Polls Can be Decieving

When you are listening to poll numbers braodcast through the media, make sure you find out more about the particular poll. What is the margin or error? What is the confidence level within that margin of error? How was the poll conducted? What kind of sample did they use? What is the poll really telling you?

For example:

Real Clear Politics Average shows Obama up by an average of 7.3% with one poll show Obama up nationally by 14% and another showing Obama only up by 1%. If you look closer at the polls you see they are all sampling the entire US population, though they are using different methods with different margins of error and confidence levels. What does that really tell you? It says if you were ask everyone in the US who they support that the research company is confident at whatever level they express (usually expressed in terms like 95% confidence level, if it is under 85% confidence the poll sucks) you will find that the real numbers are within the margin of error (usually expressed as something like +/- 3%).

IBD/TIPP Polls show Obama ahead by only 3.7%* with a margin of error of . Why the big difference? Well for one IBD/TIPP polls different people. They only poll likely voters, or people who are statistically likely to vote this year. In other words, they are confident that if you ask all likely voters that you will find that Obama's lead will be 3.5% above or 3.5% be the declared 3.7% lead.

* New IBD/TIPP Poll shows the lead for Obama down to 1.1% with the same 3.5% margin or error.

So what do these polls mean as to who will win the election? Well not a whole lot since even IBD/TIPP polls show that with a week and a half to go until election day somewhere between 9%-15% of likely voters still don't know who they will vote for.