OCCOM writes
Quote:Parados, you are wrong. Opposition, who I respect tremendously have accepted that data without hesitation. Unless you can provide opposing facts your exclamation is only a "sez me". Since I don't know you, and have no reason to trust in your expertise, that carries no weight at all. Oh, and false accusations of ignorance and stupidity don't reflect well on the accuser.
OK. lets discuss how data is to be interpreted in a poll. This is hardly a "sez me". Any class on statistics will tell you this or you can find the same information on Zogby or Gallup polling sites. (It IS ignorance if you do not understand the scientific method or scientific polling.)
The first rule of any statistical analysis is that your polling sample MUST be random. Without a true random sample the data you acquire will be skewed drastically. So, when examining a poll the first thing you always have to look at is how the random sample was acquired. A true random sample means that everyone in the set you are sampling must have an equal opportunity to be in the sampling. Without that equal chance then you can't extrapolate the data to the entire set.
In the case of the lovely charts that Lusatain posted a link to there is NO REFERENCE to the sample, to the methods of achieving a random chance for sampling nor any other mention of the methods used. Without a single mention of the methods then the validity of the charts can not be SCIENTIFICALLY judged. The idea behind SCIENTIFIC polling is that the methods should be known so that anyone can repeat the method. Since the methods are not shown then the charts are meaningless under the scientific method since they can't be duplicated or checked.
Based on what Lusatain posted earlier we can ASSUME that these charts are based on the email survey conducted by "The Military Times". With a little research it is easy to find the methods used to get these figures. The Military Times sent out emails to 31,000 of its subscribers and got back about 4000 responses. These 4000 responses make up the information they published. Since we know for a fact that not all military members are subscribers to "The Military Times' that means we can NOT extrapolate any response to be indicative of the overall military since every person in the military did not have an equal chance to be included in the sample. This is a classic failing in the use of statistics, trying to extrapolate findings to a set that was NOT used to get your random sample. For instance If I polled 30 of my friends and relatives in the military and 67% of them said they were voting for Kerry I could not apply that to the entire military since my sample was NOT randomly of the entire military. The only conclusion I could make is that my friends and relatives in the military are more for Kerry than for Bush.
The next failure would be to claim it is even indicative of the subscribers of "The Military Times". Again, you have a response structure that was not random. If one side is more motivated to answer in a survey it skews the survey. The key to an accurate sampling for a poll is that you must make every attempt to get an answer out of everyone in your sample. If your "sample" is 31,000 people and only 4,000 respond then your margin of error becomes quite high based on the fact that you only have responses from 12% of your sample.
Scientific polling has existed for over 50 years with some rather dramatic failures in its history. "Dewey defeats Truman" being a prime example and one that changed dramatically the way pollsters polled. The idea behind getting accurate polling is to reduce the factors that can contribute to a bad sample or bad responses. In order to judge the accuracy of the poll you have to know how the random sample was achieved and what questions were asked. For instance would you trust the answers in a poll that asks; "Are you voting for the patriot John Kerry or President Bush who is a traitor and responsible for over 1000 American deaths?" I know I wouldn't.
Now lets examine what "The Military Times" actually says about its poll.
Who you chose for president and why Army Times Oct 11, 2004
Quotes from the Times story on the poll:
Quote:..the results of the Military Times 2004 Election Survey are not representative of the opinions of the military as a whole..
Funny, "The Army Times" agrees with me on this one. You can't use the survey to represent the entire military or even to calculate the views of the readers. Again, I state categorically, Lusatain is misrepresenting this poll by claiming it shows what the military as a whole thinks when he started this thread. Anyone that continues to think it does at this point is no longer ignorant but just plain stupid.