Remember Gallup pushed people to make a decision, that is why the number of un-decided's is way too low around 1-2 %.

I say this being an Obama supporter, because  we need to look at a flawed poll and be realistic about these numbers.

We all would be pointing out these flaws if the republicans were leading in head to head over our democrats.

He also has the lowest ID score. I'm thinking McCain will be the easiest to beat. Bad declining numbers plus he's too old to have the energy needed to campaign full-time for a year and a half. We should hail the new Dole, as old as the old Dole.

I was about to say the same thing - Edwards is the strongest Democrat who is currently in the race.

You can't make that case when the margins are within one point, either way.   That is just silly argumentation.   So, I raise you a one-point McCain lead over a your one-point Giuliani lead?   With MoE considerations a "lead" on anyone does not exist here, all 3 are doing equally well.  

BTW, what do you base this statement on:

"In most polls trying to predict the Dem primaries, Clinton is the clear winner, probably in spite of being the least progressive candidate rather than because of it."

Voting records show Clinton to be a very progressive candidate.  

Her progressive score of 92.25% is only a smidgen behind Obama's 92.73%  (as per http://www.progressivepunch.org ) and they are ranked 14th and 15th on the progressive scale.  We don't have a progressive voting record for Edward, as he is not active, but if you take the votes we DO have to look back on, his progressive score would be well down the list, definitely not in the 90s.  

You're trying too hard. You're fiddling with small marges and making it sound as if they are big ones, etc. If just you tone it down a bit the result is almost the same, without over playing the numbers.

Diferences in % per candidate match-up.
C- E- O
3- 6- 2   MC A very slight lead for Edwards
4- 5- 5   G  Within the margin of error/simular result.
7-31-29 R  Clinton doing comparibly very bad against the more unknown Romney.

This seems to indicate that Clinton performs worse when the candidate is more dependend on brand loyalty then on name recognition. This shows a significantly more limited crossover appeal for Clinton compared to other democratic candidates at the moment.

With this I've said the same as you, but I've used for stronger numbers.

Now we can discuss how much that is going to be an influence during election day as the candidates will be well known and how much this is a weakness that can be adressed. On which you and I will doubtless differ a lot but it'll be a much more open discussion. It does't always have to be made the worst version possible in order to show a weakness in a candidate.

that's extremely naive.  don't you understand margin of error?  while interesting -- and i love to see these polls -- these match-ups tell us nothing about each of these candidate's strengths and weaknesses against potential republicans.  they are media fodder.

we can argue about who is stronger, but i don't think it illuminates anything to say that candidate x is better than candidate z when both are in the same range and well within the margin of error.  i happen to agree with your conclusion, but your reasoning is unsound...

I cede georgep's point about voting records. By some measures Hillary has voted more liberally than Edwards, who may be the progressive favorite. However, there seems to be considerable controversy about how progressive or centrist Hillary is. I think I am not alone in being influenced by her husband's record which, although pretty good overall, in spite of an exceedingly hostile congress, also included pushing some regressive legislation such as NAFTA and welfare reform. In many of Hillary's public statements over the past several years she has served as a "ventriloquist for the patriarchy", as Jane Fonda says, in spite of conservatives' efforts to paint her as ultra-left wing femi-nazi. For instance, her statement in the last debate that we are safer now than we were a few years ago, in spite of all reputable research being to the contrary. Her courting of such conservative icons as Newt doesn't help either. Maybe that is just positioning, but that doesn't make me feel that good about her either.

Many supporters claim that her positions are consistent with what she has said and done all along, which would make her fairly far to the right of center, or what they call "centrist". I was hoping for more of a real progressive; I think that is what the country is crying for. But she is certainly infinitely better than any one of the possible or impossible Republican candidates, and may be our best hope for this cycle.

In any case this is a much larger topic, and in any case is off-topic.

Yes, I understand a bit about margin of error. I have had several courses in statistics.

On the margin of error for this poll, the sample size is about 1000, putting the margin of error at about 3-4%, according to Gallup. For spreads, the margin of error is about twice that, or 7%. So none of the two-ways is significant except against Romney. There the difference between Clinton and the others is almost certainly significant (but of what? Ernst's explanation seems as good as any).

Incidentally, there is no single standard of significance--it all depends on how sure you want to be. And it is possible for individual statistics, that are not statistically significant, to add up to a significant overall result. So it is possible that it be significant that the Dems are ahead on each one of these match-ups against McCain and Giuliani even though none is significant on its own. And it could also be that the fact Hillary is behind the other candidates is also significant for the same reason (although perhaps less likely, since it is a more complex dependency).

That said, I agree with bored now that at best this doesn't tell us much. Even if a calculation said that one of the above passed some statistical test on paper, that doesn't mean that the questions are sufficiently well designed, that the sampling is good enough, or that it isn't highly influenced by some news event which people may have forgotten by next week.

i won't be voting for hillary regardless because i don't believe that political dynasties are good for our political system...