Science, ego and power

Sarah Vine in the Times (£) says something quite remarkably stupid. Commenting on the IFS’s research showing that people born in August tend to do less well at school than those born in September, she writes:

Some of the cleverest, most confident people I know are August babies…All of which just goes to show you can’t take statistics too seriously.

You don’t need me to tell you what’s wrong with this. The IFS research points to smallish differences in average attainment. This is entirely consistent with some (many) August babies doing well.

The problem lies not in taking statistics too seriously, but in not taking them seriously enough.

Of course, pointing out that a newspaper columnist is writing rubbish is as revelatory as noting the religion of the Pope or ursine defecatory habits. But because columnists are paid to echo the prejudices of their readers, I fear that Ms Vine’s remarks are significant, and depressing, in two ways.

First, they indicate a philistinistic anti-intellectual culture, which elevates ego over rational inquiry: “never mind your evidence and hard work, just look at me and my chums.” This attitude  naturally retards the growth of public knowledge about social affairs. As I’ve said before, what we need is a campaign for social science.

Secondly, there is a class aspect here. Ms Vine ignores the fact that the odds are biased against August babies and so invites readers to believe that their relative (average) lack of educational attainment is their own fault. And this error generalizes. Some inequalities do their damage by working as statistical tendencies - for example, people from poor homes are more likely to suffer ill-health and less likely to do well at school. Relying on anecdotal evidence can serve to disguise these tendencies:  “When I was at Oxford I once met someone from a poor home.” In doing so, the costs of disadvantage are diminished.

In this sense, an ignorance of the very basics of social science can serve a reactionary function. Rejecting empirical enquiry means accepting the prejudices that sustain inequality.

Comments

[Nick Rowe]

There is nothing wrong with what she says in your quote. Let me translate:

"Population statistics should not be taken too seriously when applied to a small sample, because the sample statistics may be very different from the population statistics."

Which is good statistics.

[BBFlint]

No, her statement was much more general than that. It's as if she's dismissing the discipline of statistics as a whole. But one or two dodgy stats - sometimes just badly reported - doesn't negate their value.

Chris: Another point you could make about anecdotes is that people tend to mix with people who are similar to them (interests, hobbies, tastes etc). This exposes them to a very narrow section of society that isn't necessarily representative of the whole. The implication of this is that anecdotes aren't a reliable source of information.

[pablopatito]

"The problem lies not in taking statistics too seriously, but in not taking them seriously enough."

Not sure. The media are also guilty of taking statistics too seriously. So a study showing some health benefits of broccoli becomes a newspaper headline: "Broccoli cures cancer!"

The media abuses statistics both ways, depending on the argument their trying to make.

[Keith]

Sarah Vine could benefit from a course in statistics. I am sure she can afford the fees with her Journos pay.

[Keith]

Having said that I think most people including statisticians see life primarily through the personal events that happen to them. And people they know. I do not see how you can help it as maths is a rarified capacity. Only a small subset of people can see reality mainly by mathematical relationships. Gauss was very clever at maths but this seems to have made him a spoilt brat with a big ego. Your mothers death by murder will have a big impact on you however insignificant it is as a statistical event. I agree it is useful to be able to achieve detatchment from such things when evaluating arguments but that is also a matter of training your emotions. If you prefer to indulge strong prejudices rather than employ reason that is an emotional choice any one can make regardless of their education.

[Nick Rowe]

BBFlint: "Another point you could make about anecdotes is that people tend to mix with people who are similar to them (interests, hobbies, tastes etc). This exposes them to a very narrow section of society that isn't necessarily representative of the whole. The implication of this is that anecdotes aren't a reliable source of information."

But it's the other way around that matters to (most) people. We want to know about the people we mix with. Because the people we mix with aren't a representative sample of the population, information on the population isn't a reliable source for information on the people we mix with.

[Matthew Taylor]

This post reminded me of my own howl of rage against media statistical ignorance..

http://www.matthewtaylorsblog.com/politics/selectively-grumpy/

[Metatone]

Sad thing is, no-one takes this seriously, unless they happen to have a kid who is one of the youngest in the class.

We could probably double the effectiveness of primary age schooling if we split classes in two, one set starting in Sept, and the other around March. It's not costless, but it doesn't involve any great advances of thought, it's just logistics...

[Metatone]

What I mean by "doesn't involve any great advances of thought" is that this is a potential productivity advance that doesn't involve new technology or a stunning discovery in the mechanisms of education.

You'd think we'd at least study the idea some more...

[medyumlar]

Ne düşünce çok büyük bir gelişme anlamına gelmez demek, bu yeni teknoloji ve eğitim medyumları çarpıcı bir keşif içermeyen bir potansiyel verimlilik önceden olmasıdır.

[dr ray]

"Some inequalities do their damage by working as statistical tendencies - for example, people from poor homes are more likely to suffer ill-health and less likely to do well at school"

You don't think you might be putting to much faith in the simple interpretation of statistics. Both the above examples may be true but don't show cause. Perhaps the reason is that people with ill health or poor intellect don't earn much not the other way around.

[Nigel]

Sarah Vine is married tpo Michael Gove - so lets hope that she is not advising him and sticking to beauty tips which is what she used to do at the Times

[BBFlint]

Nick Rowe

"But it's the other way around that matters to (most) people. We want to know about the people we mix with. Because the people we mix with aren't a representative sample of the population, information on the population isn't a reliable source for information on the people we mix with."

That depends on what you're trying to measure. Anecdotes would indeed be one reasonable method for finding out about people you mix with. Of course, there are still potential problems with this. The person whose opinion you are citing may just be an outlier. So, to be accurate, you would need to be systematic about it (e.g. randomised, large enough sample to be representative etc etc). Otherwise you still cannot really draw valid conclusions from an anecdote.

You say that "most people" want to know about the people they mix with; this is a very subjective statement that you did not back up with either reason or evidence. To take an obvious counter-example: policy makers vs "average man on the street". The former may be more interested in macro data (i.e. broad sections of society), the latter in micro (i.e. people you mix with). Now, whilst anecdotes may be ok for talking about people like you (see above), it isn't neccessarily going to be very useful for broader sections of society. It may not be diverse or encompass large enough groups to draw valid conclusions. Broader society may simply be too different from your friendship groups. This is especially true given the context of this discussion - the study at hand was very much at the macro rather than the micro level.