I am somewhat confused by a recent post in Teh Grauniad by our colleague Athene Donald. Part of my confusion stemmed from the fact that I had walked into an argument that was already in full swing – Donald’s piece had been a riposte by this item in which it was asked why the Royal Society was seeking opinions on whether one should make a ‘business case’ for diversity. This question occurred to me – if we all accept that diversity is something to be encouraged, why should a business case even be necessary? Some of the comments were trenchant. If the RS is scientific, what would its response be if their findings showed that diversity undermined rather than promoted business? Is the RS being careful not to bias its own outcome by taking a view, a priori, that diversity is a good thing?
This set me thinking about null hypotheses. When one is seeking to test an hypothesis (such as whether diversity is good for business) one should advance a null hypothesis against which one’s data should be tested. The question of what constitutes ‘diversity’ is somewhat slippery – what is the null hypothesis against which one judges a workplace to be acceptably diverse? Would it be no diversity at all? The current average in workplaces? How would one control for workplaces in which we have good reason to suspect diversity might be more or less welcome than others?
So in a comment to Donald’s column I addressed the much simpler question of gender, which, in all societies we know about, is always approximately 50:50. The evolutionary biology that underpins this optimal sex ratio is fascinating but need not concern us here. All we require to know is that when the ratio deviates very far from it, mechanisms come into play that return it to more or less this value.
My question about equality is about what I see as a hidden presupposition that the representation of people of different gender in any given workplace should be the evolutionarily stable 50:50 ratio seen in the population at large. If no case can be made for such a presupposition (I should like to see one made), then why should the representation of men or women in science (or government, or the civil service, or among nursery-school teachers, or any other activity) be 50:50? Why shouldn’t it be 10:90 or 90:10 or any other ratio one might choose?
In other words, how does one choose the value of the ratio that one should expect a priori, against which one can judge the significance of any deviation one might find? Is the 50:50 ratio a null hypothesis, or the ratio we wish to encourage? Given that there seems to be some confusion about what one’s prior assumption may be, the finding of any particular ratio, whether it deviates from 50:50 or not, is meaningless.
Athene’s response to my comment was very nonscientific and, if I might say so, high-handed – she dodged the question entirely. Yet I do not think this question is invalid. To take an extreme example – one would not expect there to be a 50:50 ratio in applicants for egg donation, sperm donation or surrogate motherhood. Once one accepts that, one must therefore take into account the many less extreme but scientifically attested, biologically based differences that exist between men and women. For example, a recent paper shows that men are better than women at judging the size of a person from their voice. Now, nobody knows whether this makes men or women better or worse candidates for certain kinds of employment. But the point is made – the difference exists, which leaves open the possibility that there exists in any workplace environment an optimal gender balance that need not be 50:50. In which case, one could argue a moral case (leaving aside the business case) that research be done to find a the optimal case for gender balance in any particular workplace environment, because it would be at that balance that would produce the happiest and most productive outcome. I ask again – on what basis is the assumption made that the optimum gender ratio in a given workplace reflects the fact that the natural ratio between genders tends to 50:50? Is this 50:50 ratio the null hypothesis, or the goal to which one should strive?
I do not know why Donald thought I was using gender balance as a cover for something else (read the comments and see – my question was plain enough.) Perhaps she thought I was talking about antisemitism, which Donald knows is my particular bête noire. It was quite clear that I wasn’t (in which case Donald shows that she had failed to read the question, which as every exam candidate knows, is a big no-no) but had I done so, I’d have mentioned that more Jews have won Nobel Prizes than any other segment of the population. Of the approximately 800 recipients, at least 20% have claimed Jewish descent, despite Jews comprising 0.2% of the population.
Were one to claim as a null hypothesis that people in any ethnic group should win Nobel Prizes according to the proportion of that ethnic group in the population, then one would expect the numbers of Jewish nobelists to be trifling. The fact that it is not suggests thatthe null hypothesis has been broken. Something else is at work.
Now, one can hardly say that this curious achievement has been made thanks to any number of well-meaning working parties organised by the Royal Society to assess the moral, or as it may be, the business case for diversity. Au contraire – this achievement has been made despite more than a century of systematic and indeed genocidal persecution of Jews.
What is the lesson here? What is it about Jews that makes them such excellent Nobelists? Some would say that the reason lies with the Jews being awfully well organised, and have a secret conspiracy to run the world, but one can dismiss such notions as the ramblings of racists, conpiracy theorists, and such jealous people as there are who seek to find scapegoats for their own laziness. Could it be, perhaps, that the habit of close family organization, high literacy and organised debate that has been ingrained in Rabbinic Judaism to a peculiarly high degree for two thousand years lends itself to scientific inquiry? Surely not – one must be blind to such quirks of diversity.
Perhaps adversity itself is a spur to success? After all, it clearly worked for the Jews. Perhaps, then, to further the cause of less well-represented people in various spheres of activity – women, people with various forms of disability, people from under-represented ethnic backgrounds – we should seek to deny them access to university positions, kick them out of their workplaces, send them away to death camps, and otherwise seek to marginalise them as much as possible? Ridiculous.
Ah! I have it! Perhaps when one is seeking to discover the next generation of Nobelists, the relevant committee should – purely for the cause of diversity, mind – positively discriminate in favour of women and minorities, and actively vote for any candidate who isn’t Judische? You know, it would be only fair.