Wednesday, April 04, 2018

Gender gaps in our understanding of statistics


Lot of stuff in the media at the moment about the "Gender pay gap", seems like every slimy politician going is positively leaping on the band-wagon to virtue-signal about it. The question I ask is simply, "where's the data?" Or more precisely, show me the multi-variate analysis that proves that the difference (which there clearly is) between the average pay for Men and Women doing the same job is due only to Gender. Sure, I accept that there will be some employers out there who are exploitative and yes, absolutely, there needs to be mechanisms in law to redress unfairness. But, this "blanket" approach seems all wrong to me and I fear will lead to unexpected consequences that will harm more than it will help. This kind of blanket-policy making where punitive legislation is made not for individuals but for an "identity" (for example Women, Men, Gay, Black, Jewish etc.) has a fraught history. Identity is not a great basis for any kind of legislation in my view, it smacks of quotas, committees, fixed-wages and discrimination based upon things that don't assist the viability of an enterprise, like individual competence.

I've been employing people in various companies in the Tech. sector now for going on 25 years, I must say, hand on heart, I've never perceived a single instance of this so called "Gender pay gap". In my experience people have been paid differently for the same role due to a number of factors, none of which include Gender. The factors I've seen would include things such as general and specific experience, previous salary, incentive to move, hours worked, ambitiousness, intelligence, competence, flexibility, personality and so on, the list is long and the process is a complex and varied one. Added to this, many of the factors are highly subjective. Of course, with the exception of perhaps the Civil Service and certain production-line type jobs, the complexity and changeability of the employment landscape is one of the reasons that it's practically impossible these days to compute which jobs are even "the same".

Now, is the Tech sector an anomaly? Possibly. Is my experience exceptional? Maybe, but there's no data (that I have seen) that even strongly-supports the claim the politicians are currently making. To claim that Women (on aggregate) get paid less than Men for the same role is to assume (among many other things) the following,

- Women (in general) actually want to do the same jobs as men (for example they dominate in Medicine, whereas Men dominate in Engineering)

- Women (in general) don't actually want to put the 100+ hours a week into a career and job in order to run a large complex organisation, sure some can and do, but in general it's men that step up to the plate for that because they have (in aggregate) personality traits and personal situations able to accommodate that.

- Women need to have children (most do) before they hit 35 (ideally, from a Biological point of view) and this unavoidable reality means that the all consuming top-jobs become less and less attractive to Women when children come on the scene.

- Women tend to be more agreeable in personality than men, this correlates negatively with negotiating success (there are many studies that show this) Clearly this can be adjusted with training, but in general you get what you can negotiate and not necessarily what you deserve. Clearly this is unfair, but only in the same way that good looking Women tend to attract more successful men is also unfair. No one complains about bias and selectivity when it comes to finding a suitable partner!

Clearly this whole subject is littered with emotion, strong-opinions and bias, there are obviously many exceptions to these generalizations. Talking about it often elicits the "smoking Granny" response, i.e. if you say "smoking is bad for you", some bright spark always pops up and says "but my Granny smoked and she lived until she was 99" etc. Many people struggle to understand the difference between the statistics of large populations and individual experience. Blanket laws applied to identities that are ill-defined and sit astride highly subjective "Human" processes, like employment, would seem to need pretty clear and unambiguous multi-variate analysis to back them up. 

I'm not convinced that this kind of identity-politics will achieve good outcomes, I can imagine an employment landscape where average and below-average Women (there will always be a tiny minority of high-performers) will become very unattractive propositions to employers. Not because they don't have useful skills but because their pay will be pegged (by law) to the highest male performers simply because of their Gender, regardless of competence or contribution. Of course there may well be situations where this is reversed, i.e. where below-average men receive artificially high wages simply because there are more competent women doing the same job, i.e. the men will actually hold the high-performing Women back! What is more likely however is that everyone will receive a median salary even the high-performers, and this will create serious problems for employers. As most business leaders will testify, not giving someone a pay-rise is an invaluable way of providing an incentive to that person to improve their skills or to move-on and not paying a high-performer what they're worth is a sure-fire way of ensuring they do.

No comments: