Tag Archives: Correlation and causation

Even more on the ‘value’ of university sport… oh the irony

22 Jul

Only yesterday I wrote about a story from the USA about how university sport ‘allegedly’ makes you cleverer, and then this morning the Sports Centre of my own institution tweeted a link to a UK story suggesting that university sport makes you more employable and results in you earning more money !!!!. This story comes from what seems an unimpeachable source, an academic study funded by BUCS.

As with the study I wrote about yesterday from the US, in the 59 pages of the BUCS report I can find no mention of the Socio-economic of the students surveyed. I am well aware that there are a some honourable exceptions to this (my own institution is based in an economically reprieved area but has a successful rowing club !), but it’s not beyond the bounds of possibility that universities that draw their students from higher SES groups make larger investments in sports, and students at those institutions have a higher disposable income and more free time (not working part-time) and thus more time to devote to sport. It’s hardly surprising that students from such institutions are more employable and earn higher salaries. I’m just not convinced that the sport is what is driving the higher salaries !!!

Now of course I’m not suggesting that university sport is a bad thing, but people really need to look at the source of stories. One wonders if BUCS would have funded a study that said ‘Students at Russell Group Universities (the UK equivalent of the Ivy League) have a higher disposable income and more free time that students at Post-92 universities (State unis in the US) and go on to earn more’. That’s rather more a point about the socio-economic climate that it is a point about university sport..

As ever the take home message of this is that just because you find a correlation it doesn’t mean that variable A causes condition B !!!!!!

Being a member of a gym makes you do better at University…. The return of the Mediteranean Diet Effect

21 Jul

One of the most used examples for teaching rational thinking is the ‘Mediterranean Diet’ story. Put simply this is the idea that Northern Europeans live longer if they stick to the diet of Southern Europeans i.e. Lots of salad and olive oil. A few moment’s thought demonstrates how silly this idea is i.e. Northern Europeans who consume lots of salad and olive oil tend to come from higher income brackets, and it’s the higher economic status that is the predictor of longevity, not the diet itself.

This can to mind when I read a story in the Daily Mail this week:

Capture

This story reports a study from the Michigan State University suggesting that students who join an on-campus gym get better grades and stay in university longer. Sadly the original paper is behind a paywall, but the available abstract makes no reference to controlling for the disposable income of the students. (Working in a university I am fortunate that I have access to the full-text of this paper. As I suspected, the study doesn’t control for disposable income, indeed the paper actually acknowledges that Socio-economic status could confound the results) It seems possible that students who can afford to join a gym might have a high disposable income, and indeed that those with a higher disposable income might spend less time in paid employment and thus have more time for sporting activities (and college work !!).

All of this leaves me wondering who is at fault here. It’s interesting that whilst the paper itself acknowledges a potentially dramatic confounding variable it still makes claims about the importance of the relationship found. At times I worry that in a drive to gain publicity for their research scientists may end up actually making matters worse by facilitating a lack of understand that correlation doesn’t necessarily equal causation.

Correlation and Causation – Comedy Gold !!!

16 May

xqOt9mP

For anyone who’s ever taken a university research methods or statistics course the phrase ‘correlation doesn’t imply causation’ probably sticks in the mind. This vitally important idea is central to interpreting statistics, and yet I suspect that even for those who have studied such things to university the idea isn’t entirely clear. Politicians seem to love to ‘abuse’ correlational data, and thus an understanding of this concept is important for anyone wanting to think rationally.

Most teachers presented with the problem of getting this idea across have on obviously ridiculous correlation that they roll out. My own example is ‘There is a negative correlation between number of children in a household and number of electronic devices in a household, but this doesn’t mean that handing out toasters will reduce the birth rate’.

Someone has now done a great service to rational thinking teachers everywhere and collected a great series of spurious correlations all in one place

 

Regular sex makes you rich. (I give up !)

16 Aug

richsex

My list of weird penis and sex stories has just got even longer. The Daily Mail’s story today reports a study conducted by Dr.Nich Drydakis of Anglia Ruskin University that reports an apparent correlation between income and frequency of sexual intercourse. Unusually for such stories, and very usefully for teaching purposes, the original paper is freely available on the internet.

This story could be useful for illustrating a whole range of things :

1) Correlation and causation – Does this mean that if you have more sex your income will increase ?

2) Can self-report be relied on ? – I’ve written before about the perils of self-report studies in relation to penis-size, and this seems to me like another example where it’s entirely possible that people who exaggerate about sexual activity might also exaggerate about their income

3) Generalisation – The Mail’s report of this study reports that it used 7500 participants. At face value this seems this it might add to the generalisability of the studies findings. However, I’m interested to see what my students make of all the participants being Greek. Can we draw conclusions about the UK from such a sample ?. It seems to me that even if the data is reliable, the vast cultural differences might bias the conclusions

All in all this looks like a really useful example that I look forward to trying out with the students

Illustrating correlation and causation

8 Aug

mail

I’ve written before about an alleged relationship between computer use and various psychological problems and a newspaper article I read this week made me think about that idea again. The newspaper story straightforwardly reports a paper demonstrating that children on the autistic spectrum spend twice as long playing video games as typically developing children. This seems like a really nice illustration for students that just because two things are related it doesn’t necessarily mean that one causes the other. I suspect that most students will rapidly see the idea that essentially solitary video gaming might be an obviously appealing recreation for autistic children.

What is depressing is that I suspect that those peddling a link between computer use and psychological problems (Aric Sigman, Baroness Susan Greenfield) will latch on to studies like this and claim that they support their theory. Much has been written elsewhere about Greenfield unsubstantiated claims about autism and computer use, (Ben Goldacre is particularly good on Greenfield’s strange claims) but I do wonder whether I should be pleased or depressed that my students can spot something that seems to be difficult for ’eminent’ scientists !

Things like this are always really useful for teaching. I find that the idea that they are developing a skill that others (who they think are cleverer) lack is really appealing to students.

 

 

Eating chocolate and drinking milk will make you a Nobel Prize winner

22 Jan

Screen Shot 2013-01-16 at 22.11.46

You could hardly miss this story, as it seems to have been all over the media this week. The headline of the  Daily Mail version of the story focused on a correlation between a country’s milk consumption and the number of Nobel Laureates from the country. The story went on to cover a similar correlation between chocolate consumption and Nobel prizes. The same story appeared around the world including Bangladesh, USA (Time), Pakistan and Ghana.

You might just dismiss this story and the usual newspaper rubbish, but when you read through the story it is based on material from the Practical Neurology (A British Medical Journal publication) and from the New England Journal of Medicine. When you look at the Practical Neurology article one can feel the authors tongues firms in their cheeks and in the second line of the article they acknowledge that correlation obviously doesn’t imply correlation.

This story is useful for two different types of teaching. At a basic level it’s a nice way to illustrate the idea that correlation doesn’t imply causation. In discussion with students I’m pretty sure you could come up with a lot of possible variables that might me mediating this relationship. My own mind wanders to the figures for lactose intolerance in Asian countries being 75% plus.

At a much higher level I think there is an interesting discussion to be had about the responsibility of scientists when publishing correlational data. Whilst both the papers that this story was based on were published in peer-reviewed journals (and were thus aimed at an audience who would understand correlation and causation) it’s not unreasonable to suggest that given the subject matter both the authors and the journal editors would be aware that the popular press were likely to pick it up. Under these circumstances my question is ‘Do the authors and editors have any responsibility to consider the wider audience and their lack of understanding and correlation”? It’s all very well for us to bemoan poor science reporting in the popular press, but we ar least partially responsible ?

Teaching statistical thinking might have just got easier

6 Aug

Over seventy years ago Samuel Wilkes, the then President of the American Statistical Association, wrote “Statistical thinking will one day be as necessary for efficient citizenship as the ability to read and write!”. (Interestingly he was paraphrasing the work of the the British writer HG Wells from 1903 !!). Given the welter of statistics that we are confronted with on a daily basis, it seem s perfectly reasonable to say that that day has arrived, and yet what I see of students entering undergraduate studies suggests that we have a long way to go in developing ‘efficient citizenship’.

My own discipline, psychology, requires students to have detailed knowledge of the statistics of null-hypothesis testing and yet in focusing on an understanding of t-tests, ANOVAs regression etc etc I suspect that more apparently ‘basic’ statistical ideas such as understanding distributions and sampling are often neglected. It has be to said that the fault does not entirely lay with those of us teaching in higher education institutions. Psychology students appear to arrive at university with an aversion to anything the looks like mathematics, that you have to assume is a product of the nature of pre-16 mathematics teaching.

Ive written before about using opinion polling problems as a route into teaching about sampling, but I’ve just come across a little book that seems like a perfect way of getting students to grasp ‘statistical thinking’.

‘How to lie with statistics’ by Darrell Huff is a tiny (124 pages) fifty year old book that contains a huge range of the type of examples that I like to use when teaching, and delivers them in a style that is accessible to modern-day students. It seems to me that this book would make an excellent basis for the first few weeks of any introductory statistics undergraduate course and ought to be compulsory reading to guarantee ‘efficient citizenship’.

%d bloggers like this: