Big Greedy Companies

[Chris Auld]

Originally Posted by Coasterdon

When I say stats can prove any point I mean for example:

Poverty has risen in Canada per Stats Can. Doesn't this depend on what you consider poverty to be?

Of course it does. Why is this an example of "stats can prove any point" instead of an example of a case where one must reason clearly to understand the implications of a particular statistical argument?

Crime in Toronto has gone down. Sure crime per person has gone down. Wouldn't that be the case if crime went up but the population of the city went up too?

Another good example of the advantage I previously noted of statistical arguments providing clarity of expression. If someone told you "crime has gone down" and you tried to translate that into a statistical argument, you'd immediately confront the problem that you need to know how "crime" is being measured. Is it absolute number of crimes? Is is per-capita crime? Which "crimes" are being counted, by whom, and in what circumstances?

Oracle is buying Siebel and going to lay off 5000 people. Except for shareholders how does this contribute to society in any way?

This isn't a statistical statement in any sense. (The answer, incidendentally, is that those 5,000 people will go off and do something else, and if the market is working more or less as it should they will be more productive in those other tasks.)

Notice that you have not provided any examples of "stats proving any point." As I said, any argument, statistical or otherwise, can be deceptive. The former are easier, not harder, to reveal as wrong because statistics and mathematics are so closely related.