When in comes to polling, Andrew Smith is to New Hampshire what Gallup last week was to the national scene.
His polling data is totally out of whack with other pollsters in the field just as pundits (mostly Democratic) were quick to point out last week how Gallup was so far off base when it had Romney seven points ahead of Obama. The Gallup number is down to five this week, about the same as Rasmussen's number Wednesday (Romney up four, 50-46).
However, there are no indications that Smith's numbers will move back to the realm of credibility, and I say this as someone who has respected and defended Smith for years.
It's hard to continue that respect this year. His numbers have been all over the map, and yes, I realize that numbers will vary depending on when a poll is taken. However, not only is Smith out of synch with four other pollsters who comprise the Real Clear Politics Average, but his internal numbers simply don't add up.
First the five polls. Two have Romney slightly ahead; one has Romney slightly behind; one has a dead heat; only UNH has a wide Obama lead. Here are the numbers.
Smith (the WMUR/Granite State Poll to be precise) has Obama up 9 points--51-42 in New Hampshire.
Rasmussen has Obama up 1 point, 50-49.
Suffolk University has it dead even at 47-47.
PPP (a Democratic outfit!) has Romney up 1 point, 49-48.
ARG has Romney up 2 points, 49-47.
Only the out of whack UNH data gives Obama a lead in the average, 1.4 points (48.6-47.2).
This is enough for RCP to move our dear little state back into the Obama column, but the Smith numbers simply don't make sense. Go to realclearpolitics.com, and you can click on the UNH poll to get reams of internal numbers, but here's the one which makes the entire poll suspect.
The poll has Obama leading 94-2 among Democrats, no surprise. Romney leads 91-4 among Republicans, no surprise, but given those numbers, one would be forced to intuit that to be ahead by nine points overall, Obama must be up six or seven points among Independents. In fact, the Smith poll has Obama up only two points there, 39-37. The only way these numbers add up is if the poll sampled many, many more Democrats than Republicans and since the lastest numbers from the Secretary of State's office show four percent more registered Republicans than Democrats in the state (it's about 40% Undeclared, 32% Republicans, and 28% Democrats), no pollster with credibility would oversample Democrats.
The UNH numbers make even less if we look at likely voters. Smith has Romney up overall by eight points, 49-41. However, he actually has Romney leading among Independents 37-36 while Obama leads 92-2 among Democrats and Romney 91-3 among Republicans.
How does that make sense unless the sampling is heaving skewed toward Democrats? With each candidate taking almost all his base and with Romney up one point with Independents, the race should be dead even or close to even--like the other pollsters have it.
As always, once I begin to distrust anyone (whether it be in polling or the seven Manchester Republican snakes who went back on their word on redistricting), I tend to write them off and never trust them again.
We may be headed in that direction with UNH polling. Remember, a few weeks ago, Smith had Obama up 15 points and then up only six the very next week (even when the first debate bounce was not entirely cooked into the numbers).
This is important because Smith, also unlike all other pollsters, has Maggie Hassan up beyond the margin of error for Governor, six points (37-31 with a whopping 32 percent undecided).
Until Andy Smith's numbers begin to add up, we should either take them with a shaker of salt or ignore them completely.
If Obama were up nine points here (as Smith tells us), who on Earth thinks he would feel the need to keep coming back here to campaign?
No one of course.
It's close to dead even here, and the WMUR/Granite State poll cannot be trusted.