The poll sampled adults (i.e. - not registered voters, and definitely not likely primary voters), and then extracted a smaller subset of self-described "Republicans" and "Tea Party supporters" via cross-tabulation and recorded their current preferences.
Three points about this methodology: first, it is unreliable as a matter of sampling error rate, which increases exponentially each time one extracts a response to a subset question. Two: it is far too early (11 months before the NH primary) to draw any conclusions from any such poll, that (three) is itself only a "preference" poll of undeclared, potential, and largely unknown candidates, which naturally skews the results toward familiar faces.
I think the only conclusions one might reasonably draw from the results are that Mitt Romney is well-known in NH (he has a vacation house here and has traveled extensively throughout the state) and that Sarah Palin is relatively unpopular here (for a number of reasons, I think, but that's another subject).
Understatement of the year.
Understatement of the year.