SurveyGizmo Blog

From survey design all the way through taking action, you'll find tried and tested survey knowledge on each and every page of our blog.

Can You Trust Political Polls? Maybe.

Abraham Short Oct 14, 2016 0 Comments

Listen to any newscast, or read any newspaper, magazine, or online thinkpiece and the odds are that you are going to see at least one political opinion poll being touted. This is nothing new, Political polls have been with us almost as long as there has been politics. Which is to say, as long as there have been people. But how do they work?

What is truly amazing is that for something that carries so much weight in our collective consciousness, very few people, including the people who report these polls, really know the art and science that is behind them or the difficulties that are involved in producing results that are truly representative.

In this article, we try to shed a little light onto the murky processes behind political polls, how they are used, and their importance in modern day politics.

How Are Political Polls Done?

The mechanics of how polls are performed is simple in theory, but difficult in practice. The simplest explanation is that you ask a representative group of people to answer a set questions and then tabulate the results based on their answers.

But, finding that representative group is no easy task. In fact, this is where the problems with political polling really begin: with how poll collectors choose to find their audience.

Political Polling via Telephone Polls

Sounds easy, doesn't it? Just call up a bunch of numbers, and you have your representative data – right?

There are some difficulties major that have to be overcome, though. If you want to develop meaningful results, you first have to be able to elicit responses from a truly random representative group.

Traditionally, this is most often done by utilizing a random digit dialing (RDD) system, but with the advent of caller ID and cell phones, this has become much more difficult. In fact, response rates in some areas have dipped below the 10% mark.

Not only does this mean that pollsters have to make more calls to garner a minimum usable number of responses, but it has also been found that women and older people are more likely to answer the phone and stay on the line long enough to give valid responses. This leaves men and younger people, who are less likely to even have a landline, as underrepresented in the samplings. An unbalanced sampling leads to skewed results.

Online Political Polls

You may think that moving polls online would be the answer to all of our polling problems. Unfortunately, that's not the case. Online polls aren't without their own set of headaches.

While it may seem like the internet is everywhere, it is currently estimated that 16% of the people in the United States still don't use the internet at all. In addition, online polls tend to over-represent men and people who are unemployed. Again, this skews the data.

How (Some) Political Polls Use Weighted Averages

To overcome the bias that is hard to avoid in either system, pollsters will often attempt to adjust their results with the use of weighted averages. That is, they will adjust their results to more closely reflect actual society.

Say a pollster knows that only 25% of the respondents in their sample group were men, but census numbers tell them that 50% of society is male. They would give the information received from male respondents twice as much value or weight when calculating their results. Problem solved.

But not really. This is a far from a perfect solution and introduces added odds of miscalculations as well as the opportunity for data manipulation. So polls that are "fixed" using weighted averages can be just as unreliable as those with more obvious representational deficiencies.

Are Unified Polls the Answer for Political Pollsters?

Another approach to finding more representative sample groups is what is called unified polling. This method incorporates sampling from both landline and cell users with data gathered from online polls to help create a more universal and equitable picture of the population. For organizations that have the deep pockets required to support it, this is quickly becoming the sampling method of choice.

However, it does require investment of time and money. Not every media outlet or research group is willing to put in that kind of effort!

 

What Methods Are Good for What?

There are many different methods used by various political groups and the pollsters they employ. For purposes of privity, we will limit ourselves here to the three most common types of political polls you are likely to find. In truth, this section is probably the most important when it comes to determining the true meaning of the numbers that you will come across in your daily lives.

Random Polls

A random poll is what most people think of when they see polling data. It is a poll developed with one of the methods in the above section and provides, as much as possible, a snapshot of the current overall attitude of the population at large. There are many possible flaws, but these polls are common and easier to administer.

Many politicians use random polls as guides in determining what policies are most important to their constituency and how the community views recent political moves that have been made or are about to be made.

Tracking Polls

Tracking polls are very much in style during this current election cycle. They are similar to random polls, but with one very important difference. Where random polls are constantly in flux and subject to large swings due to them basically being recreated with a fresh sampling each time, tracking polls deal with a consistent group of respondents because they ask the same people are contacted over time.

In the majority of tracking polls, extra effort is made to avoid extremes in the sample group and stay more towards the middle of the political spectrum. These are what politicians would call the swing voters. Once this group of swing voters in established, it doesn't change. The same group is canvassed repeatedly and the ebb and flow of their positions closely monitored.

Tracking polls excel at determining small shifts in attitude and opinion that might not be as apparent in other polls. And, since the group is politically involved enough to become committed to the polling process, if done right, they are more likely to more accurately represent actual voters and their sentiments.

Be Wary of Political Straw Polls

Straw polls are a breed unto themselves.

A straw poll is one in which a pollster asks a group, any group of people that happens to be around to give their opinion. Classic examples of this are entry and exit polls performed at voting centers on election day and used in this way they can be valuable for predicting the final outcome of an election hours before the polls close and the votes are counted. They do have several disadvantages, though.

They are very limited in scope. This was most recently illustrated in the first Presidential debate between Donald Trump and Hillary Clinton. A straw poll of the media types present, immediately following the debate, gave Hillary top marks and declared her the winner by a landslide.

However, by the next morning, it turned out that both the random polls and tracking polls told a very different story and saw Trump gain ground in almost every demographic category.

Straw polls, due to their limited time window, are also subject to the vagaries of daily life. The Classic 1948 Chicago Tribune headline "Dewey Defeats Truman", is a prime example of this weakness. Dewey was indeed in the lead when they sent the paper to print, but what the Tribunes reporters failed to take into account was the late voting by union loyalist that turned the tide of the election. These poor blokes were still having to work when the paper declared Dewey the winner.

 

Are Political Polls Biased?

This is the $64,000 question: are polls biased? The honest truth is... sometimes.

There are polls called "Push Polls" that use manipulative worded questions to not so much gather valid statistics as to either sway voters or create numbers that can then be used to strengthen a candidate's perceived support.

Outside of these, the majority of political polls are created with the hope of gathering true and useful insights into how voters actually feel and think. This isn't to say that they are always completely neutral, though. Pollsters, in the majority of cases, work very hard to carefully word their questions and even to arrange the order they ask them so as not to lead members of the sample group toward or away from certain answers. However, they are human and the people they question are human so there is always the danger that some bias can sneak in unnoticed.

As an example, if they are taking what is known as a "Horse Race Poll" (that is, a poll where people are asked will you vote for candidate "A" or "B"), they will ask this question before any others. The idea is to not influence the person being asked with any information that might be included in other questions.

 

How Do Campaigns Use Political Polls?

It might be easier to make a list of ways that campaigns don't use polls. To a greater extent than most people care to admit, politics in the United States, from a local level to the presidency, have become as much about appearance and appeasement as it is about what the candidates actually stand for.

Add to this the simple fact that the costs of political campaigns have gone through the roof, and it is easy to see why polling has become so important to every phase of a campaign.

 

In Politics, Knowledge is Power

For major campaigns, before a candidate even decides to run, there will be polling performed to see if they have or can develop sufficient name and face recognition to make running worth the effort and expense. Then they will have polling performed to see how their past voting or statements on key issues will line up with the demographics they consider must haves to win. It is only after this initial research that a major party candidate will officially enter a race.

From this point on, many candidate's lives will become ruled by the numbers they receive daily. Everything from the subjects the include or leave out of speeches will be decided by the what the polls tell them people like or dislike about their core message.

As absurd as it may sound, even the clothes they wear and how they wear them may, to some extent, be influenced by data gathered in opinion polls. Take off the jacket and roll up the sleeves when you visit Iowa. They like working men there. Visit Texas, throw that tie away, Voters there prefer the open collar look. Heading for New York, put the suit back on but wear the red tie, not the blue one. It projects power and authority that will play well to NYC audiences. All of this type of information is gathered from studies and followed very closely.

Then, of course, you have the true political polls that must be carefully monitored throughout the campaign. These are numbers you see in the paper each day and on the news each night. These let the campaign managers know where they are gaining ground or losing the fight and where they need to focus their resources. You have a comfortable lead in the 18-25 female demographic, but seem to be slipping with males over 55. It may be time to reel in on the free childcare program and spend more money and time spreading the word about your support of increasing and protecting social security benefits for the future.

The point is that no bit of information is too small to be of some value to a candidate that is scratching for every vote they can get. On the campaign trail knowledge is power and the only way most candidates can really gain that knowledge is through the use of polls.

 

Political Polls Have Influence

The other side of this story is that polls, when properly promoted, can be used as a weapon to influence the outcome of an election.

Voters have become lazy. Roughly only 60% of Americans bother to vote in any given election. If a candidate can make it appear that they have an insurmountable lead in a race. It can make it very difficult, if not impossible, for their opponent to mobilize the support they need both on the campaign trail and at the polls.

Then, of course, you have the Push Polls that we discussed earlier. Though greatly frowned upon by honest polling companies, their use has greatly increased since negative campaigning has become the norm. The idea here is to do a poll that focuses not on your candidate's strengths, but on the negative aspects of your opponent. This makes it possible to make them look bad even if you haven't done anything to make yourself look good. This is a very strong tactic used in campaigns that have image issues of their own or who find themselves in weak positions.

As you can clearly see, Political Polls are not a simple subject. We have only scratched the surface in this article. Still, I hope that you have gained some insight into the complex world of pollsters and better understanding of the major role that political polls play in our lives.

Spread the word
Try SurveyGizmo out!
TRY BUILDING A SURVEY

By accessing and using this page, you agree to the Terms of Use. Your information will never be shared.

Be in the know!

Subscribe to get regular updates and best practices for survey creation, data collection, and data analysis.