This item by TDS Co-Editor William Galston is cross-posted from The New Republic.

Recently, three respected national surveys--Gallup, Pew, and now Battleground--have given Republicans a double-digit edge among likely voters. While I'm no expert on this history of public opinion research, I can think of no parallel to these findings during my three decades of involvement in national politics.

There are only two possibilities: Either this election is so distinctive that existing likely voter models, which are derived inductively from past experience, are simply inapplicable, or we are looking at a potential Republican sweep of historic proportions, larger even than 1994, long regarded as the ne plus ultra of contemporary swings. If so, the oft-repeated characterization of this election as a "wave" seems inadequate; tsunami would be more like it.

In particular, these findings have implications closely contested Senate races, which are numerous right now. During recent decades, three elections--1980, 1986, and 2006--have featured tossup races that all ended up falling in the same direction. If Republicans enjoy anything like a double-digit edge on November 2, 2010, may well be another such election.

This is a time of testing--for Democrats, but also for the profession of survey research. On November 3, one or the other will have to go back to the drawing board.