Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Politics : Electoral College 2000 - Ahead of the Curve -- Ignore unavailable to you. Want to Upgrade?

To: Ilaine who wrote (4364)	12/3/2000 12:50:41 AM
From: TraderGreg	Read Replies (1) \| Respond to of 6710

<<would the rareness of the occurrence and the size of the sample tell you anything about causation?>> YES, depending upon what it is that you are observing and measuring. A coin flip is a very, very concentrated event in the sense that you can remove/control external error sources. Sample size is everything. IF you toss a coin 10 times and get 7 heads, you might not be surprised. But believe me, 70 heads out of 100 tosses is a much different thing. It is nearly 4 1/2 standard deviations away from expectation...on the order of 4 in a million. I'm not going to tell you that it can't happen. Just that if you conclude the coin is bogus because you got 70 heads out of 100, your error likelihood is .0000036. It is called Law of Large Numbers and contributed to a great extent by this guy named Tchebycheff(ChebbyCheff) and is essentially what we call the "Law of Averages". The reason that you can say there is a problem with the coin is that the investigator can closely monitor the experiment and can control it. In the case of the optic vs punch card issue, he has to know a lot before concluding that it is the system rather than the users that caused the difference. Now, if the optic and punch card systems were BOTH distributed WITHIN each precinct, then you would know that user characteristics were uniformly distributed since your data would show that similar people used both systems and thus remove those potential sources of errors. But since they were not, he would have had to get very accurate data on specific voter profiles (demographic analysis of the various counties) to do a paired type of comparison...data that I don't think is available because of privacy issues. He claimed he got that data in "general" terms and concluded the people weren't the cause...sounds iffy to me considering the limited time he has had to do all this. But who am I to criticize a Berkeley magna? TG