How does 'normalization' of statistics work?

[paulobrad]

I recently put my face on hotornot.com and I am told I have a score of 9.1 (out of 10). But they also give you the raw results which are quite spread out for me and if I work out the mean average it's about 5.75. I read their FAQ and they say they 'normalize' the result and something about how some people only rank on a scale of 1 to 5. What does this mean and why does my score seem a lot highier than the raw scores say it is?

[3s_a_crowd]

Don't take it seriously. It's only a game. What it means is that because people thought that you were dog rough they bumped your score up so that it wouldn't damage your ego.

[RCP]

A batch of 1.6 litre Ford Cortinas assembled in Israel once arrived in the UK but the '1.6' badge on the back had been put on upside down. Maybe something similar has occurred here ;-)

[gilf]

Good answer RCP. I would question a score of 9.1 It just boast about it :)

[paulobrad]

I'm really not bothered about having my ego dented or having something to show off about, I just want to know what normalization is, the FAQ says they use something similar in the Olympics.

[darth vader]

Normalisation is a type of biasing to the set of results, to achieve a range of standard results which are more acceptable. This sounds like "doctoring" results but let me explain a bit better - Say a class of kids sits a GCSE exam and averages below par (say for example this class was expected to get grade Cs and Bs based on previous results/work done, but for some unknown reason only came out with Ds)- the examiners would look into the factors behind the below par average, perhaps the teacher was substandard, or even replaced half way through the year, or perhaps the exam itself was unusually harder. Anyway the bottom line is that everyone agrees that the class should have been getting better results than this, and that the results were perhaps not accurate of the class's abilities, so they agree to increase each result by a set amount, so that the actual class average is closer to the expected average. Another example (as you mention) is the Olympics in qualifying times - to go to the olympics for a certain event you need to reach a minimum standard, say for example you have to be able to do the 100 metres in sub 11 seconds (i'm just hypothesising here, i don't know the real qual times). Now, athletes who attempt the qualifying by attending events held at high altitude, would have a harder time making the grade than events at low altitude, so the times have to be "Normalised" to account for this, and other factors which may affect the results (weather, time of year etc). Hope this is understandable!

[darth vader]

ps. to answer your question i would say that 3s_a_crowd is possibly correct in that they normalise all results up, so that no one ever gets a poor score (and goes away thinking they look like the back end of a bus). If you give us the link we'll happily give you a 2nd opinion!

[paulobrad]

Thanks darth vader, thats is a great explanation. Then link to my pic is: http://www.hotornot.com/r/?eid=NZARH8K&key=EUN

[darth vader]

definitely a minus three ha ha ha......... cheers, Darth

[LeMarchand]

If you're female, showing a lot of cleavage always seems to get you a high score on that site.