Saturday, September 13, 2008


My granddad used to like to say, "Figures don't lie, but liars sure figure". It was his gruff way of reminding people that numbers and statistics, no matter how technically accurate, can be presented in such a way as to imply much more than they really say.

The point was also driven home to me in my college statistics class. I did ok -- Math is not my passion, and statistics really pushed my limits. I escaped it with a C -- and that was just introductory statistics. But what DID stick with me, no matter how much the actual mathematics started to swamp me, was the professor's constant reminder to us whenever we were presented with statistics to ask questions -- never accept uncritically the assertions made by those presenting the statistics. Ask about the methods used to gather the information. If, for instance, the statistics are from a survey, and the presenters claim that X% of people favor Y, ask questions -- how many people did you ask? How were they selected? How representative of the population in general was the sampling? What exactly did they ask the people surveyed? How were the questions worded? Was it a yes/no question? If not, how many responses were available? How were THOSE worded? As you can see, statistics involves more than just numbers.

I have been struck by the importance of this recently, given the fact that one of the Obama campaign's main points of attack against John McCain has been "John McCain voted with President Bush 90% of the time".

Ninety Percent.

It's an interesting statistic to me, because it's the exact same number quoted by the Jeff Merkley campaign in attacking Gordon Smith -- "He voted with President Bush 90% of the time".

So let's start asking questions.

The first question that sprang to my mind was "How did you come up with that figure, 90%?" It seemed the obvious question to ask, but I soon realized that the answer to that question was dependent on another more fundamental question -- How do you define "Voted with President Bush"?

Obviously you can't mean "Voted the same way that Bush did", since Bush, not being a member of Congress, doesn't vote. So what DOES it mean? Voted in favor of 90% of the bills that Bush then signed into law, and against 90% of the bills that eventually passed but he vetoed? Voted 90% of the time the way Bush actively urged congress to vote? Voted the way the Republican Party leadership urged you to vote? Voted with the majority of your fellow Republicans? Please clarify.

But ok, for the sake of argument, let's accept the 90% claim for now. My next question is, how many times did Obama vote "with President Bush", using the same standard to determine the percentage? 0%? I doubt that10%? 20%? 50%? And if Obama voted "with President Bush" say, 30% of the time (as an example, I'm not making the claim that he did), would it still be fair to criticize McCain or Smith for all 90% of their pro-Bush votes? This is important to ask because just saying "they voted with Bush 90% of the time" and presenting that as a criticism implies "They voted wrong 90% of the time". That's unfair to both the senators AND President Bush. NO matter HOW bad a president you believe Bush is, it would be unreasonable to argue that he's been wrong about everything.

Here's what I mean: If Obama and the Democracts and the Republicans and Bush could all manage to agree on, again, say 30% of all the laws passed, that means that the Dems and Bush disagreed 70%, and the Dems and McCain disagreed 60% of the time (90-30=60), with McCain and the Dems agreeing with each other and against Bush 10% of the time. So even from the Democratic POV, arguing that Bush was wrong 70% of the time, McCain was only wrong 60% of the time. Given the way that the country is so split down the middle on so many issues, do you really think that's a strong criticism of your oponent?

No comments:

Post a Comment