National polls show us one view of the race and state polls show us another view. Individual pollsters such as Gallup, Susquehanna and Rasmussen show a 2012 electorate likely to be favorable to the GOP. Other pollsters such as PPP, NBC/Marist/Quinnipiac show the President in a strong position even as preliminary indications in several states point to serious drop-offs in early voting totals among his base. So what gives? Who is right and do the polls really tell us anything heading into Tuesday?
Ever since the beginning of the year there has been a noticeable divergence between different pollsters samples, demographically as well as partisanly. Gallup and Survey USA’s numbers since the introduction of their LV models have pointed to a demographic and party ID electorate much like 2004. Other pollsters such as PPP, and yes even Rasmussen, have pointed to a more Democratic and racially diverse electorate then 2008. This as everyday we have been inundated with stories of how depressed the Democratic base is and how excited Republicans are.
National polls have trended Romney’s way since the first debate in October. But many state polls, which show a more Democratic electorate in numerous states then even 2008, have given Obama the edge. Gallup’s surveys of battleground states (12 of them as whole) have found by contrast Romney ahead by narrow margins. So who is to be believed? It honestly is not a question of whom to believe but a question of methodology standards.
Gallup and Survey USA both believe the electorate will be more like 2004 racially and politically. Other pollsters disagree. Considering how racially polarized American politics is today it is thus no surprise that these differences thus lead to widely divergent results. Hispanics and African-Americans are strong Obama backers but erosion among his white support means he needs 2008 level turnout to offset his losses among whites. Romney on the other hand needs 2004 to 2006 level white turnout to win a close race.
Politically, Gallup has found that the public has become more Republican while other pollsters disagree. Gallup found that the electorate is now almost 5 points more Republican than 2008 while Democrats have fallen two points. Politically this means Republicans outnumbered Democrats in all Gallup surveys conducted since October 1st 37%-36%. No doubt these numbers are influenced by Gallup’s racial polling samples.
Individual state polls have shown a distinct proclivity to over sample or at least find Democrats far outnumber Republicans. Even when GOP candidates or Romney win independents by double digits they still trail. If this holds true then Democrats and Obama will win a route on election day. Confusing things even more is that the Census Bureau estimated white turnout in 2008 was 76%, not 74%, and in key swing states such as VA and OH, white turnout was far below the norm for an electorate that was 76% white. For the Romney campaign this bodes well.
Both Democratic and GOP pollsters have also used widely divergent samples. Democrats see an electorate like 2008 or even more favorable to them. The GOP sees a sample much closer to 2004 or 2006. More troubling to many Republicans is minus Gallup and Rasmussen many pollsters have opted to go with Democratic samples for their polling samples.
All these factors add up to the inconvenient truth that we will not be able to call many states until well into November 6th (if the polls are right). Maybe pollsters have become used to wave elections but this may be the beginning of the new normal when pollsters go with widely different samples and hope they are right. The political parties and individual partisans may well simply have to hope the pollsters samples that benefit their side are right.