Great News for the GE: Long term trends favour Dems :)

No One Wants To Be A Republican
by DemFromCT

<snip>Hey, it's not just the candidates that want nothing to do with the tarnished, corrupt, competence free Larry Craig Republican party of the 21st century, it's the people. Pew has a fascinating trend poll that bodes ill for the shrinking number of Republicans in November<snip>

http://demfromct.dailykos.com/

Fewer Voters Identify as Republicans
Democrats Now Have the Advantage in "Swing" States
March 20, 2008

http://pewresearch.org/pubs/773/fewer-voters-identify-as-republicans

The balance of party identification in the American electorate now favors the Democratic Party by a decidedly larger margin than in either of the two previous presidential election cycles.

In 5,566 interviews with registered voters conducted by the Pew Research Center for the People & the Press during the first two months of 2008, 36% identify themselves as Democrats, and just 27% as Republicans



The Electoral Landscape

The decline in the number of self-identified Republicans is evident in all parts of the country, but is perhaps most significant in the politically important "swing" states that were closely contested in the 2004 presidential election (see "How the States are Analyzed" below).

Four years ago there were about as many Democrats (35%) as Republicans (33%) in the 12 states where the voting was closest in 2004, and the balance was similar in the 2000 election cycle. But so far in 2008, Democrats hold a substantial 38% to 27% identification advantage in these states.



Inside Key Swing States

The growing Democratic identification advantage in swing states generally holds true when several of the largest swing states are analyzed individually. In Ohio, 37% of voters identify with the Democratic Party, while just 25% identify with the Republican Party, based on surveys conducted in 2007 and 2008.1

This is a 10-point drop in Republican ID since 2004, and a four-point gain for Democrats. The pattern is similar in Pennsylvania and Michigan, where a rough balance in the number of Democrats and Republicans in the last two election cycles has shifted to a substantial Democratic advantage.



Whether the Republican losses in these swing states will translate into an electoral advantage for Democrats remains uncertain. There is not a one-to-one correspondence between the balance of party identification in a given state and the electoral outcome in presidential elections.

Nowhere is this gap more evident than in many of the states Bush won handily in 2000 and 2004. Bush carried North Carolina, Georgia and Virginia by eight points or more in 2004, even though the balance of party identification in those states was about even -- and actually favored the Democrats by four points in the case of North Carolina.



Inside Key "Blue" States



Race and Party Identification





GObama!
Go Dean's 50 State Strategy!
Go Grassroots!

Only the braindamaged will stay RED

BLUE is the new color....

I read it on DU, so it must be true.

Nice work!

Thanks for the fascinating data!

When you look at a statistical analysis of polls there are four factors to look at: 1) Nominal, 2) Ordinal, 3) Interval, 4) Ratio.

We know who did the polling. However we need to know how they arrived at the study clusters.

There are two techniques that can be used to derive data: 1) Stratified, 2) Convenience.
Under these two techniques are things like was the poll random, how large was the study groups, how was the questions asked, how was the data collected: phone (land line vs cell), door-to-door, the percentage of ethnic groups, household income, etc.

Just by looking at the size of the study group, 5,500 people in itself would not be representative of the ~150,000,000 voters.

What this means is all the polls that have been posted are irrelevant. Don't ask me any questions because I can't tell you what this all means other then I was relating what my partner told me.

Pew's guesstimates have gotta be in the ballpark.

Did your partner really say that a sample of 5,500 people couldn't be representative of 150,000,000 people? Certainly is doesn't have to be, a funky sample is a funky sample, regardless of size. The issues your partner raises are good ones about which to be concerned. OTOH, Pew is a pretty experienced polling outfit with some pretty good statisticians on board, I'd be suprised if they made any egregious errors. I'd want to get some specific problems raised before I'd make a blanket statement that What this means is all the polls that have been posted are irrelevant.

He has taken statistics classes in the past and he seems to think polls are irrelevant. He went on to say even the people whom are professional statisticians will admit these polls are irrelevant. He also said his statistics professors made the same statement about polls. That's all I can say about this issue because I really don't know enough to elaborate.

based on any sample size :-).

I use statistics as a working tool and make predictions based on it that help guide actions. Since I'm going to have to watch the shipwreck if I send people off in the wrong direction I'm pretty conservative in that area. Pew's analysis has information in it - i.e. - there is enough signal to overcome the noise and still carry meaning. It's certaily NOT absolute, but then no statistic ever is. It does tell me something about voter preference, though, and probably shouldn't be dismissed out of hand.