The experiment we should never do (but did)

The problem with conducting educational research in schools is that you immediately bump up against a knotty tangle of ethics. And, like hitting a dog with your car, once that bump has happened, there's no returning to the point where you haven't hit the dog with your car. The ethical bubble has been breached.

From this breach then emerge all sorts of unwelcome 'what-ifs', and each 'what-if' is a reason not to continue the research. What if your hypothesis is flawed and the students in your test group go on to underachieve? And from thence to economic depravation and low-wage misery? What if the headteacher, whose permission for the research to be conducted must rightfully be sought, is fearful of bad press? What if the students in the control group are disillusioned for life if they find out they were 'only' the control group? What will the parents say?

Educational research is real world research. It cannot be conducted in a laboratory setting because then it would be a film by Stanley Kubrick. The test subjects are students, wide-eyed and trusting, and hopeful for their future. They have but one chance at education, and here you are, T-boning this one, tiny, portentous moment in time that will frame the rest of their years on Earth.

Any research design which has the audacity to interfere with the trajectory of a child had better be so secure in ethical red tape that the angels themselves could present it to the school governors.

For example, if your hypothesis suggests that students would benefit from later starts to their day, you will then want to test this. You will create a test group of participants who will, over a period of time, start their school day later than their peers. You will analyse their attainment and how it compares to their bleary-eyed peers, and you'll probably conduct some qualitative analysis into their sense of well-being at the same time because that will be irresistible, and why not.

But if your hypothesis is proven, have you not then disadvantaged the students who are not in that test group? With UK grades norm-referenced, students across the country are, in effect, competing with one another for a finite number of 9s, 8s, 7s, and, of course, the 4s which will open the doors.

And then there is the Hawthorn effect, which posits that a test subject, merely being aware that they are a test subject, will unconsciously raise their game. In our norm-referenced grading system, the Hawthorn effect will then mean that the test subjects bite off a larger chunk of the norm-referenced grade cherries than their pedestrian, everyday peers. Even choosing test subjects could be advantaging them, before they have even had the chance to deface their first questionnaire with a crudely drawn penis.

Without a time-machine or ready supply of student clones, there is an argument that educational research is inherently flawed. You could run the same research design with 2 groups of test subjects chosen for their similarity, and get entirely different results. It is impossible to adjust for all the variables that make students temperamentally, culturally and economically different from one another. Yet still this bad science is out there, with school-based reforms happening because one handful of students made more progress than another handful of students when, say, they were given an apple to eat.

Happily, teachers make serendipitous research-worthy mistakes all the time, and there are no research designs needed for these, governed, as they are, by the Gods of happenstance and our occasional incompetence.

I once managed to conduct an accidental, small-scale, extremely unethical piece of research myself. I am not proud of it, but it makes for excellent dinner-party chatter.

One student, let's call him Chris, looked very much like another student - let's call him James, and you can see where this is going. As part of my school's well-intentioned drive to secure as much data on each student as we could, teachers were regularly asked to supply information such as effort grades, predicted grades, and current performance grades.

I had just started teaching a year 10 GCSE drama class (14-15 year olds). Chris and James were like anime twins, all dark hair, large eyes and fun to be around, and the Gods of mischief had deposited them both in this one class. Chris was happily able; James was outstanding.

Now, in my English classes, the seating plans would support a quick study of who was who. Martin with the oddly dark eyes sat at the back right of the room. Ashleigh with the blue eyeliner who went on to become a lawyer, was back left. In contrast to the regimented codification of my English classes, my first few drama lessons were a rainfall of exercises with students fizzing around like happy but volatile molecules.

The call came for data, and these were the days when we did not have a sheet of photographs to help us. All the students' photographs were either locked in cabinets or dutifully displayed at home. I am also the sort of person who likes to tick off jobs the nanosecond that they are completed, so waiting for days until the next lesson in order to tweezer these two boys apart in my consciousness before completing the data request was not an option. After a good two minutes of flicking my eyes from Chris's name to James's name in the register in the vain hope that I could decipher their faces in the blue ink, Chris received an "A" for his current performance, and James a "D". Note that these were not final grades but 'just' current performance grades collected and issued right at the start of their GCSE courses.

Of course, thanks to my haste and the dastardly workings of the mischief gods, I had given the wrong grade to both boys. This mistake hoved into my awareness one day when Chris sidled up to me as I leant upon a piano, ignoring my class. His facial features were beginning, slowly, to distinguish themselves from James's, and I realised what I had done. Having just received his report, Chris was lit with a confident energy and brightness that went on to secure him, of all things, an "A" two years later.

From James and his "D", there were no eyebrows raised, no parental complaint (those were the days, although in this circumstance it would have been entirely justified) and he continued in his excellence, eventually even being recognised for it by his muddled teacher and, finally, the exam board, who also awarded him an "A".

I had conducted an experiment, the design of which would never have passed the ethics test, but which was seductive in its suggestion of the power of praise.

Educational literature is bloated with evidence of the negative emotional impact of grades, and an empathic human being, unfamiliar with our history and culture, might wonder at the cruelty of awarding a numerical grade to our children as they try to navigate their way through the landmines of adolescence. Despite his early "D" James had the genes, the temperament and the resources to go on to get his "A". Chris's spirits had been lifted by his erroneous "A", but he had also clearly possessed a hitherto latent skillset that had been ignited by this lift.

Without the resources to replicate every feature of this 'experiment' (time-machine, clones), I will never know if Chris would have gone on to secure his "A" without my mistake.

My 'evidence' that he did, like much academic research, is anecdotal, specific to one set of variables that can never be replicated, and entirely unethical.

"Chris" went on the inherit and then build the family business into being now worth over a million GBP. He has won numerous business awards, as well as having an "A" in drama GCSE.