The next presidential election is still eleven months away yet we’ve been bombarded with numerous political polls for over six months. Have you ever noticed how rare it is for two polls taken at the same time in the same area to produce the same results?
Each polling source uses different questions, target groups and sample sizes among other variables, yet most of the media just report the numbers as if they are the gospel truth. A good example is the two latest polls taken in Iowa.
The CNN poll of possible GOP voters in Iowa has Donald Trump leading with 33%, Ted Cruz with 20%, Ben Carson with 16% and Marco Rubio with 11%.
The Monmouth University poll of possible GOP voters in Iowa has Ted Cruz leading with 24%, followed by Donald Trump with 19%, Marco Rubio with 17% and Ben Carson with 13%.
So why the big difference between the two polls supposed taken from the same general group of GOP voters?
Michael Traugott, a polling expert with the University of Michigan offered an explanation:
“The first thing I notice is that the sample size is only 425, with a [margin of error] of 4.8 (5)% points. They also modified the sample design from the previous poll to include some general-election voters as well as caucus participants. This has two consequences: The differences between the leading candidates is unclear (MOE) and the trend from the previous poll is not easily discernible (design shift).”
Those behind the CNN poll also noted the difference between their poll and the Monmouth poll, stating:
“The Monmouth poll interviewed a sample drawn from registered voter lists that primarily comprised those who had voted in state-level Republican primary elections in previous election years.”
“The CNN/ORC Poll drew its sample from Iowa adults, asking those reached about their intention to participate in their caucus, interest in news about the caucuses, and past participation patterns to determine who would be a likely voter.”
I recall my advanced statistics professor in college stressing the point that anyone can make statistics say whatever they want by controlling the source of the statistics and then how they manipulate them. We were given an assignment to design and conduct a poll for the same purpose. Three weeks later, we all submitted our poll and results to the professor. Even though we were all working on the same topic, all of us got different results and different conclusions.
That lesson taught me years ago that you can’t trust polls. The past few presidential election early polls have also taught me that you can’t trust poll results. Most of the early leaders rarely end up as their party nominee.
While it’s fun to watch and report on the different polls, please remember that every poll is very subjective and may not accurately report reality.