08-06-2010, 07:23 PM
Part of my job is interpreting and reporting statistics on engineering experiments.
It doesn't take much training in statistics before you can identify easy ways to misuse statistical data to skew results. It's pretty easy to do on accident if people have "just enough training to be dangerous." And even easier to do on purpose. And once you are experienced, it's pretty easy to do on purpose in a way that a layman won't realize the "mistake."
I guess what I'm saying is that statistics in politics should be approached like an advertisement. Look first at who is paid for the statistical data to be generated, then at who will benefit if the data were skewed. Then make your assessments on the data with the appropriate grains of salt.
You pretty much have to be a statistician with access to the raw data and how it was collected to understand the validity of any given statistic, which means that pretty much any statistics you see publicized can generally be assumed invalid in some way or another. If honest and objective, and trained people interpreted all data and statistics, there would not be "lies, damn lies, and statistics." However, the reality is that most people dealing with statistics are not honest, or objective, and many who think they're trained are really only trained enough to be dangerous, and make frequent fatal errors.
It doesn't take much training in statistics before you can identify easy ways to misuse statistical data to skew results. It's pretty easy to do on accident if people have "just enough training to be dangerous." And even easier to do on purpose. And once you are experienced, it's pretty easy to do on purpose in a way that a layman won't realize the "mistake."
I guess what I'm saying is that statistics in politics should be approached like an advertisement. Look first at who is paid for the statistical data to be generated, then at who will benefit if the data were skewed. Then make your assessments on the data with the appropriate grains of salt.
You pretty much have to be a statistician with access to the raw data and how it was collected to understand the validity of any given statistic, which means that pretty much any statistics you see publicized can generally be assumed invalid in some way or another. If honest and objective, and trained people interpreted all data and statistics, there would not be "lies, damn lies, and statistics." However, the reality is that most people dealing with statistics are not honest, or objective, and many who think they're trained are really only trained enough to be dangerous, and make frequent fatal errors.
Conc / Concillian -- Vintage player of many games. Deadly leader of the All Pally Team (or was it Death leader?)
Terenas WoW player... while we waited for Diablo III.
And it came... and it went... and I played Hearthstone longer than Diablo III.
Terenas WoW player... while we waited for Diablo III.
And it came... and it went... and I played Hearthstone longer than Diablo III.