…this is the title of a song by one of my favourite Australian bands Mental as Anything. This song has nothing to do with statistics (the band members went to Art School), but it’s the launchpad for a discussion of the *average *of something*,* or in statistical terms, the *mean. *To do this I want to make reference to the recently released report by the Austrian Institute of Health and Welfare (AIHW), which annually publishes IVF pregnancy rates for Australia and New Zealand.

Firstly, this data is very valuable. Australasia can be very proud that every IVF cycle and every successful outcome has been recorded (anonymously) in a central database since IVF began 30 years ago. I’ve never, for example, seen national data for the outcomes from naturopathy or detox diets. The data show that we have high pregnancy rates by international standards and we tend to keep problems like multiple pregnancy to a minimum. The good success rates have been maintained over successive years.

However, the results give only the *mean *percentage rate of all treatments.

What is the mean? Put simply, it is the average outcome of all the things you are measuring. For example, if you took every Australian male over 18 and measured their height, the mean height would be an addition of all of their heights to give a total (rather tall) number, and you then divide that number by the total number of men you measured. In IVF terms, if you took 10,000 women that had IVF in one year, and 3,000 got pregnant, the mean pregnancy rate amongst those women would be 3,000/10,000 or 30%.

Now, if that is the average number, it means that about half of the individuals did better than that and half did worse. About half of the men in Australia are above the mean height and half are below it. So even if the average Australian man’s height is 170cm, it’s no consolation to you if you are only 145cm and wished you were taller.

The national IVF data therefore mask the fact that some women will do better at IVF and some much worse than the average national figure. The data also mask the fact that some IVF units do better and some worse (it is statistically possible that all units do exactly the same, but I very much doubt it).

The mean of something becomes much more relevant when you divide the data down into lots of different relevant subgroups. In IVF the obvious one is female age – for example what is the average result for 40 year olds, compared to 30 year olds. However you can’t break things down too far, because then each subgroup has such small numbers that the results in each set are no longer statistically significant (that means they could just as likely have happened by chance and not really be representative – if I only measure the height of three tall men, the average won’t be representative of all men).

So, when you ask your health practitioner what the result is from a particular treatment, they don’t necessarily mean to be mean, if they just give you a mean. But they would be better giving you as specific a mean for your individual situation as they can – ie your age, your background medical history and the results from *their treatment in their unit*. Don’t forget, however, that your return could well be higher or lower than the number they are giving you.