SurveyUSA's 50-State Poll Tells Only Half the Story

SurveyUSA just released 2 50-state polls for the 2 potential matchups this fall:

Theres good news here for all of the remaining contenders. Hillary is shown winning Florida and Ohio and thereby narrowly winning over McCain. Obama is seen stealing away Ohio and red states like North Dakota and Virginia (and even 2 of Nebraska's EV's). McCain, whose national numbers have slipped lately, still shows only behind by 1 pt in many swing states giving him the potential to flip just one and win the Presidency.

But something is amiss. SurveyUSA used "registered voters" to conduct the poll. Historically (the recent polling in Texas and Ohio for the DNC primary as a clear example) there is quite a difference between "registered" and "likely" voters as an indicator. Focusing on just a few states in the SUSA data, and averaging it against that of polling using Likely Voters, we come to slightly different electoral maps.

Polling in Florida amongst LIKELY voters in the last month (Rasmussen, MasonDixon, etc) gives McCain a lead over HRC there, flipping that map 289-249 and giving John McCain the win.

Data from Ohio amongst likely voters shows it far closer there in either potential matchup. Rasmussen's pre-primary poll there gave McCain a 1-3 pt edge over either Democratic candidate. Still, it now averages as a weak Obama gain.

West Virginia in the HRC-McCain matchup by SUSA averages out against Clinton. WV has been polling significantly in favor of the GOP candidate up until this one poll which gives HRC a 5-pt edge. Going with the voting trends from 2000, 2004, and general election polling there, this stays weakly Republican against her.

Wisconsin in the HRC-McCain matchup by SUSA, when averaged with Rasmussen's late February poll of likely voters, also goes McCain's way, and the lead Obama has over McCain there shrinks when the same Rasmussen poll is taken into consideration.

The North Dakota results showing an Obama victory is laughable. First, RasmussenReports recently conducted polls in South Dakota, with a population demographically and politically identical to it, and gave McCain a solid 10 pt lead there. SurveyUSA shows SD barely in McCains camp. Suffice to say ND isn't straying from the Republicans this year.

Nebraska has been polling safe GOP/McCain with likely voters and the trend will not shift despite the SUSA's robopolling of registered voters.

Lastly we have Virginia, in which SurveyUSA gives to Obama despite a narrow 49.xx% victory. Averaged with RasmussenReports' late February poll, VA goes to McCain by 2.5pts.

With all the new data taken into consideration, averaged with existing polls from Rasmussen, Mason-Dixon,et al., heres what we end up with:

Even when factoring in other polls, you still end up with an incredibly close race, perhaps the only thing NOT to change from previous election cycles. Go ahead and toss out the old state maps, and Democrats, pick your poison. With Obama, you assure yourselves gains in the Mountain West but you lose Pennsylvania and New Jersey. With Clinton you get Ohio but lose the Pacific Northwest and and a good portion of the Upper Midwest. The Survey USA poll is a reality check for those predicting a blowout for the Democrats or the Republicans. One thing is clear: we are facing a contest unlike any in forty years.