Professor Stephen Newstead, president of the British Psychological Society, is calling for a national examination at university level, and a radical overhaul of the student assessment system to offset the "worrying variations" in marking.
He cited the example of two experienced markers differing by 70 per cent in the marks they gave a history script. In another study, Professor Newstead gave six essays to 14 experienced examiners in psychology and found "dramatic" variations.
Professor Newstead, Professor of Psychology at Plymouth University, said that in the US no one now trusts the degrees awarded by individual universities. They rely instead on a national exam, the Graduate Record Examination, taken by graduates. "We are in danger of that happening here unless we take steps to address it now . . . The system is in danger of being undermined," he said.
Speaking on the opening day of the BPS annual conference in Brighton, Professor Newstead said that examination marking is unreliable; standards of assessment are inconsistent over time, institution, and subject; markers are biased, and it is too easy for students to cheat.
The assumption that a First Class honours degree means the same as it did 20 or 30 years ago is open to question, he added. The number of firsts awarded nationally has increased from 6 to 10 per cent since 1980, while the average degree is now upper second, not a lower second as it was just a few years ago. This is at a time when almost one-third of 18-year-olds enter higher education compared with just 5 per cent in the 1970s.
"It is difficult to believe that the proportion of these modern-day students who merit a good degree is so much higher than the proportion of their much more highly selected predecessors," Professor Newstead said.
He pointed to wide variations in the awarding of good degrees - an upper second or higher - between disciplines.A student had a 50 per cent chance of getting a first or upper second in philosophy and history, but only 30 per cent in accountancy. Just over 40 per cent of maths and education students received good degrees compared to 60 per cent in engineering and technology.
Marker bias is also a key factor, based on an examiner's knowledge and opinion of the student with a tendency to make sure that their own students do better than those of other lecturers. They may be marked higher, or helped before the exam with hints about questions or revision classes on exam-related topics.
There was also evidence of discrimination against women students who tended to be marked down compared to male colleagues who get twice as many firsts.
Cheating in exams was also "disturbingly high", Professor Newstead claimed.
A study of almost 1,000 students found that more than half used material from another source without acknowledgement, nearly half had invented data, and a similar number allowed fellow students to copy their work.