The most frequently cited polls were those by Zogby International, which showed Dean leading Kerry among likely New Hampshire primary voters by 38% to 17%, with all other candidates' support in the single digits. Zogby's national poll of Democrats in mid-August also showed Dean in the lead, receiving support from 17% of "likely Democratic primary voters nationwide," compared with 11% support for the next candidate (Dick Gephardt).
In the national and the statewide polls, Zogby's figures for Dean are considerably more positive than those measured by other organizations. The reasons for the differences may be Zogby's use of likely voters rather than registered voters, as well as the timing of the different statewide polls.
The National Polls
Four prominent news organizations conducted national polls of Democrats in August. Three of the polls show Lieberman ahead -- with leads of 4 to 10 points over his next-closest rival. The Zogby poll, by contrast, has Dean in the lead -- with the other three major candidates essentially tied for second.
|
CBS |
CNN/ |
Zogby |
Fox News/ |
|
|
Aug 26-28 |
Aug 25-26 |
Aug 16-19 |
Aug 12-13 |
|
|
% |
% |
% |
% |
|
|
Lieberman |
14 |
23 |
10 |
18 |
|
Gephardt |
10 |
13 |
11 |
12 |
|
Dean |
11 |
12 |
17 |
11 |
|
Kerry |
5 |
10 |
9 |
13 |
The polls conducted immediately before and after the Zogby poll -- by Fox and Gallup, respectively -- show similar results: Lieberman in the lead, with the other candidates essentially tied for second. After the Fox poll, did Zogby pick up on a real surge in support for Dean that dropped a week and a half later, as well as a real drop in support for Lieberman that surged 10 days later? Such short-term fluctuations almost certainly did not occur. Nothing in the news would explain such changes. Thus, the differences between Zogby and the other polls are most likely not due to the different interviewing dates.
Dean Fares Better Among People Paying Close Attention to Nomination
One of the major differences in the polls is the identification of the sampled population. CBS, Gallup, and Fox all interviewed registered voters who identified themselves as Democrats. Zogby interviewed likely Democratic primary voters nationwide. This characterization implies a subset of registered voters, though Zogby does not indicate how the likelihood of voting was measured.
Pollsters typically develop likely voter models by asking respondents how frequently they have voted in the past, their intentions to vote, and how interested they are in the campaign. Gallup did not ask all of those questions in the latest poll, but it did ask respondents how closely they had been following news about the Democratic candidates.
Among respondents who said they were following the process "very" or "somewhat" closely, Dean surges into second place. Lieberman wins among this more attentive group with 26% support, but Dean receives 19%, Kerry 13%, and Gephardt 10%.
These results suggest that when potential voters are identified by how interested they are in the campaign, Dean's standing improves. More evidence to support this hypothesis appears when we identify a subset of voters who say they are familiar with all four major candidates. In this group, Gallup shows Dean receiving 22% support, followed by Lieberman with 20%, and Gephardt and Kerry each with 14%.
The New Hampshire Primary
Two polls were conducted recently in New Hampshire -- a Zogby poll, Aug. 23-26, and an American Research Group (ARG) poll, Aug. 15-18.
|
New Hampshire Primary |
||
|
Zogby International |
American Research Group |
|
|
Aug 23-26 |
Aug 15-18 |
|
|
N=501 |
N=600 |
|
|
% |
% |
|
|
Howard Dean |
38 |
28 |
|
John Kerry |
17 |
21 |
|
Dick Gephardt |
6 |
10 |
|
Joe Lieberman |
6 |
4 |
Again, Zogby's results are more favorable to Dean than are the other polling group results. While ARG showed a seven-point lead for Dean in mid-August, Zogby showed a 21-point lead a week later. Because of the week difference, timing could have caused the wide gap. The ARG poll showed a nine-point increase for Dean and a four-point decline for Kerry from a poll conducted in July, so there was some indication Dean was gaining momentum before the Zogby poll.
The ARG poll also used likely voters, people who were registered as either Democrats or "undeclared," and who indicated they voted in all or most New Hampshire primaries. That would make the ARG sample closer to the "likely voter" sample that Zogby identified.
Many pollsters do not rely on likely voters this early in the election season, believing that it is difficult to identify them until the election is much closer. In fact, some elections show that by the time people actually vote, there are few to no differences between registered voters in general and likely voters.
Nevertheless, the results here suggest that the decision on whether to use some kind of likely voter model, especially early in the campaign, can have a major impact on what the polls say. What is not clear is whether interviewing likely voters, or interviewing all registered voters, provides the more useful measure of public sentiment.