Friday polls: Diverging margins in Mississippi, and the state of do-over primaries

The Democratic candidates are preparing for two more contests in the next few days -- Wyoming and Mississippi -- on which there has been no public polling. Until today, that is. Two polls were released of the Mississippi primary, which is expected to go heavily for Obama. The share of the African-American population in this primary could be higher than in any other contest's -- but these polls suggest that the higher a state's black population the more whites vote for Clinton. We had witnessed this racial polarization in South Carolina and Georgia, and now here it is in Mississippi:

• ARG shows Obama leading 58% to 34%, a very large margin that hides big disparities: Clinton wins white voters 61% to 22%, while Obama carries African American voters 87% to 11%.


• Insider Advantage -- which has had a very good track record in most states it has polled, and which picked up Clinton's Texas rise first -- shows Obama up only 46% to 40%. Now, IA itself acknowledges that Obama's margin is likely higher because a full 20% of black voters tested in this poll are undecided, putting Obama at 66% -- he will almost certainly be much higher on Tuesday. But IA also shows Clinton winning an astonishing 71.8% of the white vote.


• Another very strange result in which both polls are consistent is that Clinton does better among men than among women, which suggests that Mississippi white men are not willing to back a black candidate...

If this level of racial polarization holds on Tuesday, it would provoke some very strange delegate distribution scenarios, as Clinton and Obama would both be close to being shut out in different parts of the state depending on the racial make-up of different districts.

Meanwhile, Rasmussen came out with two very interesting polls today from Florida and Michigan, testing the two candidates in possible do-overs. The results are very different:

• In Michigan, Clinton and Obama tie at 41%. In January, Clinton had gotten 55% of the vote without any opposition.
• In Florida, Clinton gets 55% to Obama's 39%, which is a healthy lead. On January 29th, Clinton had beaten Obama 50% to 30%.

It goes without saying that Clinton is likely a few points below Obama in a Michigan caucus if she can only tie him in a primary, so that poll is not a good sign for the Clinton campaign. One of their most important arguments is Obama's inability to win the largest states, and an Obama win in Michigan -- particularly after Clinton broke 50% on January 15th -- would make it much more difficult for Clinton to put forward her big state edge.

Another interesting poll this morning -- also from Rasmussen -- tests the South Dakota presidential race:

• McCain beats Clinton 50% to 38%, and beat Obama 48% to 38%.
• Yesterday's SUSA poll of South Dakota had McCain up Obama by 4% and up Clinton by 12% as well.

A very Republican state that Bush won by overwhelming margins, South Dakota is much closer than anyone would expect in both of these polls.While it is very unlikely that either Democrat will end up within single-digits in South Dakota (and if they have a chance of winning here it probably means that they are on their way to routing McCain). But it does coincidence with a lot of evidence that the Mountain states -- usually among the most reliable for Republicans -- are flirting with Democrats this time around. SUSA showed Obama with a narrow lead in North Dakota and taking some electoral votes in Nebraska.

Finally, we have a few polls from congressional races:

• Rasmussen accompanies its presidential poll with a Senate survey (it's actually the other way around), and it confirms that Democratic Senator Johnson has very little to fear next year. He crushes both his (declared and potential) Republican opponents by significant margins and gets 62% and 63%.


• Meanwhile, Michigan pollster EPIC releases a survey from MI-07, where Democrats are confident they can get freshman Rep. Walberg in trouble. The poll shows Walberg leading Democrat Schauer 51% to 40%, but after a biography of both candidates is read the Democratic candidate edges out Walberg 49% to 48%.

Polls with biographies are obviously very shady, since no candidate gets to define himself in a vacuum, and no House candidate can possibly introduce himself to 100% of voters (not even close). The poll does show a baseline vulnerability for Walberg, but he does get above 50% in the first match-up, and that should reassure him a bit. But given the financial advantage the DCCC has over its counterpart this cycle, they are sure to test incumbents like Walberg and unless the RNCC can step in to help he is likely to see his numbers drop in the coming months.