This past election has been one of the most polarizing and heated on record. Both sides engaged in name calling and other not so classy campaign strategies. But no matter a persons political beliefs, not many people expected the outcome that was delivered on November 8th. This is due in large part to the countless polls that were taken before the election, almost all of which projected a Clinton win. And the polls weren’t coming from the local high school stats class, these polls came from reputable sources such as the Wall Street Journal, and CBS news. Even Fox News had Hillary winning by 4 points. These polls were constantly shown and talked about on the news and many in the political world gave trump next to no chance just days before the election, but in the end he pulled it out. So this begs the question, is there something wrong with the science of polling?
Going into this my null hypothesis is that there is nothing wrong with the science of polling, while my alternative hypothesis was that their is something wrong with the science of polling. To try and answer my question I first wanted to try my hand at polling. So a couple days after the election I set out to ask people how they had voted. I ended up asking about 30 people, when I asked them I would ask them two questions, one, who did you vote for, and two, did you change your mind about your vote at anytime within two weeks of the election. The second question is to ensure that the answer I got would be the same answer that a pollster got had they called the person. of the 30 people I asked, 22 did not change their vote, and of those 22 people, 13 stated they voted for Trump, 8 said they voted for Clinton, and one said they voted for Gary Johnson.
Had I taken my poll before the electio, and simply taken the data at face value, I would have correctly predicted the outcome of the presidential election. Now this is most certainly due to chance, and my poll committed just about every sampling error in the book, but nonetheless it was still right. And this raises the question of whether or not chance plays a bigger part in polling in today’s world than it did in previous decades.
Many articles were written about possible reasons for why the polls were so wrong and one reason stuck out to me, many people suggested that it simply much harder to poll people accurately today than it was 20 or 30 years ago. This is in large part due to the increase in cell phones and the ability to screen calls. Almost all polling is done over the phone, and 30 years ago most families had one house phone, and a person couldn’t tell who was calling until they picked up the phone, so more people answered their phone because they had no reason not to, and thus pollsters were able to get more information. But today this is not the case, house phones are more or less a thing of the past, almost every adult has a cell phone, and they are not likely to answer it if they don’t recognize the number calling. This is due to the rise of telemarketing as well as scams. And this presents a problem for pollsters, because their predictions rely on getting reliable, well distributed information, and if they cant get that then their polls are likely to be skewed. Based on these facts, I reject the null hypothesis and conclude that there is something wrong with the science of polling, in that it has not adapted to modern times, and unless it does, people will continue to see inaccurate polls.