Tilted Forum Project Discussion Community - View Single Post

quicksteal · 11-01-2004, 03:54 PM

Polling that is performed is "reliable" generally within 3-4 percentage points. This means that a statistician is 95% certain that the results are accurate within the given range. If I was to say that in a poll taken this weekend, 85% of Halloween party-goers like wearing costumes, and this poll has a 4 % margin of error, then it is 95% likely that between 81-89% of Halloween party-goers like wearing costumes. Of course, it is most likely to be 85%, and the probabilities drop off from there.

Statisticians do a lot of "fuzzy math" work to try to make the polls as accurate as possible. For example, they try to poll a certain number of Democrats, Republicans, and Independents. This way, they won't poll too many members of one party, who are likely to choose their candidate, and therefore sway the poll. There was an argument that in Ohio, the Gallup organization was polling too high a percentage of Republicans (35%) because they were using a number from the 2000 election, and the Democrats have reported that they have registered more voters than the Republicans in Ohio since then.

To me, it's faily foolish to judge the opinions of everyone based on one poll. They seem to do better when averaged together, and when they are used to watch trends (post-convention or post-debate bounces). This is why I like www.realclearpolitics.com, since they do a good job of averaging polls and showing such trends. Be warned--their commentary is Republican-leaning, so they might sway some of you folks!

To answer your thread topic, I agree to an extent that people should stop polling before the election, and since polls always open on Tuesdays in America, I think it would be a good idea to at least discourage pollsters from releasing their results the week of the election. I feel as though it creates more confusion, especially when the race is this close. Right now, news organizations should be trying to convey the two platforms of the party, not bickering over who's ahead in what state, and the only way to do that is to take away their bait.

11-01-2004, 03:54 PM	#8 (permalink)
quicksteal Crazy Location: Auburn, AL	Polling that is performed is "reliable" generally within 3-4 percentage points. This means that a statistician is 95% certain that the results are accurate within the given range. If I was to say that in a poll taken this weekend, 85% of Halloween party-goers like wearing costumes, and this poll has a 4 % margin of error, then it is 95% likely that between 81-89% of Halloween party-goers like wearing costumes. Of course, it is most likely to be 85%, and the probabilities drop off from there. Statisticians do a lot of "fuzzy math" work to try to make the polls as accurate as possible. For example, they try to poll a certain number of Democrats, Republicans, and Independents. This way, they won't poll too many members of one party, who are likely to choose their candidate, and therefore sway the poll. There was an argument that in Ohio, the Gallup organization was polling too high a percentage of Republicans (35%) because they were using a number from the 2000 election, and the Democrats have reported that they have registered more voters than the Republicans in Ohio since then. To me, it's faily foolish to judge the opinions of everyone based on one poll. They seem to do better when averaged together, and when they are used to watch trends (post-convention or post-debate bounces). This is why I like www.realclearpolitics.com, since they do a good job of averaging polls and showing such trends. Be warned--their commentary is Republican-leaning, so they might sway some of you folks! To answer your thread topic, I agree to an extent that people should stop polling before the election, and since polls always open on Tuesdays in America, I think it would be a good idea to at least discourage pollsters from releasing their results the week of the election. I feel as though it creates more confusion, especially when the race is this close. Right now, news organizations should be trying to convey the two platforms of the party, not bickering over who's ahead in what state, and the only way to do that is to take away their bait.