The Great Five GCSE Myth

One of the more interesting and widely cited, empirically verified, facts to emerge from the field of psychology in the last few decades is the fact that people have a marked propensity to overestimate their own abilities relative to other people.

To put another way, if you ask 1000 randomly selected individuals to assess their own intelligence relative to the general population, on a scale of 1 to 100, more people will afford themselves an above average score, i.e. above 50, than is statistically possible.

This phenomenon has acquired several different names over the years in its various guises – illusory superiority, overconfidence, the Lake Wobegone effect, Dunning-Kruger effect – but it all add up to the same thing. Across a wide range of abilities, skills and personal qualities, including intelligence, social skills, sense of humour, professional ability, etc. humans have consistently proved to be rather poor judges of their own capabilities and, to compound the problem, the less able or competent an individual is, the more likely they are to both overestimate their own competence and underestimate the competence of other people, even if those other people are considerable more skilled tan they are.

Its a classic double-whammy. Incompetent people not only fail to recognise their own incompetence but they also fail to see recognise competence in others, depriving them of the opportunity to address their own failings and improve their abilities by learning for the people around them.

This phenomenon has been found in everything from schoolchildren and college students, to college lecturers – a faculty survey at the University of Nebraska found that 68% of faculty members rated themselves in to the top 25% for teaching ability – and even amongst tournament chess players, a field in which – in theory – no one should be under any illusions as to their ability relative to other players due to mathematically rigorous ranking system used in tournament chess.

If you understand this phenomenon, and other related issues and limitation in the general field of intelligence research, including the Flynn effect and ubiquitous nature of the normal distribution, i.e. the famous ‘bell curve’, then this news report from the BBC will make sense to you in ways that complete and routine escape both politicians and journalist.

Just one in 15 (6.5%) pupils starting secondary school in England “behind” for their age goes on to get five good GCSEs including English and maths, official data shows.

The government data published as part of secondary school league tables suggests the majority of schools are failing struggling pupils.

Nationally 58.2% of pupils reach the five good GCSEs benchmark.

Minister Nick Gibb said schools which let pupils down would be tackled…

Only a third (34%) of these children achieve the government’s benchmark of five GCSEs – or equivalent qualifications – graded A* to C, including English and maths.

In 909 schools, not one low-attaining pupil (those who did not reach Level 4 at the end of primary school) reached this threshold.

At the other end of the spectrum, 95% of pupils who started school “ahead” for their age (achieving Level 5 at the end of primary school) got five good GCSEs, including English and maths.

And of those who started school at the expected level for their age, (Level 4 at the end of primary school) some 45.6% failed to progress to five good GCSEs.

Overall, 58.2% of pupils in England’s state schools got five good GCSEs including English and maths (including equivalent qualifications).

When these qualifications, such as BTecs and NVQs, are excluded, 52.4% of pupils gained five good GCSEs.

In truth, this raft of information tells us very little about actual school performance, but its does tell us quite a lot about the education system generally and the manner in which its been ‘weighted’ over the years to generate particular outcomes, outcomes which politicians and policy-makers believed to be politically desirable.

Examinations, and the examination system generally, are not an empirically fixed, strictly objective measure of student performance. rather they are a weighted system in which examinations, qualifications and grade boundaries are set in such a way as to generate certain statistical outcomes.

For example, if its considered that an average student should be capable of attaining a grade C in a mathematics examination, then the grade boundary between a C and D – the pass mark – will be determined, based on an evaluation of the degree of difficulty of this year’s questions and a statistical analysis of student performance in previous year’s examinations, in such way that it will be expected around 50% of all examination candidates will get at least a grade C.

So, when we’re told that 52.4% of students, nationally, gained 5 ‘good’ GCSEs (58% if you include other ‘equivalent’ qualifications) all that actually tells us is that the system, as a whole, is weighted is such a way that it is assumed/expected that an average student should be capable of gaining 5 good GCSEs, or at least any five good GCSEs as the headline figure falls to just 34% if we insist on both English and Mathematics being part of the package.

Just exactly how ‘good’ those GCSEs are and what, if anything, this tells us about the abilities of those students who did make the grade is anyone’s guess as these statistics are almost entirely detached from any kind of solid empirical framework against which we can reasonably evaluate these results over anything other than a very short period of time.

For example, back in the days when I was at school and ‘O’ level were considered to by a elite examination system targeted primarily at the top 10%-15% of student, many of whom were, in those days, expected to continue their studies through to university, the overall system was weighted in such a way that an average student would be expected to leave school with a reasonable number of CSE qualifications at grades 3&4, with a CSE grade 1 being considered to be equivalent to a grade C pass at ‘O’ level.

This being the case, does this mean that young people leaving school today with 5 good GCSE are actually no more intelligent or better educated than those who left school 30-40 year ago with the clutch of average CSE’s.

Well, yes and no.

From the Flynn effect we know that, overall, average human intelligence, as measured by IQ and general intelligence tests, has been increasing substantially and in sustained manner over the course of the last century although recent evidence suggests that this trend is starting to level off in some developed countries.

On paper, we’ve become much more intelligent than our parents and grandparents in a relatively short space of time, but what’s not entirely clear is just exactly how much of that is a genuine improvement as opposed to it being an artifact of our increasing complex social environment. Unfortunately, even if a sizeable proportion of the Flynn effect is a genuine increase in human intelligence, this has to be offset against the fact that the world we inhabit is a considerable more complex and, in many ways, intellectually demanding place than the world our parents grew up in and so, overly, its likely that when you offset the gains in intelligence against the increase in social and environment complexity, the net effect of our apparent gain in intelligence is likely to be small and confer only marginal benefits relative to the situation in which our ancestors found themselves.

One thing that has changed over time and that is relevant is our expectations of what will happen to young people when they leave school.

Forty to fifty years ago, Britain’s economy and labour market was structure in such a way that it was assumed, if not expected, than an average student leaving school with average qualifications would go on take up an apprenticeship and learn a trade or get a job as semi-skilled manual worker. Average meant more or less ‘blue collar’ and the examination and education system was weighted accordingly.

Today, the economy and labour market has changed considerably such that our working definition of ‘average’ is pegged to the demands of the service industries and, for political reasons, to a much larger proportion of young people should be going into higher education and on to university, hence the need for the average student to gain 5 ‘good’ GCSEs in order to be considered to have been educated successfully.

This shift in emphasis in perhaps best illustrated if we consider where, exactly, this 5 GCSE benchmark originated.

Its commonly assumed that its origins lie in the standard entry criteria for universities, i.e. it became the benchmark because it was one of the entry requirements for a university education at the time when politicians and policy-makers decided that an increase in the number of young people going to university was a desirable policy objective. This is not, however, the case. If you look back to the days before the introduction of the GCSE examination, what you will find is that universities did not insist on or look for 5 ‘O’ levels at grade C or bertter as a basic entry requirement. 2 or 3 ‘A’ level passes, depending on the insititution and course, yes, but not 5 ‘O’ levels. The only ‘O’ levels that universities required were a couple of grade C passes, or better, in English and Mathematics and if student obtained ‘A’ level passes, in these subjects, in addition to one further ‘A’ level then they could readily obtain a place at a university, or more likely a polytechnic, on the back of having actually passed only three ‘O’ level examinations.

At the time, the only institution that specifically required entry candidates to have the magic 5 ‘O’ levels was the Civil Service, and its from their entry standards, and not those of the universities and polytechnics, they we get the 5 GCSE benchmark.

This, of course, marked a shift in expectations for the average student – as the labour changed over the course of 1970s and, particularly, 1980, average ceased to be synonymous with  the declining blue collar sector and became linked, instead, to the white collar sector in the burgeoning service industry sector.

So, by the standards of the 1960s and 70s, our expectations for the average student would be considered to be higher than the prevailing expectations of the period, a white collar job being generally considered to sit higher on the career ladder, and in terms of social class, than a blue collar job. However, if you look at how the service industries have actually developed over the last 30 years and the impact that this has on the white collar working environment relative to the environment that existed in the 1960s and 70s then the general trend in the workplace has been one of increasing systematisation of white collar roles, with an attendant loss of autonomy and an overall downgrade in the intellectual demands of white collar work. Where, at one time, a white collar job was a skilled role, today most such jobs tend to be at best only semi-skilled roles which require on a very narrow and sometimes basic skill set. The sector hasn’t quite got to the point of operating on a ‘monkey push button’ basis that one used in large-scale manufacturing but in many environments, particularly call centres, the work that is on offer is not far away from that level and the working environment in call centres is arguably a lot worse that it was in factories. People who did dull, repetitive, routine assembly jobs in factories could at least get through the working day by chatting to their workmates, an option largely unavailable to call centre workers in the kind of galley slave conditions that have sadly become the norm in that part of the service sector.

So although it might appear, on paper, that gains have been made and that this is reflected in the education and examination system, when you look more closely at what some of these statistics mean then it becomes much clear that things have actually moved on or improved appreciably over time, particularly if you drill down through the system and look close at how the curriculum has changed over time, especially in the natural sciences where, at GCSE level, things have quite clearly changed for the worse in terms of both content and intellectual rigor.

This being the case, what should we make of the following statements..?

Just one in 15 (6.5%) pupils starting secondary school in England “behind” for their age goes on to get five good GCSEs including English and maths, official data shows.

At the other end of the spectrum, 95% of pupils who started school “ahead” for their age (achieving Level 5 at the end of primary school) got five good GCSEs, including English and maths.

In the eyes of politicians this is evidence that at least some schools are failing their students.

To someone who is conversant with the findings of several decade of human intelligence research and the general workings of education and examination systems, these results are only to be expected and are very much in line with what we know about intelligence and its statistical ‘relationship’ with the normal distribution.

What we are looking at here is data relating to individuals whose overall performance sits at opposite ends of the distribution, the most and least intelligent students – at least on paper. Intelligence is not a thing in itself, it a conglomeration of cognitive skills and abilities and this where much of the debate around the fairness or otherwise of examinations and assessment systems lies. For example, if you set examinations which primarily test student on their ability recall information but not their reasoning skills then its likely that a rather dull-witted student with an excellent memory will score more highly than student with excellent reasoning skill but a relatively poor memory and because we place so much store in exam result when it comes allocating places in higher education this can have the unfortunate effect of allowing less capable student to advance through the system with relative ease, until they reach a level at which their limitations are exposed while holding back student whose abilities would be better suited to the kind of demands made higher in the system.

In our hypothetical example, one might expect the dull-witted student with a good memory to get a university place with relative where the pre-entry examination system  focusses primarily on knowledge accumulation and recall but struggle to complete their degree because degrees course are generally expected place greater demands on students’ reasoning abilities, while our other student may struggle or even fail to obtain a university place, despite their abilities being better suited to that environment.

Even after taking these uncertainties into account, we should nevertheless expect that figures for groups at the far ends of a distribution of test scores or examination result will largely reflect the genuine abilities of most of the students in these groups and this is, in turn, an accurate reflection of those students’ cognitive and intellectual abilities – most, but not all.

Intelligence and other cognitive abilities are not the only determinant of examination performance. The quality and style of teaching matters, as does the mode of assessment used in evaluating performance and, of course, motivation and a willingness to work at your studies also plays a part – and the latter is also susceptible to a range of influences of which the quality of teaching provided by schools and the amount of support provided in school and in the family home are also relevant.

Taking all this into account, we should firmly expect that although for many students their examination and test results provide a fair and accurate reflection of their abilities, some students will also either overperform or underperform at particular stages in their education and also that, of the students who do under/overperform, at least some of these students will go on to to find and perform at the ‘natural’ level of their actual abilities as the progress through the system.

So, when we compare student performance at 11 and 16 and look specifically at te top and bottom ends of the performance distribution and how these compare over time, the one thing we should absolutely expect to see is that a proportion of the students will either move up or down the distribution and out the end groups purely on the basis of changes in student performance that amount to nothing more than regression to the mean.

So, although the data for the lower end of the distribution, which the end that politician are most concerned with and most likely to comment one, shows that some students successfully moved up the distribution between the ages of 11 and 16, it doesn’t tell us anything at all about why this happened or, indeed, why it didn’t happen for the majority of students who started out at the bottom of the pile and stayed there. It could be that they were let down by their school, or it could be that their position on the distribution is nothing more than a fair and accurate reflection of their actual abilities.

I don’t know. You don’t know and you can be absolutely certain that Nick Gibb hasn’t got a Scooby either.

At best, he’s just blaming it on ‘poor schools’ because that what ministers always do and it keeps the media happy, even if its nothing more than the usual handwaving in the face of questions to which politicians don’t have any real answers.

At worst, he may genuine believe that this all just down to poor schools and poor teaching, in which case we have an education system which is being shaped, to a considerable extent, by the Dunning-Kruger effect such that the only thing we can actually expect is that although, in twelve months time, the actual figures may have changes to small extent, the excuses for the fact that some young people don’t leave school with five ‘good’ GCSEs will stay exactly same, just as they have for at least the last 25 years.