We’ve all heard the line, “the only important poll is the one on election day.” That is true, but too many Americans and media personalities place too much emphases on polling as the indicator about how the race is going to end up, despite the fact that more and more over the last several election cycles, the polls did not predict exactly what the outcome was. People don’t understand that scientific polling and balloting are by design two different things and the only actual accurate measurement is the results of the ballot box tabulation (computer hacking, suppression, cheating and fraud not withstanding). Polling involves asking a fraction of the number of people who may or may not vote and predicting what the outcome would be. However, it’s not as simple as you would think.
Representative sample doesn’t mean that the answer everyone gives is counted equally to the others asked. In true scientific polling, the demographics of the person asked are subjected to historic voting trends. Those demographic measures taken, applied to a complex algorithm and each participant’s answer is weighted differently. Some think that if 25% of 100 people asked said they would vote Democratic then 25 of the 100 people asked said they would vote Democratic. That’s not how it works. Demographics coupled with historic trends give each respondent a different statistical weight.
The demographics measured include party affiliation, sex, race, age, socioeconomic status, voting history, etc. and those demographics are measured and weighted to determine the likelihood of that person actually voting the way they said they would or even at all. Demographics and historic voting trends are measured and placed in an ever changing algorithm of data to determine the relevant statistical weight of the response given, compared to the statistical weight of the response from people with different demographics. The outcome of the weighted results is what you see when polls say 25% of the people in a certain region support John Smith over Alice Jones.
The problem in today’s scientific polling results reliability is that not only are the demographic factors changing, resulting in the algorithms needing change for an accurate result, they’re changing faster that the pollsters can keep up with. As a result, they may keep a certain statistical trend in the algorithm that no longer applies to today’s world. You will note that the say that Democrats are far less likely to vote in midterms of special elections and as a result, Republican voters who are polled get a higher statistical weight in the algorithm. However, recent election results have shown higher Democratic support than the polling indicated. Why? Because there is record enthusiasm for Democrats and Independents to vote in this cycle, more so than ever before and the algorithms haven’t kept up with that changing dynamic so Democratic responses are weighted lower than Conservative Republican results.
Another constant trend that’s changing and Pollsters haven’t been able to compensate for is how the data is collected. Traditionally polls are done by phone. Only recently have pollsters gone from landlines for polling to cellphones. Elderly people tend to be regular reliable voters and often have no problem responding to polls. Younger people use mostly cellphones and are not likely to vote or answer polls. However, there’s indication that younger people are energized to vote this midterm, more so than ever before but that’s not to say they are so willing to answer pollsters calling them on their cellphones. Younger voters, millennials do not have a high historic trend of voting so their results aren’t weighted as high as elderly folks with landlines. If in fact that’s no longer the case, polling results will be undercounting pro Democratic responses. So in short, today’s “representative samples” used by pollsters to present results to the public may not be as representative as they were last year since the demographics are changing and the algorithms used are outdated.
Now I like to look at poll results as much as the next guy, but I don’t take them as Gospel, never have and never will. For any voter out there the only opinion that you should rely upon for voting is your own, regardless of what other people may or may not believe. If you want your views to be taken seriously, to be acted upon and passed by your government you have to register and demand your right to vote in each and every election regardless of whether the polls are going your way or against. Even if polling was 100% accurate, those results are only opinions of a sample representation of people taken regarding an issue or issues. They mean nothing if the voters don’t come out and vote their views.
Take polls, good or bad with a grain of salt and vote. Your views and our democracy depend on it and you.