Hillary Clinton’s victory in New Hampshire was a shock to anyone who followed the pre-election polls. At Rasmussen Reports, our final numbers suggested a 7-point victory for Barack Obama. In the end, Clinton won by three. Other polls also foreshadowed a solid victory for Obama, some projecting a double digit margin. The campaigns themselves expected a significant victory for Obama (the Clinton campaign was even saying that anything less than a double-digit loss would be a victory of sorts for their candidate).

It is hard to remember a time when the polling and expectations were so universally different from what really happened. At the same time, It is worth remembering that polling was generally on target for the Republican race. John McCain won, as expected, by splitting the Republican vote with Romney and winning big among Independents. Independents accounted for 37% of the Republican Primary voters, a bit higher than projected.

So what happened with the polling on the Democratic race? There are several possibilities.

First, there may truly have been very late changes in the race. Hillary’s tearing-up moment may have played a role (another powerful moment came in the debate on Saturday night where the only woman in the race reminded everyone that she embodies change). There is some evidence to support this theory, even if we only recognize it in hindsight.

In Rasmussen Reports polling, our final trend was in Clinton’s direction—our tracking poll showed Obama’s lead declining from 10-points following the Sunday interviews to seven points after the Monday night calls. Extrapolating that trend another day would have pointed to a much closer race. Additionally, the Rasmussen Reports surveys showed that Clinton supporters were somewhat more certain that they would stick with their candidate than supporters of Obama or Edwards. If this is the case, why didn’t the late trend get more notice? Perhaps because few other firms polled on Monday night. So, the last polls reported by many continued to show an uptick for Obama.

Further support for this theory comes from Exit Poll data showing that an astonishing 38% of voters made up their mind in the final three days of the race (after Iowa). Of these, more than a third ended up voting for Clinton. These last minute decisions gave Clinton 14% of the vote overall (more than a third of her total vote). It’s easy to imagine that many of these voters had been leaning towards Clinton before Iowa, were impressed by Obama during his weekend “wave,” but came back to Clinton by Election Day.

Another possibility is that the polls simply understated Clinton’s support. At one level, Clinton’s campaign organization may have been great at getting out the vote. One analyst noted that “The Clinton turnout operation in Manchester their strongest area, was very good, and turnout soared 33% over 2000. In Rochester-Dover-Somersworth, another strong Clinton area, turnout was up 94% from 2000.” That could account for a several percentage points, but not the ten point gap between our final poll and the actual results.

The problem may also have resulted from the greatest challenge in polling--determining who will actually show up and vote. This is especially difficult in a Primary Election. It is possible, perhaps likely, that the polling models used by Rasmussen Reports and others did not account for the very high turnout experienced in New Hampshire. Rasmussen Reports normally screens out people with less voting history and less interest in the race. This might have caused us to screen out some women who might not ordinarily vote in a Primary but who came out to vote due to the historic nature of Clinton’s candidacy. The final Rasmussen Reports poll anticipated that 54% of the Democratic voters would be women while exit polls showed that number to be 57%.

A third possibility is that John McCain may have taken some independent voters away from Barack Obama. On Tuesday, Rasmussen Reports noted this two-front challenge but at the time thought it might represent a greater threat to McCain than Obama.

Given the fact that everyone was surprised, it’s likely that no one factor can explain why. It is quite possible that each of the factors mentioned above—and others—may have played a role.