It’s sort of funny, after having posted the previous item this morning, to come across some analysis of the ‘National IQ test’ last night. If you have a look through there you can see some good examples of using graphs to exaggerate differences, by simply changing the y axis scale.

And that’s aside from the debates about the futility of assigning a single number to ‘intelligence’. The psychometric literature has a lot to say about this. And that’s aside from all of the issues about socioeconmic effects biasing the outcomes.

I guess what really pisses me off is the general trend to give people a quick summary of information, without indicating to them that it is a summary. So people get these sound bites of, for example, ‘HRT increases the risk of breast cancer’, without an explanation of how much, for whom, who these numbers were obtained by, how they were obtained, or even that there may be more to the story than that.

It’s this sort of crap that led to the Asteroid Flap of a week or so ago. If you looked at the actual findings, it was pretty clear that there was a pretty wide area the asteroid was going to pass through, and because of the limited accuracy of observations, the earth happened to intersect that area, creating the possibility of an impact in 2019. Of course, it was reported as ‘End of the world to come via asteroid in 2019′. And that’s it. No wonder people panicked.

Again, it would be nice for people to simply ask ‘Where is this information from? How reliable is it?’ before getting too excited about stuff they see in the news.

Ross Gittins, as always has a very interesting article in todays SMH, about the ways people manipulate statistics. Since doing stats at uni, and being a naturally suspicious kinda guy, I’ve always taken reports that doing $random_thing will increase/decrease your risk of $some_random_undesirable_thing with a grain of salt. The obvious questions are things liek: what’s my chance if I do, and what’s my chance if I don’t?

For example, if I like to hop around on one leg screaming about the social construction fo reality at the top of my lungs, and it’s been shown in some study that this increases my chance of being attacked by a thousand rabid marmots by 10%, should I be worried?

Not if my chance is, say, 10 in 10 billion to start with. If that’s the case, the impressive-sounding increase of 10% in my risk raises my chances from 10 in 10 billion to 11 in 10 billion. Oh no, I’m really scared. The marmots are coming! The marmots are coming!

I wish people would switch on their bullshit detectors more often when using the media.