> . Napster users are 45 percent more likely to
> purchase music not
> because they are music aficionados, but because
> Napster itself is a
> catalyst for sales. Even when Jupiter researchers
> controlled for
> other factors that influence music purchasing, such
> as time spent
> online, income, age and gender, the use of Napster
> still had a very
> strong influence. "Other than existing level of
> music spending,
> Napster [use] was the greatest predictor of music
> spending," remarks
> Jupiter analyst Mike Saxon.
>
Whichever side of the argument you're on, this is pretty mouldy statistical practice. The study described simply doesn't justify this claim. Controlling for income, age, gender, time online etc is a pretty pointless pice of OLS frippery; it clearly doesn't control for "being predisposed to both buy music anyway and use Napster, ie, being a muso-head", which would be the null hypothesis of interest.
I can't remember who said it: "The Ordinary Least Squares regression model is the internal combustion engine of statistics; its extreme convenience has allowed us to forget how toxic it is".
dd, arch pedant.
===== "The banker must at all times affect a respectability that is more than human ... for this reason, he is typically the most romantic and least realistic of men"
John Maynard Keynes
____________________________________________________________ Do You Yahoo!? Get your free @yahoo.co.uk address at http://mail.yahoo.co.uk or your free @yahoo.ie address at http://mail.yahoo.ie