In the United States one of the biggest covered stories for the build up to the election is about election polling, why is the news so focused on them and what affects their accuracy.
The purpose of a poll is to get an accurate representation of what a population thinks on an issue or candidate. The election polling in the U.S. is done to predict the outcome of the U.S. election. The U.S. election and primaries are themselves polls because only 60 percent of eligible voters actually vote in the national poll. So, how do polls work?
Polling in the U.S. is done by private companies and several colleges. There are a few National pollsters such as Rasmussen, and the Gallop poll. The vast majority of organizations are focused on local polling. Then there are a few poll aggregators who take the data from many polling organizations and try to combine them to try and represent a more accurate outcome for the election. The key to making polls accurate you may have learned in a statistics class. The two things you have to know are the population, everyone who will be voting and sample size, which is the number of people taking the survey. Its accuracy will be affected by the ratio of population to the sample size. The larger the sample size the more expensive the process will be, but the more accurate it will be. Based on this you get the margin of error and confidence interval. Confidence intervals are how sure you are that these results will be reflected by the outcome. Margin of error is the range of values used to change the confidence interval. For example the odds that (insert the name of your candidate here) is going to get 50.000000 percent of the vote is very slim, but the odds of them getting between 60 and 40 percent of the vote is a lot higher. You could say there is a 100 percent chance your candidate is going to get between 0 and 100 percent of the vote, but that would not be useful. Statisticians play with these numbers to get a number that is accurate, but has a high chance of being correct. This can be problematic in some cases. For example in the Brexit debate the polls Bremain was supposed to win at 51% with high confidence, but the margin of error was 3 percent. That means that when Brexit won it was still in the margin of error, but this is not the only way polls can be wrong.
The other key part of this is whether or not the sample you used is representative of the population. In theory you would do random sampling where every person in a state has an equal chance of being selected, but since that is impossible statisticians have come up with a few clever ways to try and make samples more accurately represent the population. The most commonly used technique is stratified sampling which is where people are put into groups. For simplicity sake the groups I will use are gender and race, but these groups are not homogenous and high variance within those groups. If you poll your sample, and then ask them demographic questions. You can then average their responses by group. You then take for example white males and multiply it by their percent of a state’s eligible voters and add all the groups up to get a more accurate prediction of election outcome.
One error that can occur is selection bias. If you asked at the unemployment line about the economy or outside a catholic church after mass about Roe v. Wade that would affect the results of a survey and its accuracy for the general public. The method of polling also effects this. Robo polls, or automated phone calls, use to be the most commonly used method. As the number of people with a landline go down this method becomes more and more skewed. Some pollsters use online polling, but that has the problem of not being able to limit the number of times a person can vote. The presidential primaries have selection bias built in by having a closed system. That means it is not open to non-party registered voters. If either side allowed independents to vote in their primaries it would pull it more towards the center, and more likely to win a general election than their competition. Skewed polls don’t really help anyone. It just makes predictions less accurate.
Polling does have some effects though. It is used in the selection process for the debates, and was why Gary Johnson was not present. It also is used by candidates to see their chances of winning, and it is used by the candidates to adjust their platform to try and appeal to more voters.
I hope this has informed you for the upcoming election and come November 8th I hope you vote for whatever candidate you believe to be best, and November 9th we can all come together, and fix any problems America may face.
I hope you enjoyed this article, and maybe even learned something. Feel free to share with your friends on the social media. If you want to read more of my other content you can here or follow me on twitter. Please stop by next week for more articles on news, economics and international affairs only available at Odyssey.
