Monday, January 7, 2008

Why I Don't Trust Opinion Polls

Polling is probably one of the biggest farces in the political arena. It is the best way to tell the common man that he isn't all that common and should conform to a new standard. More often than not, polls are used by media outlets to provide a political bias while pretending to be "above the fray." I will illustrate this simple fact in a little bit, but I need to better define what opinion polling is.

Opinion polling is simply surveying a small group of random people in order to get attitudes related to important news stories and political leanings of the general public. The basic idea is that a random sample of people can really represent attitudes, opinions, or projected behavior of all people. Call it psychohistory-lite, but it is the belief that there are more people than attitudes (or options) for a single subject and thus with a randomly selected sample of a certain, we can determine the attitudes and opinions of everyone within a certain region.

Already you may be seeing the problems inherent in this form of study. Firstly, because everyone is an individual with their own set of opinions and beliefs, there could always be a different perspective on any issue from abortion (I once had a roommate who believed half-jokingly that women who have third trimester abortions should be forced to consume the remains), right to bear arms, health care, etc. Frankly the premise of this form of information gathering is patently ridiculous.

Now for some not so obvious problems:

1) Polling questions and answers can be skewed to meet a particular result. For example, you if I asked you if you wanted to fund the local government skool with higher taxes or government bonds, most likely you would answer government bond. But you would be cheated since the scope of the question and the answers were limited to my boundaries. I never gave you another option like having parents pay some form of tuition, allowing for a school voucher program, or not spending that money at all. The question of whether the government skools really need it or where it would go never comes up.

2) Polling samples can easily be skewed either deliberately or inadvertently. This usually depends on the samples taken and the methodology used. For example, I could call 1000 people over a period of two weekdays between 10:00 AM and 3:00 PM and publish the results. If I publish them in a scientific magazine (assuming someone in the scientific community would publish it), then I would given the equivalent of a public whipping for such a flawed sample. Yet I probably could get it published in a mainstream newspaper as a fact. But the fact is, if I were studying a general case, this method of sampling would be seriously flawed because I would only talk to jobless bums or retirees (old jobless bums). In more extreme cases, people can really skew their results purposely to get desired results. Alfred Kinsey's studies on human sexuality can fall into this category as one of the most glaring frauds of the 20th century. Assistants have said he threw out 3/4 of his data when it came to adults and he probably allowed pedophiles to rape children in order to get desired results in his study (no one ever asked how he managed to get the data in relation to children orgasms or what qualified as children having an orgasm).

3) The sample size. Really, are we suppose to believe that 1000-1500 random people would accurately reflect the attitudes of 150-200 million people? Think about that.

4) The reporting on the polling results is bias toward a particular agenda. I can't think of a more glaring example of media publishing their own bias in polling then from this poll result on Socialized medicine. While the poll results indicate that people want the government to guarantee health care to all, the article associated with it either failed to mention or toned down other results like the fact that 77% of those polled were satisfied with their own health care and that the majority of those polled (44%) didn't think the government could do a good job (there were three options: private sector, government, don't know). Yet the headline of the article and the content of the story basically said that the majority of the American people wanted Socialized health care. If you analyzed the poll results, disregarding the skewed questioning, poor data collection methodology, and small sample size, then you'd see that most people are worried about other people's health care and not their own. But that wasn't what was reported at all.

While I have touched on the various reasons why polling data is probably the most useless information, save for celebrity gossip, you could read, I'm sure there's a lot more. In any case, I don't trust polls for those primary reasons listed above, regardless of the favorable result they give. But that doesn't mean I won't stop paying attention to them. Why would I pay attention to such garbage even though I don't trust it? Simple: because, as Rush Limbaugh once said, politicians believe it, especially those in Congress. I guess being a lawyer was a better option for them then being a statistician or mathematician or engineer. And people are dumb enough to keep on voting for these folks.