS scores which are already pointing very strongly in a particular direction.
"GCSE is 'take your pick' almost, and 'pot luck' and various other expressions of randomness."
In a second study commissioned by Ofqual, the National Foundation for Educational Research looked at how exam systems in 14 different countries reported results and how much the reliability of assessment was explained.
It found that: "Few examples were located of reporting uncertainty or error in their results to students.
"An introduction of the reporting of error in high stakes qualifications would need careful handling to ensure this did not result in misinterpretation and a loss of confidence in the system."
Both studies were part of a two-year probe into reliability by Ofqual.
Due to end in December, the investigation is examining the consistency of assessment results and the factors that can affect their reliability.