A WPRI-TV poll released October 30 had Congressman David Cicilline clinging to a one-point lead over his Republican challenger as the campaign entered its home stretch. And it was hardly the only survey suggesting a tight race.
The television station conducted a poll a month earlier that gave Cicilline a 6-point lead. A Brown University poll released shortly thereafter found the same gap. And the campaigns' internal polls showed a tight race, too - nothing like the 12-point win Cicilline would eventually register on Election Day.
So, how did the polls go so wrong?
There is, of course, the possibility that they didn't. Public opinion surveys, as WPRI's respected pollster Joe Fleming pointed out in a chat this morning, are snapshots in time. It's possible that voters moved, in big numbers, to Cicilline in the closing days of the campaign - after the final polls had pulled out of the field.
The race's remarkable fluidity - in February, a WPRI poll gave Doherty a 15-point edge on Cicilline - lends some credence to this argument. Butit's a little hard to believe that Cicilline's lead really jumped from 1 percent to 12 percent in just a couple of weeks.
More likely: pollsters undersampled certain segments of the electorate. Some sources have suggested that the Latino vote in areas like south Providence and Central Falls may have been underrepresented. Fleming agreed that could be an issue. But that wouldn't explain the full gap.
Pollsters, as Fleming himself suggests, may have simply undersampled Democrats - failing to account for a big turnout by the party faithful in a presidential election year.
The state does not yet have data on who voted in the election, by party. But the circumstantial evidence supports Fleming's theory. While turnout in the First Congressional District actually declined some 9.8 percent between the last presidential election, in 2008, and this year's contest, the drop-off was not as sharp in three urban, Democratic strongholds I examined - three cities that were vital to Cicilline's victory.
Turnout dipped just 6.2 percent in Providence (which is divided between Cicilline and Congressman James Langevin's districts) and 5 percent in Pawtucket. In the much smaller Central Falls, which had a feisty mayoral contest on the ballot, it actually jumped 1 percent.
Fleming's polling sample roughly matched the party breakdown among registered voters in Rhode Island. He said he would consider upping the Democratic quotient in the next presidential year, 2016. If he had done the same this year, we might have learned, a bit earlier, what emerged as a central truth on Election Day: the race was not as close as we imagined.