Polling in the United States

The strengths and weaknesses of a powerful political tool


By Meshach M.


Public opinion is not something tangible that one can see, it is something much more abstract that has to measured.  There are many ways to measure public opinion, but they all have their drawbacks, and the results are not a direct reflection of the people’s opinions, they are probabilities.  However, polls generally reflect public opinion fairly accurately and give information to politicians that they would not otherwise be able to obtain.

Polls are generally conducted per request from a polling company.  A newspaper or another source that presents a poll will contact a polling company.  They give the company the questions they want asked, the number of answers they want to receive, and how they want the data to be presented as well as other logistical information.  The polling company then contacts a call center with which they are affiliated and the call center makes the actual calls.  The call center will pick phone numbers at random to call.  They often use quota sampling, where they categorize the population into subgroups and then represent the data in the subgroups proportional to the population.  Quota sampling has been proven to be less accurate than other forms of sampling because it often fails to represent equal numbers of people from various demographics. Another popular form of polling is random sampling, which uses the probability theory to equally represent a population.  The probability theory states that if you pick a large number of people at random there is a high probability that you will pick an accurate sample of the entire population.  However, when this kind of polling is conducted the results are still skewed because many people choose not to answer the phone or refuse to answer the poll.

A telephone poll is often conducted either by an automated system or actual people.  If an automated poll is conducted, where a computer operates the phone, the poll costs much less, but more calls have to be made because people are less likely to pick up the phone.  The automated polls also cannot call cell phones and the demographics of people who answer the polls are skewed.  This is because fewer individuals respond, and people under thirty are much less likely to have landlines.  However, when actual people conduct a poll, although they can call cell phones, the respondents still are likely not to answer or refuse to take the poll.  The real people conducting the polls are often third party, part-time workers, who are unfamiliar with the issues and not well trained to conduct the poll.

People answering the poll questions are often asked about issues that they are unfamiliar with, but will answer the question anyway so as not to appear ignorant about the issues. The results from these people are not a reflection of the American population, but the polling companies do not have a way to remove them from the data.  People answering the polls will sometimes purposely answer a question against their personal actions or beliefs.  This is often in cases when the questions asked by the poll have some kind of social merit and people feel obligated to answer one way or the other.  For example, if a question was whether or not you voted, people may be prone to answer in the affirmative even if that was untrue.

The way that questions are worded can also determine the outcome.  If a poll has questions that directly favor one candidate or issue, that candidate or issue will generally receive a higher percentage.  To receive accurate results from a poll, the questions have to be without bias, so sometimes a poll can be skewed because the organization requesting the poll has given questions that favor one side over the other.

There could be bias in something as simple as the order that the candidate’s names are read.  For people who are unfamiliar with the candidates or undecided which candidate they will vote for, this could make a difference. The Pew Research Center suggests that order of the names of the candidates or the issues that are read could be randomized, thus ensuring that there would be greater accuracy in the data.

In a state poll, there are usually between 500 and 1,000 people polled overall.  However, because so many people do not answer the phone or refuse to take the poll it may take up to 10,000 calls to get the 500 to 1,000 results.  Even if everything aforementioned did not occur: the bias in the questions, asking people that are unfamiliar with the issues, or people feeling socially responsible to answer the question a certain way, there is still a margin of error.  This means that the actual information could be slightly higher or slightly lower than the recorded data.  This only occurs in cases where the poll is supposed to represent a population bigger than that polled.  For instance, if an online poll were looking at how many people visiting their website liked a certain product, the poll would not have a margin of error because they are not using this information to represent a larger population.  When recording public opinion, the entire population cannot be sampled so there is always a margin of error.

In a poll of 1,000 people, there is a margin of error of about three percent.  This means that if a candidate had a 52% approval rating in the poll they could expect to have an actual approval rating anywhere between 49% and 55%.  As the number of people polled increases, the margin of error decreases, in a poll with 5,000 responses the margin of error would only be about 1.4%.

Polls are still the most accurate way to determine the thoughts and opinions of the American People.  They can often be skewed, but they give insight into public opinion, and show the way that people thought about issues throughout history.  In a country that is based on the idea of self-government, polls can help represent the ideas of the American people and hopefully those ideas can help shape the future of this country.


Work Cited

“Margin of Sampling Error Credibility Interval.” APPOR. http://www.aapor.org/Education-Resources/Election-Polling-Resources/Margin-of-Sampling-Error-Credibility-Interval.aspx. Accessed December 5, 2016.

“Election Polling.” Pew Research Center.  http://www.pewresearch.org/methodology/u-s-survey-research/election-polling/. Accessed December 2, 2016.

“Quota Sampling.” Explorable. Last modified, September 1, 2016, https://explorable.com/quota-sampling. Accessed December 2, 2016.

Patterson, Thomas E. We the People: An Introduction to American Government, Eleventh Edition. New York: McGraw-Hill Education, 2015.

“Polling FQA.” Electoral-vote.com. http://www.electoral-vote.com/evp2016/Info/polling-faq.html. Accessed December 2, 2016.

Whiteley, Paul. “4 possible reasons why most of the election polls were wrong.” Business Insider. http://www.businessinsider.com/4-possible-reasons-why-most-of-the-election-polls-were-wrong-2016-11. Accessed December 2, 2016.