Poll Cats

The only poll that matters is the one on Election Day.

   Telephone surveys, the method most pollsters use, is at best a guesstimate, and pollsters always hedge their results by saying a report is within several percentage points of full accuracy.
   This means, for example, that if a poll is said to have a margin of error of three percentage points, and depending on which point spread you choose, either candidate can be declared the "winner" by six points, or they can be said to be tied.
   So who won, really?
   Even with a lead of ten points, that margin can be narrowed by selecting numbers within the margin of error spread. Other polls conducted by other agencies will often show different ranges and results.
   A second factor is phrasing. What is the question, and how is it presented? Questions can be phrased to elicit answers favorable or unfavorable to any given issue. Reliable, ethical polling firms try to avoid posing slanted questions, while others deliberately phrase questions designed to prompt answers favorable to one side.
   As for telephone surveys, they face a problem endemic in America today. Many young Americans do not have landlines, and laws prohibit calls to mobile devices and cell phones.
   Therefore, many surveys are inherently biased toward older citizens with traditional landline telephones, and their political views are likely to be quite different from those of younger folk.
   A classic example of built-in survey bias to telephone households is the presidential election of 1948. At the time, telephones were more commonly found in middle- to upper-income households than in workers' homes, so the survey population was inherently biased toward Republican candidates.
   Moreover, voting places throughout the country were in many areas still open in the evening, when workers -- who leaned Democratic -- were more likely to go to polling places.
   Result: The classic Chicago newspaper headline, Dewey Defeats Truman.