Conventional wisdom holds that Bernie Sanders' and Donald Trump's big wins in New Hampshire’s presidential primary earlier this month were driven by hordes of irregular and first-time voters flocking to the polls.

But a review of preliminary voting data doesn’t exactly back up that premise.
Updated numbers from the New Hampshire Secretary of State’s office show that, statewide, close to 12 percent of all voters in the Feb. 9 primary registered to vote at the polls. Those were either first-time voters or voters who were registering in a new locality. (That's virtually the same rate of voters registering at the polls in 2008, the last time both political parties had contested presidential primaries.)
That figure varied considerably across the state this year. Communities with the highest rate of new voters on Primary Day included the state’s college towns (Hanover, Durham, Plymouth and Keene) as well as bigger cities like Manchester, Concord and Dover.

If you follow the theory that Sanders and Trump drew lots of new voters to the polls, you would expect those locations to have supported the two candidates at rates higher than the statewide average.
The story is not that tidy, however, especially for Trump.
Let’s look at the 15 communities with the highest rate of new voters as a share of the Primary Day electorate. Just four - Plymouth, Rindge, Manchester and Boscawen - voted for Trump at a greater rate than the state as a whole did. And they were all within 5 percentage points of Trump’s statewide rate.
Next, look at the New Hampshire towns where Trump performed best.

Only one of those towns (Wentworth’s Location) saw new voters turning out at greater rate than the state as a whole.
It’s the same story for Sanders: of the 10 communities where he saw his biggest victory margins, just one had higher than average new-voter turnout.
For Sanders, though, there is a wrinkle in this analysis.
Look back at the first table, and you’ll see that six of seven communities with the highest rate of new voters backed Sanders by higher than average rates.
But four of those places - Durham, Keene, Plymouth and Rindge - are college towns, which would have higher rates of new voters in any case: many college students have only just recently become eligible to vote.
But once you get past those towns, there appears to be little correlation between a town’s rate of new voter registration and its support for Sanders.
One thing we can’t tell from this data is how many “infrequent” voters were drawn to the polls on Primary Day to cast ballots for Sanders or Trump -- that is, how much of those candidates' support came from registered voters who rarely vote. To calculate that, we'd need access to the state voter file and do a painstaking analysis of voting patterns, voter by voter, town by town.
Still, this preliminary data doesn't seem to support the premise that Trump and Sanders owe their Granite State wins to new voters alone.