CBS/NY Times Lies again: New "poll" uses same tainted sample as old poll (Vanity).

Those numbers (39%D,34%R,27%I) sound like the flawed VNS exit poll numbers from the 2000 election. I think that the 2002 numbers were a lot better in accuracy, but the break-downs must be reevaluated every year...

There are so many cross-tabs which can be generated by pollsters to support which way the electorate will, or did vote. Race, income, gender, age, views on domestic and foreign policy, religious affiliation, etc. All well and good for the statistical analysis that goes on at Universities and polling firms.

However, without question the individuals political party affiliation (Rep,Dem,Ind) is the best predictor of how their vote will be cast. Somewhat less useful is the Conservative, Moderate, Liberal self-identification. Now the problem with many polls is that they use identification break-downs on the general adult population in the United States, and extrapolate those numbers to the actual voting public. Most recent presidential elections have around 55% of the potential electorate actually vote. Here is a link to data put out by The Pew Research Center titled "The 2004 Political Landscape" which shows a pre-9/11 voter breakdown as 27%(R) and 33%(D) which means 40%(I). The post-9/11 voter landscape according to the Pew Research is 30%(R), 31%(D) and 39%(I). This is for the general public at large, not just voters.

The best case for Independents to show their numbers would be for the 1992 Presidential election which had Ross Perot get 18.91% of the popular vote. When there are claims that 40% of the American electorate are Independents, it always seems that when ever there is a viable third-party candidate, those numbers are never realized. Many who self-identify as 'Independents' when asked are actually very consistent in voting for Republican or Democrat candidates, but for ego reasons wish to avoid being categorized as a 'party regular'. The Gallup organization pushes folks to eliminate some of this 'reluctance' effect when they came up with their numbers. Historically the American voting electoral has been divided into 40%-40%-20% groups, no matter the party label since John Adams was the second President. We are now in a time where the true number of actual political Independents has shrunk so there is very little possibility that 37%+ of the voting electorate are Independents. Those that do not vote do not count.

Personally I don't believe any national poll which currently has the internals on the Republicans and Democrats separated by more than 2%. If the 'Independents' hit 35% to 40% the poll is also suspect. Just my point of view. I don't want polls to say what I want to hear, but their sample universe must be logical and consistent...

dvwjr

Good stuff from all your posts. There is so much information out there that you can't get in the newspapers, much less the networks. I know this might be a Herculean task, but wouldn't it be great if someone could normalize all poll results along an accepted (and realistic) turnout model, as you did for this Slimes poll. I bet that a consistent pattern would emerge that would be stripped of any propaganda effects desired by the various media outlets. We could use 40-40-20, as a baseline. Of course, I think only likely voter polls would be applicable to this.

Furthermore, because the Times used a previous sample, how does that skew results. One usually strives for randomness in polling, don't you. A previously polled person might be reluctant to change his response, because he doesn't want to appear wishy-washy.

Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.