Statistics are a powerful tool in order to understand error, probability and optimization. The problem is that there are too many statistics and not enough people educated to understand them.
Polls and statistics permeate through our culture: discounts, percentages and opinion polls. As a mathematician, it irritates me that this beautiful and elegant tool has been used to manipulate and take advantage of good people.
I remember in the 2000 presidential election, hanging chads, dimpled chads, pregnant chads, butterfly ballots, computer malfunctions and other factors caused inaccuracies in Florida.
As Brian Greene put it in his book, “The Best American Science and Nature Writing 2006,” “Usually such sources of inaccuracy can be ignored because they’re too small to have any effect on the election’s outcome. But when the vote differentials are not in the hundreds of thousands or in the tens of thousands or even in the thousands, the built-in inaccuracies compromise the entire process.” The appropriate question should not have been “who received more votes?” but “what should we do when two candidates, to within the accuracy of our measuring apparatus, are tied?”
I remember in the third week of 2004, the Gallup and the Pew both released public opinion polls. You would think that these polls would be identical (or at least pretty close) considering they were released on the same week, but the Gallup poll showed President Bush and John Kerry at 52 percent and 44 percent respectively and the Pew poll showed Bush and Kerry tied at 46 percent. Why is there such a difference?
In September of 2007, Army officials argued that there was a 75 percent decrease in sectarian attacks. Of course what they didn’t say was how the statistic was constructed.
“If a bullet went through the back of the head, it’s sectarian, if it went through the front, it’s criminal.” So all they did was redefine what a sectarian attack was in order to cherry-pick favorable statistics. I wonder if this angers anybody else.
Here is a common trick of the trade: when a magazine or poll says “average,” that word is specifically used to be vague. Average can mean any number of things, which include the arithmetic mean, the median and the mode, and depending on how the sample is skewed, the numbers can be drastically different. Which brings me to the fallacy of the day: equivocation.
This fallacy is committed when a key word is used with two or more different meanings. For instance, “All jackasses have long ears. Flank Huntz is a jackass. Therefore Flank Huntz has long ears.” Likewise, depending on the statistic and what people would like to report, average can mean any number of things and can vary in millions of dollars. Only when distributions are normal do we get all three to line up, and when we are talking about money and income (for those of you not in touch with the rest of the world), it’s almost never without skew.
I’m not saying that all statistics are useless or that all polls are pointless. But before you become convinced by some sales-person or politician or by all of the pretty pie charts, pictographs, numbers and percentages, ask these questions: What does the statistic actually mean? Who is providing the statistic? What would they want the statistic to mean?
We live in a commercialized world and there are people who are more than willing to exploit your trust and manipulate you. Your vote, your money, your signature, your time and your patience should all be very important things to you and should be guarded thoroughly.
As Darrell Huff put it in his book “How to Lie with Statistics,” “The crooks already know these tricks; honest men must learn them in self-defense.”
By the way, it has been calculated that 95 percent (plus or minus two percent) of all statistics are B.S.
I made that up. It’s just that easy.
About the writer:
Ken Ueda is a senior math major and can be reached for contact at [email protected].
