Melbourne can get VERY hot. I have been here for 3 summers now, and each year we've had a few days that hit 42-44C. I am from New York/Boston originally. There is a saying in New England that if you don't like the weather, wait a minute. Well, this is so much more true for Melbourne than it is for New England. We have something called the "cold change" here. A shift in the wind can come and the temperature can drop 10-15 degrees C in 30 minutes. Its amazing. Melbourne is a fantastic city. I absolutely adore it and you are very lucky to be moving down here.
I hadn't realized that detail. But this of course would fit in nicely with my statement that the problem isn't the sample size but that its a non-random sample
I'm less convined Kerry is going to win, but the cell phone bias in polling is true...
Here is a link to one of many articles that talk about the cell phone problem here.
Now I can't find a link, but I'd swear I read an article this morning saying that Gallup (or some other big poller) just did an SMS survey and found that Kerry was way ahead with that group...
Nothing wrong with this. Pick up any introductory stats book and look up confidence intervals. Assuming that your 3000 voters are an acurate random sample (which is of course the big problem, not the fact that a sample size is only 3000) you can accurately predict outcomes with some probability. I assume that they are using a 95% confidence interval which would state (assuming a random sample) that there is a 95% chance that the true scores are within +/- 5% X% for Bush and Y% for Kerry. You can increase the certainty to 99% if you want, but you then as a trade-off you increase the size of +/- percetage...
I hate to nitpick, but the XT was 4.77MHz, or did you have something special? And did you have a hard drive? The XT came with one. The PC was floppy only by default (and also 4.77MHz)
Slashdot Mirror
Melbourne can get VERY hot. I have been here for 3 summers now, and each year we've had a few days that hit 42-44C. I am from New York/Boston originally. There is a saying in New England that if you don't like the weather, wait a minute. Well, this is so much more true for Melbourne than it is for New England. We have something called the "cold change" here. A shift in the wind can come and the temperature can drop 10-15 degrees C in 30 minutes. Its amazing. Melbourne is a fantastic city. I absolutely adore it and you are very lucky to be moving down here.
I hadn't realized that detail. But this of course would fit in nicely with my statement that the problem isn't the sample size but that its a non-random sample
Here is a link to one of many articles that talk about the cell phone problem here.
Now I can't find a link, but I'd swear I read an article this morning saying that Gallup (or some other big poller) just did an SMS survey and found that Kerry was way ahead with that group...
Nothing wrong with this. Pick up any introductory stats book and look up confidence intervals. Assuming that your 3000 voters are an acurate random sample (which is of course the big problem, not the fact that a sample size is only 3000) you can accurately predict outcomes with some probability. I assume that they are using a 95% confidence interval which would state (assuming a random sample) that there is a 95% chance that the true scores are within +/- 5% X% for Bush and Y% for Kerry. You can increase the certainty to 99% if you want, but you then as a trade-off you increase the size of +/- percetage...
I hate to nitpick, but the XT was 4.77MHz, or did you have something special? And did you have a hard drive? The XT came with one. The PC was floppy only by default (and also 4.77MHz)