Originally Posted by YardleyLabs
In response to #1..you go from decades of analysis to centuries of "probable theroies"
you know I have this friend, he says the wieredest stuff...like.."you know Im probabbly 100% right that this might work". or when we Fish..I mean seriously when we FISH..."I guarentee, there is probably fish over there" So tell me the difference between these statements, probable theroies, and faith. While I do agree that alot of it depends on how a question is phrased, it almost proves my point that when you phrase a question such as pro-life or not a republican is more likely to say "no comment" where my experience is a democrat is gonna say "pro-abortion all the way".
Im just saying.
and in response to #2
hey, if that how your twisted mind read what I was saying who am I to stop you...;-);-) the fact still remains the same while extreme conservatives may out number democrats 2:1...for every one democrat that says he or she is such, there is usually only .5 republicans that will say they are....
your right, just like I said, young republicans are becoming less bashful when it comes to pronouncing there party ties....
but your analogy of this board doesnt hold water....that is because the majority of us that speak out on this board do so becuause we are in the company of others with like minds...
Get 5 republicans together and they all admit to it amoungst themselves and they talk all day long....they are comfortable about it..birds of a feather and all that jazz...
But I would and do challenge that if you were to sit outside your local store and ask people that entered..if they were Republican or Democrat (and nothing more) the number of people saying they are Democrat would be higher the number of republican would be low and the number of people saying "no comment" would be higher again with the majority of those people actually be republican....
out in public republicans tend to be GUARDED, or conservative (go figure) in announcing there political ties...
To me this is a better example of how a poll and how it gains its numbers should look...and even then I dont (and this is just my opinion from what I have gathered from my own conversations) buy it as totally accurate...but at least the pool is extensively larger..
Rasmussen Reports tracks this information based upon telephone interviews with approximately 15,000 adults per month and has been doing so since November 2002. The margin of error for the full sample is less than one percentage point, with a 95% level of confidence.
15 x 12 =180,000 people a year
x 7 years 9 months = 1,395,000 people polled....just a bit better dont you think....
and this woudl show that DEMOCRAT's still hold a larger identified party affiliation.
I didn't say "probable theories", but "theories of probability". It's an area of math that has been in development since the 18th century and addresses the specifics of how properties of large populations may be estimated with a high degree of precision using samples that are relatively small. It also addresses factors that can make such estimates unreliable. The fact that your friend's comment is ignorant and that you have no basis for knowing how either a republican or democrat would respond unless you have conducted experiments (using appropriate samples and solid analyze of results) has nothing to do with the accuracy of polls. The fact that your gut tells you how a republican or democrat would respond is an act of faith since, by your own statements, you would need to interview 50000 or so people to verify your assumption.
Originally Posted by Nor_Cal_Angler
All the major pollsters, including Rasmussen and Gallup among others, interview thousands of people each week. Rasmussen's work requires a little more "interpretation" than most because they use computer calls to obtain their data. This result in a higher non-response rate and a different pattern of responses in some circumstances (you can find this information on the Rasmussen web site). As with all legitimate pollsters, part of the process is analyzing biases in the responses received so that these may be taken into consideration is interpreting the results. Rasmussen does not use larger samples than other pollsters. The fact is that polls tend to fall into a limited number of sample size ranges: 300, 1100, 3000, etc. In each case, a dramatic increase in sample size is needed to hit the next level of precision.
The primary reason for expanding sample size is to obtain results differentiated by sub-group. That is, if I want to be able to draw conclusions about differences in opinion among democrats (or republicans) of different age groups, I need to have a sample big enough so that the subset of respondents that are democrats is still large enough to be statistically significant. Whether you believe in sample based studies or not, your life is affected every day by the results of such studies. Thus includes the routing protocols used for transmitting telephone conversations, the studies done to validate medications, the criteria used to select tax returns for audit, etc., etc.
As with many things, and as I said originally, this is really a matter of faith vs. science. I'll stick with science. That doesn't mean that I believe anyone should simply buy into poll results. However, you should know enough about the process to have a basis for differentiating between results that are likely to have some validity and charlatans manipulating the process to sell snake oil. The fact that you agree with the results should make you more cautious, not more accepting. We are all more likely to accept BS when we like the results.
EDIT: By the way, each sample is unique. You cannot add the results together unless you ask the same question in the same way during the same time period each time. However, when you repeat the same polling process every week for years, as all the major pollsters do for key questions, you are better able to recognize the skewed samples when they happen.
Originally Posted by YardleyLabs
The first bolden protion of the quote...
I do know enough about the process to have a knowledge of how it works and thus the reason for my calling it BS....
The second bolden and red lettered protion of the qutoe...
The fact (as you put it) is that I DO NOT agree with the study and I DO NOT agree that American is more conservative (based on the results of the poll of 1100 people)
I feel that america is, has been, and or becomming WAY TO LIBERAL...
I guess you could make an arguement that I contradict myself because I am saying on the one hand that conservatives are more Hush Hush, thus resulting that more people are holding back and actually feel conservative to the core. But reality is and what I should have said more clearly is that what ever that poll says about Americans being more conservative, what we are actually seeing on a daily basis is that
AMERICA UNDOUBTABLY IS BECOMMING MORE AND MORE LIBERAL, with there view of society and how it should be run, formed, and lived....
edit: about your edit....I do not know why your explaining that you cant add statistical "evidence" (oxymoron) unless your saying that I attemped to do that, maybe you were just trying to inform me of that. But you should know, I know that...lol
I did it based on your earlier comments about the total numbers of people questioned by Rasmussen over the course of a year and more. The cumulative total of 1.395 million (your number) does not affect the validity of their survey results in their daily surveys.
Originally Posted by Nor_Cal_Angler
very true as I understand it in America....1/3 is and always will be the same as 33000/100000...but I contend that if you wanted to convince a large audience a sample of 100000 is better than a samle of 3....
As I see it, you suggest that a sample taken at 0800 will be different if the same sample is taken at 2000....or if I ask the same question phrased differently...or if I use emphisis and puncuation differently on key words..etc,etc,etc....your words not mine...
"The problem with polls is that everything depends on doing it correctly. How you select people, how you contact them, how you handle non-respondents, how you word the question"
So using your own words again...it would seem that the Rasmussen poll would be a better indicator...
"All the major pollsters, including Rasmussen and Gallup among others, interview thousands of people each week. Rasmussen's work requires a little more "interpretation" than most because they use computer calls to obtain their data."
furthermore, again just making sure I understand you correctly...
"The fact that your friend's comment is ignorant and that you have no basis for knowing how either a republican or democrat would respond unless you have conducted experiments (using appropriate samples and solid analyze of results)"
If I were to sit outside my local store and ask 1100 people (total sampled in the orgional article posted) and analized it properly maybe using say a professor of mathimatics majoring in probability and statistics I could then extrapolate that information to fit the bill of Americans and there political ties...
OR would I have to as you say.....
"The fact is that it is more accurate to predict opinion on a discrete issue for the entire US population using a carefully selected sample of 1100 people"
Carefully select my sample population.....
ahhh, the silver lining.
Actually, sitting outside your local store would provide a non-random sample of people shopping in your store who agreed to talk with you. It would not be projectable to any broader population at all. The reason samples taken over the course of a year cannot be added together is because you are measuring a moving target: public opinion. If you were measuring hair color, a sample taken over the course of a year would not be bad.
All in all, this is like talking to those who condemn FF and e-collars without ever having done or used either or having seen the results. Both tools can be used inappropriately. However, those who simply oppose them out of ignorance are only showing what they don't know.
Originally Posted by YardleyLabs
Well put. While there are certainly entire fields of mathematics and statistics involved in polling, it is very rare that conditions will allow a scientifically sound, mathematically solid study or poll to be designed and executed. (from my experience in medicine....we're always hitting a moving, complex target with uncontrollable variables) The *good* statisticians temper the math and science with intuition, judgement, and a little luck. That's why polls are never perfectly accurate, but the good ones consistently come close. And that's why they get paid the big bucks from media and interested parties.
Like much of this world, the science must be polished with the art.
Enjoy the weekend, all.....I'm got tagged with the w/e duty this time.:-x
The judgment criticism is completely accurate. One of the classic pieces of stupidity from the earlyyears of political polling was a Lou Harris poll done for a gubernatorial candidate. What it showed was the following a single speech, the candidate's support among Irish Catholics had fallen 50%. The campaign was in panic as they tried to figure out what they could do to win that support back. What nobody looked at was that their sample included less than 10 Irish Catholics. By the next week the numbers had swung back again. Opinion polling is actually very difficult to do well and is extraordinarilly expensive. In a campaign it is one of the biggest expenses behind advertising. Sometimes these polls are deliberately biased to create evidence of support when a candidate is having trouble with fund raising. However, strong candidates work hard to avoid tha form of bias. What they need are accurate indicators that may help them win in the voting booth.
Originally Posted by dnf777
The same is true of companies trying to sell their products to consumers.
Politicians and marketers have paid for most of the improvements in polling techniques since the 50's when opinion polling first became a significant force. To meet their needs, a lot of progress has been made and the reliability of polls has improved dramatically. The biggest problems today come not from weaknesses in polling methods, but from deliberate efforts to distort the results to push specific positions (In fact, these are called "push" polls.). This may be seen all over. From examples like the NRA "polls" that I mentioned before (liberal groups do them also, but most on this forum never see those) to near fraudulent aboses of statistics such as those where auto insurers talk about how much subscribers saved by swapping from one company to the other (did you really think that people who received higher quotes would switch anyway?).
My only weekend duty is a litter of eight four day old pups. That keeps me on a very short leash.
I've always likened opinion polls to the Bible. Two exactly opposite points of view can be equally "proven" and supported from the same sources! Hence the endless arguments!