The NY Times yesterday printed a letter from three lawyers claiming that the Amer. Bar was prejudiced against conservative judicial nominees. They did this by studying whether more appointees of Republican presidents received the highest possible rating from the ABA compared to Democratic appointees. I am not questioning (although I have not verified) their facts. But suppose I wanted to establish that, on average, Republicans made poorer appointments than Democrats. I could use the exact same statistics and claim that they proved my case.
In fact, I do believe that the second case is true, but those data don’t establish that proposition or, likely, any other. I base my belief on dumber decisions, not statistics like these.
There’s a radio commercial here for a real estate investment company that states “Ninety percent of what most people think about money is wrong.” It kills me every time I hear it.
The statistical finagling which always pisses me off is newspapers (and they all do it) which print graphs whose vertical axis doesn’t start from zero. “Unemployment plummets” goes the headline - accompanied by a nifty graphic with the axis going from 7% to 7.5% and a squiggly line showing a tiny drop from 7.2% to 7.1%. You just know there’s got to be thousands of people looking at it with half an eye saying to themselves “Wow! Unemployment has halved!”
There’s a great book on innumeracy called 200% of Nothing. One of the many examples:
The electric company sent out a flier stating that 90% of burglaries took place on inadequately lighted streets. It further stated that 95% of streets were inadequately lighted.
Their conclusion: pay your electric bill.
Mathematical conclusion: 10% of crime takes place on 5% of streets. Lights attract criminals.
I just saw David Simon (creator of The Wire) do an interview with Bill Moyers. He brought up the point that all the statistics that measure progress in our institutions are being misrepresented by the officials in charge of those institutions.
What makes it worst is that the quality of reporting is going down as well. No one caught our current economic crisis because no mainstream reporter was smart enough to figure out that the AAA loans Wall Street was selling were really toxic loans.
Part of the distrust of statistics is the ease of which they can be tweaked or misrepresented and the average person doesn’t understand them well enough to know if they are being presented properly.
Kind of like lawyers. Most of them are good, but it’s the bad ones that make the news.
Well, the average number of arms on a human being is less than 2.
Statistics don’t lie, but most people don’t know what the hell they are hearing. Well, 90% of the time most people don’t.
Averages are not predictors. Weather people destroy the reputations of statisticians day in and day out. “Hey, the average high temperature for April 23rd in this area is 67 degrees. We are gonna be five degrees above normal today with a high of 72! Guess there is an upside to global warming!”
And I just want to choke these idiots, because to realize an average of 67 degrees, that means that some days were 72, some were 62…maybe one was 60 and maybe one was 74. The average works out to 67, but 67 is NOT normal! One might even be able to demonstrate that if it were 67, that’d be ABNORMAL!
Ugh. I would love to rip off the 1.9988 arms the average weather person has.
IIRC from another story, they claimed that a lower percentage of Republican nominees received “well qualified” ratings from the ABA than did Democratic nominees. This was, they decided, evidence that the ABA was prejudiced against Pub nominees.
It is at least as good evidence that Republican nominees are not as well qualified as Democratic ones.
On a show about cars on PBS the other night (NOVA) they stated that one type of car produced “4 times less emissions” than another car. I hear this kind of thing a lot; what does this mean? Seriously, do they mean 1/4 of the emissions? If so, why not say that?
