By Stu Weisbrod
One of the most frustrating aspects of our politics is the failure to debate issues using facts as the basis for rational discussion.
I’d like to illustrate this calling upon my background as a statistician and the polls of this election cycle. Some on the Right were absolutely convinced that Mr. Romney would win on the basis of information they received from Fox News, and others in the mainstream media, which hoped to promote a horse race to the very end. In fact, the overall polling picture as seen in sites like Nate Silver’s 538 blog, the Princeton Election consortium, or Electoral Vote (.com) all showed President Obama with a lead, sometimes sizable, throughout the entire election cycle and all these sites just about hit the final election results right on the head.
For example, the Princeton site had Mr. Obama with a 90% of chance of winning the election back in July on the basis of the state polls that were being conducted. This probability grew eventually to 99% right before the election. I have abstracted a table from Mr. Silver’s 538 blog here to illustrate where he shows the final election results for each pollster. The average error is how far off their polling results were compared to actual results, while the Bias represents whether they had offsets toward either the Democrats, indicated with a “D”. or the Republicans, indicated with an “R”.
There is one primary conclusion that can safely be reached. Reliance on ANY single pollster can be a recipe for failure.
Those who ONLY chose to look at Rasmussen or Gallup, typically on the Right side of the political spectrum, and Fox, were left with a 4-7 point GOP bias which conveyed an incorrect picture of the election. The results were a total surprise to many on the Right because they refused to look at ALL the data.
The solution in this case is to look at ALL polls, which these other sites did. One can never be sure which pollsters will be correct in any given cycle. They all make assumptions on how to weight the data and how they contact their subjects. As can be seen, most pollsters actually had a bias toward the GOP this cycle, although there might have been a break toward Democrats in the last week or two of the election that is not fully comprehended in this data.
The use of facts to explore issues, and find reasonable solutions, extends beyond the reporting of election polling to other significant problems our country faces today.
It becomes extremely difficult to solve these issues when data, or facts, become irrelevant and are replaced with spin, or ideology. Whether it is man-made climate change, the deficit, taxes, or education, the discussion has to start with facts and data, not spin.
Consider this as you make your choices on what sources you read, and whose ‘news’ you listen to, in order to make more informed decisions about the key issues we Americans face today.
- “Which Polls Fared Best (and Worst) in the 2012 Presidential Race”, Nate Silver, November 10, 2012, NY Times 538 Blog
Leadership Voices 