The Wallet Test

[sledghammer]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
Have you ever taken a statistics course?

[/ QUOTE ]

No. But you don't need to be a statistician to conduct a test such as this. You also don't need to be an expert movie maker to film people stealing wallets. You don't need to be a criminologist to figure out that wallets that were never returned were probably stolen. You don't need to be an anthropologist to tell if a person is a man or woman, young or old, black or white. You don't need to be a statistician to figure out that 43% is over twice as likely as 21%... This in not rocket science or brain surgery - this is careful observation, good record keeping and high-school-level math. [img]/images/graemlins/smile.gif[/img]

[/ QUOTE ]
You have just demonstrated your woeful ignorance, my friend. This is not high school level math. Yes, 43% is higher than 21% but you cannot necessarily rely on that figure.

[/ QUOTE ]

There is so much more wrong with his 'study' than just his 'analysis' of the data. The design, the collection, the assumptions, everything about it was either wrong, or not specified.

Just a few things:
He didn't collect all the data at the same location, and didn't record what locations different races/ages were at, or how location affected pick-up rate.

He assumes that whoever took the wallet stole it. At some locations, this would make sense (in front of a police station.) Anywhere else, there might not be an obvious place to return it. Certainly different locations would have a profound effect on this.

If he had talked to any statistician (or anyone who had even taken a stats course) his study wouldn't have those problems, but its obvious that objectivity was not the purpose here. And of course the sensation of this garbage was enough to get all over the news.