there has been alot of talk about the statistics. The more I study statistics, the more I've seen that you can really make them say whatever you want... work with the right groups, word your questions appropriately, look at the right data or with the right angle, you get dramatically different results... It bothers me when I see certain polls online. For instance, you see an online poll asking something like "Do you approve of Bush's policy on so and so?". They request a simple yes or no. No varience. So, you answer No. You don't approve. Do you think he is too far right? Too far left? There is no difference. There are alot of polls on CNN that do this (I tend to frequent that web site). I'd rather a poll have more options.
Consider another poll... you go to DC and ask what they think of some kind of policy (lets say gun control). Depending on the policy or question, you get dramatically differing results than if you asked in Texas or Kansas.
So, point is... lets see where these statistics are coming from. Is it good science just because a few statisticians have problems? are there more reasonable explanations for some of the anamolies? Would you call it good statistics if you get your results from a narrow group? *scratches head* you -might-, and of course, the results will not be presented in such a fashion... so, in order to make decisions regarding the validity, I'd like to see sample groups.
For instance, this was mentioned before, and laughed off, but it actually has a degree of merit. Alot of people that are on welfare are likely not working, and would possibly vote earlier. Alot of retired people would vote in the morning, and the fear tactics of Social Security might possibly push them democratic. Is it that impossible that Republicans vote late? There are alot of polls going on, informal of course, but exit polls that give an idea of how the election is going. they are not the FINAL vote, but a predictor. If I saw my candidate was winning by 20 points, would I be quite as likely to go vote? If I saw it was a close vote, would I be as likely to vote? What -would- be compelling is if this never happened before. I've not read all the web pages presented (not had time, trying to graduate soon!), so if I missed this, please forgive and point out.
Anyways, I've always had a beef with certain statistics... this discussion is not necessarily relevant, just a beef I have with the presented "facts" I often see...
MrH
Consider another poll... you go to DC and ask what they think of some kind of policy (lets say gun control). Depending on the policy or question, you get dramatically differing results than if you asked in Texas or Kansas.
So, point is... lets see where these statistics are coming from. Is it good science just because a few statisticians have problems? are there more reasonable explanations for some of the anamolies? Would you call it good statistics if you get your results from a narrow group? *scratches head* you -might-, and of course, the results will not be presented in such a fashion... so, in order to make decisions regarding the validity, I'd like to see sample groups.
For instance, this was mentioned before, and laughed off, but it actually has a degree of merit. Alot of people that are on welfare are likely not working, and would possibly vote earlier. Alot of retired people would vote in the morning, and the fear tactics of Social Security might possibly push them democratic. Is it that impossible that Republicans vote late? There are alot of polls going on, informal of course, but exit polls that give an idea of how the election is going. they are not the FINAL vote, but a predictor. If I saw my candidate was winning by 20 points, would I be quite as likely to go vote? If I saw it was a close vote, would I be as likely to vote? What -would- be compelling is if this never happened before. I've not read all the web pages presented (not had time, trying to graduate soon!), so if I missed this, please forgive and point out.
Anyways, I've always had a beef with certain statistics... this discussion is not necessarily relevant, just a beef I have with the presented "facts" I often see...
MrH