Spurious correlations

Or as I seem to remember from my dim and distant past, there is the dreaded intervening variable

Of course correlation is not the same as causation. There may be a statistical correlation between aggression and playing violent computer games, but that does not mean playing violent games causes aggressive behaviour. It could be that aggressive people are more likely to choose to play violent games.

That is it, varian. Without properly recorded and interpreted statistics, the link between smoking and lung cancer, for instance, might not have been made. However, at first glance statistics may seem to say something definitive, but a closer look reveals that they don't mean what you thought at all.

The press often reports marvellous medical breakthroughs which are revealed in the actual statistics of the sober trial reports to be a slight improvement on current practice, no more, no less. The researchers seldom claim they have a miracle cure, but the reporters often don't understand the statistics.

I disagree. Statistics, correctly collected and analysed are the basis of scientific discovery.

The corruption of statistics by poorly collected data incorrectly analysed leads to the sort of spurious conclusions that become sensationalised in the gutter press.

Sadly the low standard of numeracy in the population at large is responsible for poor comprehension and resorting to "lies , damned lies and statistics" cliches.

I think some people I discuss with use this type of statistic. Later in my nursing career we had to use evidence based practice meaning some past practices were proved ineffective, the placebo effect. Also just touching and caring for someone is good medicine and giving someone a bottle of coloured medicine in the right shaped bottle worked. So It made me aware to use use evidence based information. Yet many people express their opinions on their own experience of what happened to them or what their family or community do and won't consider the evidence when it contradicts their experience.

But I do agree that you can make statistics prove anything depending on your sample and how you count and as this op proved what you compare it with. So statistics are not evidence.

A fun post

is that counting two people in every marriage ...lol

You can, I believe, massage statistics to say whatever you want them to say

Lies, damned lies and statistics - and other clichés....

Who knows the truth? Maybe hundreds of husbands were enraged over getting marge on their toast instead of butter, and retaliated by filing for divorce?

The classic is the margarine/divorce correlation.

tylervigen.com/view_correlation?id=1703

Missed out a word - more prevalent than marriage.