It seems research has shown that school-leavers who achieve some form of exam qualification are likely to live 10 years longer than those with no grades at all. A northern source tells me that this little stat prompted one visiting in-service trainer to quiver with emotion the other day. He produced an extraordinary, almost tearful deduction: if the local authority’s schools could increase the GCSE pass-rate by just two percentage points, it would mean 10 bonus years of life for 3,500 pupils in the authority each year.

This was, of course, a madcap reading of the data. A few exam grades are helpful in life, yes, but they are not life-preservers. Using the same flawed logic, we might equally persuade all parents to change their children’s names to Jemima and to tell them to keep a Latin dictionary in the house, since households with both Jemimas and Latin dictionaries nearly always enjoy high-level GCSE success and a longer-than-average life. It is the logic of the apocryphal new head who, on finding that most of the low grades in her school were from pupils who had been assigned a learning support teacher, promptly sacked all the learning support staff.

Yes, a pupil leaving with no grades will tend to have a shorter life expectancy, but the relation is not one of cause and effect. Both are the result of something else: namely, the personality, lifestyle and social environment of the 16-year-old concerned. That’s what the world needs to improve, not the exam pass rate.

And here, in the nuttiest of nutshells, is perhaps the greatest problem with allowing exam data to multiply so uncontrollably. Never mind that all these figures over-empower the data-obsessives in education - no one much listens to them anyway. Far more dangerous is the growing number of talented, intelligent but essentially data-dimwitted people now in positions of power.

In many cases, all their academic training from the age of 16 was devoted to avoiding all things mathematical. But now, as in-service trainers, senior teachers or Ofsted inspectors, they are expected to big it up for all those notorious “data packages”. No wonder the result is so often a well-intentioned nonsense.

Sometimes the basic maths is wrong, too. In fact, closer scrutiny of the in-service man’s “life-expectancy” data revealed that his number-crunching should have led him to 350 local lives supposedly being extended each year, not 3,500. It’s only a decimal point, and what’s another 3,150 people dying 10 years sooner between friends? Take most of this exam data out of our lives and, who knows, we may all stop losing the will to live that extra 10 years.