Pages Navigation Menu

The blog of DataDiggers

Categories Navigation Menu

Should we trust polls?

Should we trust polls?

With election season fast approaching, we have decided to make a discussion on this topic and see if the polls we all seem to be so keen on, are to be trusted or not.

It’s no surprise that the news cycle is brimming with polling. You’ll be bombarded with the new “statistics” about the number of people from planet Earth who believe in God; the distribution between dog-lovers and cat-lovers; and how many people between the ages of 65 and 85 own a Nintendo.

First, some clarification: The object of any poll is not to predict what will happen. Instead, a poll is like a public opinion barometer which tries to measure how people feel about politicians during a short window of time as accurately as possible. Clearly, those feelings have a major impact on who wins elections at the top, so they’re what most pundits use to forecast an election result.

Political polling wasn’t always so scientific. In the late 19th and early 20th centuries, journalists would conduct informal straw polls of average citizens to gauge public opinion on politicians and upcoming elections. In the 1930s, the influential “Literary Digest” magazine published via mail and telephone public opinion polls of its large subscriber base, assuming that a large sample size would inevitably produce infallible outcomes. The magazine failed to notice that its readership was richer and more likely to be Democratic than the typical U.S. elector, causing the paper to erroneously forecast Alf Landon’s landslide victory over Franklin D. Roosevelt in 1936.

 Today, predicting the future is difficult, but some suppositions are stronger than others. Later polls are less likely to be reliable, as there is more room for some game-changing event (scandal, economic collapse, meteorite obliterating D.C.) to occur between the survey and the final voting. Perhaps candidates could drop out of the race, freeing up potential voters to give someone else their support. Then, there are individual flaws that can affect outcomes: people may say they’re going to vote and then choose not to. Can we blame them?

It must be said, from the outset, that the science of public opinion polling is not accurate. There is always a margin of error. In ordinary surveys, this margin can go up to 3%. It doesn’t seem like much, but it can be the difference between 4% and 7%. That is, whether or not to have Members of the European Parliament. Or win the presidential election.

The problem is that sociologists’ lives are compromised by certain factors. Another such is the so-called “spiral of silence”. When people feel their view is a minority, or even stigmatized, they would refrain from openly sharing it. But they’ll convey that in the vote booth’s safe. Another thing that can happen is that some people are simply not honest about their preferences.

Now, given the state of the polling in 2016, and despite Hillary Clinton’s inevitably mistaken expectations of a victory, can we trust polls that running into 2020?  Here are five reasons why we should be more distrustful of political polls than ever before.

  1. Media can be fallible

You may have noted this: a new poll – often a poll commissioned and released by a media company – comes out and is heavily publicized by the media. In doing so, the media often increase the name recognition of whichever candidate is in the lead. As a result, that candidate becomes even more well-known and gets an extra boost the next time people are surveyed. Donald Trump was 2%polling before announcing his candidacy. Immediately afterwards (when the newspapers committed 20% to 30% of the campaign headlines to one candidate) his interest soared to 11%.

  1. Journalists can be fallible

You’ve read that right, even journalists who are scientifically literate, may be stupid sometimes. It’s the reason that just about everyone has failed to predict the spectacular rise of Trump. That’s not just because the polling methodology itself is faulty (more on that below), it’s also because the person performing the research is influenced by it. Personal experience and personal beliefs get in the way. Many of the journalists who dismissed Trump with projections that he had a mere 2% chance of winning simply didn’t know any potential Trump supporters in their personal lives – they didn’t get it. And their personal beliefs also encouraged a bit of wishful thinking – they didn’t want to get it.

  1. Predicting the future is not easy

Consider how different your present and future responses would be to questions like: “What would you eat if you had to eat your lunch right now?”Or” If you had to pick a romantic partner right now, who would you choose?”You might argue that food, romance, and politics are not all that similar, but the answers all point to a fundamental and consistent human preferences truth: things change.

  1. It’s hard to contact people

41% of U.S. households had a cellphone in 2013, but no landline, and that number is rising. This poses a problem for polling companies because the Telephone Consumer Protection Act of 1991 means that these cell phones cannot just be autodial. Not being able to autodial means that interviewing certain people is incredibly costly and time-consuming for companies.

  1. Most people don’t want to be contacted

Response rates (that is the percentage of people who respond when questioned to a survey) have plummeted. It was over 90 percent in the 1930s; it was 9 percent in 2012, and since then it has continued to decline. Whether they’re cold calling or they’ve created a panel of individuals that they’re able to survey repeatedly, polling firms all face the same basic issue of how to encourage people to answer questions.

We thought of explaining some recent examples of polls that garnered big headlines but that turned out to be innacurate.

  • On August 13th, reports blacked about a Fox News poll rating the Democrats and saying Donald Trump would lose to each of the top three Democratic candidates if the election was held today. This poll reached the mainstream news echo chamber for a week, as a variety of doomsday articles were created by reporters. The statement of course hit the news for another week. Here too the results appear to be distorted. This points out that 8% more Democrats than Republicans were interviewed by Fox News.
  • Only remember the Trump vs. Clinton campaign in 2016 for evidence of the distorting results of such polls. Of the twenty polls conducted by some of the US’s most reputable news sources, only one (IBD / TIPP) gave Trump the lead on election day. Most of these polls gave Clinton a 70 percent chance of winning and some were persuaded by 99 percent. Not unexpectedly, a month before the election, CNBC released an article titled, “Trump Win All but Possible Based on Recent Races.” CNBC was obviously pretty far off the mark. Once, attention to the data. 43% of the people interviewed in the October 2016 study were Democrats, 36% were Republicans, and 16% appeared to be Democratic. It’s probable that seven percentage points are more than just a rounding error, based on their estimate and the actual election result.

As said in the beginning of our discussion, the election period is approaching, including in our country of origin, Romania. It is important not to be fooled by appearances and not to rejoice earlier than is the case. This is true in any other state. This type of polls cannot be trusted, for the reasons mentioned above, so it is preferable to be cautious, not to believe anything that appears in the media and to look only after reliable sources (if any).

Of course, it is up to everyone to choose whether or not they want to believe what is predicted in election campaigns. We do not want to influence anyone, but we hope you will remember what was said and make the right choice on the ballots.

 

468 ad

Leave a Comment

Your email address will not be published. Required fields are marked *