6 Answers | Add Yours
I think the accuracy of polling depends on the questions. It is very easy to create a poll that will slant the results toward whatever direction you want. I also don't think exit polls are good, because they will discourage people from voting. If people think the election is decided before they vote, they will not vote. It's a self-fulfilling prophecy.
Whenever there is political polling for state politics, our phone rings and rings. But, in twenty + years of living at the same residence, we have received only a few calls from national pollsters. This fact makes us think that some states simply are not polled much because they have already been designated as "conservative" or "liberal."
And, judging from the "results" of some polls, polling seems almost idiotic. Do these people who are polled not watch the news or talk to anyone who is knowledgeable?
Honestly, after living in Iowa--where it's virtually all politics all the time--I'm pretty burnt out toward the end of a political season. I don't think that's an uncommon feeling, and the ads are getting uglier and and more dramatic. Across the board, people are tired of being bothered by "robo-calls" and pollsters. That cynicism may or may not carry over into their answers to polling questions. I wouldn't be surprised if it did, anyway, and that's one more variable in the equation.
If you take the polling average of several organizations, throw out the "outliers" or wildly varying results, and you usually get a pretty good indicator of a range of possible outcomes. Some are better than others. Gallup is usually considered pretty good. Nate Silver has been pretty accurate in the past two election cycles. Rasmussen has not. Zogby has not.
The trouble with this particular election is that the pollsters are having a difficult time telling us what a "likely voter" will be, as the electorate they are sampling keeps changing on them from election to election. Big variables include whether minorities or young voters will show up at the polls in the numbers they did in 2008. Differing interpretations on that is making for some variance between poll results.
In general, it does, but this is coming to be less true right now because of one particular problem. That is the problem of people who have only cell phones (no land lines).
The reason that this is important is because polling is typically done by calling land lines. When people have cell phones only, they cannot be reached by conventional polls. It is very expensive to do polls that include cell phone only people.
So what? The so what here is that cell phone only people tend to be different -- mostly younger and more educated -- than others. When you leave them out, you skew your sample and it becomes harder to get good results.
Please follow the link for a longer discussion of the problems that cell phone only households are posing for pollsters.
I think that political polling, for the most part, does provide fairly accurate results. In all honesty, there can be no really accurate read with the exception of counting the ballots. I think of the Florida Presidential Contest in 2000. Exit polling was seen as highly reliable until that election when news organizations had to put the state in different candidate columns. It was at this moment that all polling was cast in a light of some level of skepticism. I think that political polling is an excellent indicator of how the body politic might feel about a particular candidate or issue and can provide a good guideline about where strategy can be formed and direction formulated. However, I think that there can be no real substitute for the tally of the final results. The danger that consultants fall into is when they believe polling to be a substitute for counting the ballots.
We’ve answered 319,632 questions. We can answer yours, too.Ask a question