Avoiding biases in political journalism should be understood, but many largely-trusted political polls lean either left or right in both their audiences and results.
Most polls have slight publicly-understood biases, just like major news outlets are understood to be more conservative or liberal. That said, the polls with huge right or left biases like TCJ Research (Rep.) or Research 2000 (Dem.) shouldn’t be trusted whatsoever because the bias overtakes fact. For a full guide of polls, visit FiveThirtyEight’s ranking here.
There are various types of polls as well, which can affect how accurate a poll actually is. A straw poll, which is conducted by interviewing random voters at an election site, is extremely inaccurate because pollsters are only in one area and are not scientifically choosing which people to interview. On the other hand, random-digit-dialing, or calling phone numbers by literally randomly choosing sequences of numbers, usually yields better results because the sample is more diverse. This method, however, doesn’t include the demographic of people without phones.
It’s become a common practice to scoff at polls, scream “BIAS” and ignore the actual trend of the election like in the tweet above. For instance, after the DNC, it made sense that Hillary Clinton was leading Trump; almost all candidates experience what’s called a convention bounce or increase in support following the nominating convention. Trump, one of the most unconventional presidential candidates in history, even experienced said bounce. Now, the polls are evening out since convention season concluded. To understand a poll, you must understand political trends that have historically occurred in elections. In 2012, we witnessed polling bias as some polls, such as Gallup, showed that Romney was up six points against incumbent President Obama, which didn’t make any sense in an election against a fairly popular sitting president. That said, it’s much easier to look back and see bias rather than identify the leaning of a poll in the moment before the results of the election are in. Below, RealClearPolitics polls battleground states prior to Election Day in 2012, and while, of course, the election didn't end in a tie, the polls are more fair than Gallup at least.
While looking at a poll, there are multiple things you must consider and remember. First, check if the poll followed the “Law of Large Numbers,” meaning at least 1,000 to 1,500 people were polled. The more people polled, the more accurate the results will be. Also, beware of push polls because these polls were conducted with a loaded question and attempt to influence – not educate – the public on an issue. For instance, a pollster could ask, “Don’t you think Donald Trump is racist for his immigration policy?” and even if someone didn’t agree, the question pressures them into answering a certain way. Lastly, do yourself a favor and compare simultaneous polls. By doing this, you’ll discover the sampling error between the two surveys, and you’ll get a more accurate picture of reality.
Despite the failings of polls, we’ll all be glued to them until November, when we’ll finally get closure to this eventful election. With that in mind, SurveyUSA only has a slight liberal bias, and Research & Polling, Inc. leans only a little right. After November, of course, we’ll find out which polls were consistently most accurate, and until then, we have to be smarter than the polls.
Lead Image Credit: Pixabay.com