By Michael Barone
Tuesday’s tracking numbers are bad news for Dean
Up through Monday, January 12, a week before the Iowa caucuses, the poll numbers from Iowa and New Hampshire since the Christmas holiday had mostly been stable and consistent with each other. The Iowa tracks and polls consistently showed Howard Dean with a lead, though not always one that was statistically significant, and they consistently showed Dick Gephardt second, John Kerry third, and John Edwards fourth, though again sometimes separated by statistically insignificant margins.
In New Hampshire, Howard Dean was well below the amazing 45 to 46 percent results he was getting in December but still led by a statistically significant margin. Wesley Clark and Joe Lieberman, who are not competing in Iowa and have had New Hampshire mostly to themselves, were gaining, to the point that Clark was coming in a solid second and Lieberman was in a statistical tie with John Kerry for third.
But the Monday and Tuesday night results were sharply different in the Zogby/MSNBC Iowa and the ARG New Hampshire tracking polls. First, look at Iowa:
|
Jan. 11-13 |
Jan. 10-12 |
Jan. 9-11 |
| Dean |
24 |
28 |
26 |
| Gephardt |
21 |
23 |
23 |
| Kerry |
21 |
17 |
16 |
| Edwards |
15 |
14 |
12 |
According to Zogby, Kerry actually led on Monday night, with 25 percent to 18 percent for Dean. Of course, one night’s sample is not statistically very reliable. Zogby’s three-night rolling track includes only 501 respondents, and the error margin on a one-night sample of 170 voters is on the order of plus or minus 10 percent; that’s why he reports the three-night average, on which the error margin is plus or minus 4.5 percent. But the numbers do support the Des Moines Register’s veteran columnist David Yepsen, who says that many Iowa caucus-goers do not make up their minds until the last minute and give serious consideration to two or more candidates in the last daysand even in the caucuses themselves, where supporters of candidates who fail to meet a 15 percent threshold at the caucus site are invited to support other candidates.
Something similar may be going on in New Hampshire. Here are ARG’s numbers:
|
Jan. 11-13 |
Jan. 10-12 |
Jan. 9-11 |
| Dean |
32 |
34 |
36 |
| Clark |
22 |
20 |
19 |
| Kerry |
13 |
11 |
10 |
| Lieberman |
9 |
9 |
10 |
| Gephardt |
4 |
4 |
4 |
| Edwards |
3 |
3 |
3 |
This looks like pretty clear movement away from Dean and toward Clark and, perhaps, Kerry. Dean has long had a larger base in New Hampshire than in Iowa, but Clark may be getting within shouting distance. And if Dean finishes second or third in Iowa, his support in New Hampshire could plunge further, and the support for Clark and/or the Iowa winner could rise.
I have some caveats. These results have not been matched in other polls. In other years, especially in 2000 in New Hampshire, there have been sharp differences between polls. There has been less of that this year, so far, but these apparent surges could be the results of one or two nights of aberrant samples. Second, it’s very hard to construct a tight enough screen in Iowa, where only about 100,000 people attend the Democratic caucuses in a state of 3 million people. Many respondents will tell interviewers they’ll participate in the caucus who will never show. Turnout matters very much. Dean supportersopponents of the Iraq war and Bush hatersseem more highly motivated than Gephardt’s elderly union backers and Kerry’s veteran political activists. But that may be counterbalanced if, as I observed, many of Dean’s workers are out-of-state kids while most of Gephardt’s and Kerry’s are Iowans.
But there is some evidence that Dean is seeing the same numbers in his internal polls. He has put up an antiwar ad in Iowa, which in the 1970s and 1980s was unusually dovish, and he has scampered to make an appearance today (Wednesday) in New Hampshire, to get some time on Channel 9 to counterbalance Clark and Lieberman. Stay tuned.