With the presidential election coming up we may be looking to the polls to try to figure out it Trump or Clinton will prevail. You may wonder, “can I trust the polls?” Well, it helps to know a bit more about the process and be informed.
One thing you may be wondering, “what is margin of error?” You may see a margin of error of 3% with a 95% confidence level. Most people see that but don’t pay attention to it. So, “what does that mean?” That means that if this poll was conducted the result would be within 3% of the reported result 95 out of 100 times.
So if Trump is leading Clinton by 1 % in a certain poll or vice versa, and that poll has a 3% margin of error with a 95% confidence level, that means we can’t really say for sure that one candidate will win over the other. In other words, don’t let some poll that the media displays on their news convince you not to vote because your candidate is losing or your candidate is, “so far ahead it doesn’t matter.”
You may also be wondering what else can affect poll results. Well many things. One could be researcher bias. You really have to look at who conducted the poll. Polls may vary depending which news channel conducts them. You may have heard about bias in the news media. If there is such bias it can translate into how the poll is conducted.
People can change the result of the polls by only surveying certain people. For instance if you surveyed a higher percentage of Democrats or Republicans you would get different results. This is just one of the many ways you could introduce bias into a poll. Sometimes a pollster may affect the outcome without even meaning to.
Let me give you an example. The morning after the debate I went to a Time and Fortune magazines to see the polls on who won the debate (according to the poll, I have my own opinion). At the bottom of each page was a poll on who won the debate. I cast my vote. The problem with this is even this poll has introduced bias without trying to. It seems completely random, right? It is just polling whoever went to the website.
Unfortunately, even with the seemingly randomness of the poll, it could be an inaccurate representation of the voting population. First, do we even know if people taking this poll are in the voting population? I could have been five and took this poll. Second, there could be something about people who visit Time and Fortune websites. There could a certain demographic that visits these websites. I’m sure there is. (Now, even with all that said, it could turn out to be a pretty accurate picture of America’s registered voters, we just don’t know.)
Here is another example of a poorly conducted poll. CNN’s article, “Post-debate poll: Hillary Clinton takes round one,” they even say, “Although the survey suggested debate watchers were more apt to describe themselves as Democrats than the overall pool of voters, even independents who watched deemed Clinton the winner, 54% vs. 33% who thought Trump did the best job in the debate.” Read this closely, “Although the survey suggested debate watchers were more apt to describe them as Democrats …” Seriously? This is a great example of a biased poll. So how would you make your poll unbiased?
Ideally you would want to conduct a poll by making your sample completely random. This could be done by using a random number generator to call registered voters, making sure to call a similar percent of Republicans, Democrats, and Independents that are represented in the U.S. voter population. You also have to factor how many of those voters called will actually vote. There is probably a lot of other things I’m not even thinking of right now. It is a complicated process.
Unfortunately you don’t get a lot of insight into how a poll is conducted by a certain news station or polling group, usually you just get the results. Does that mean all polls are bad? Not necessarily, but you must analyze a lot of polls from a lot of different sources to gain some real insight into what is going on. Usually that will give you a better idea.
That being said, polls do have some margin or error and most polls unfortunately probably have some bias introduced, whether intentionally or unintentionally. Ultimately you want to make sure you vote in your beliefs. If everyone does that we will get an accurate view of the population, so long as the election is fair … (but that’s a whole other topic)