Statistics Still Do Lie

Ferrer and Undecided lead muddled Democratic primary pack

The latest numbers in the headlined new public poll by Quinnipiac are an optical illusion. Since last Wednesday, the Big News this week was that the Democratic primary is tightening, with Gifford Miller and Anthony Weiner rising, Virginia Fields sliding, and front-runner Fernando Ferrer holding steady at 33 percent, still seven points shy of the 40 percent he will need to avoid a runoff. But the reason Miller and Weiner appeared to gain two and five points respectively between Quinnipiac's July and August polls is that the poll changed, not their popularity.

In July, Quinnipiac polled a sample of registered Democrats, as it has five other times in 2005. In August, it did two polls—another one of 1,601 registered Dems and, for the first time, a sample of a measly 489 Democrats likely to vote on September 13. If the media focused on the same poll of registrants they've been covering all year, we would've found out that Miller actually dropped five points, from 15 percent to his all-time low of 10, while Weiner remained at the 11 percent he's been at since January.

So when the press instead reported that Miller and Weiner had surged to 17- and 16-point highs, it was comparing incomparables. The two of them may have been getting those percentages among likely voters for months but no one was polling "likelies," much less pushing them, as they were pushed in this poll, to say which candidate they were "leaning toward." Likelies and leaners are a totally different universe than the registrants sampled in all the prior polls. This close to the primary, it makes sense for Quinnipiac to survey the leanings of likely voters, but with no comparative baseline from prior polls, both Quinnipiac and the media should have gone out of their way to make clear that the results don't signal candidate trends since July.

Miller: Dominates debates, subverts reforms
photo: Kate Englund
Miller: Dominates debates, subverts reforms

In fact, the registrant results, which are the only ones measurable over time, reveal that the Big Gainer in the last month was that very popular Democrat, Undecided. He had been stuck in place all year, hovering between 20 percent in February and 18 in both June and July. But all of a sudden, in August's poll of registrants, Undecided hit 29 percent, a stunning 11-point hike in a month. It's also 13 points higher than the 16 percent of likely voters who say they're undecided. It is the clearest statistical change since primary polling began this year.

Undecided has, after more than a half-year of savvy campaigning, actually tied Ferrer at the head of the Democratic pack. As quiet as it is kept, Ferrer fell four points between July and August among registered Dems, dropping to the same 29 percent as Undecided. Ferrer's support only looked unchanged because he got 33 percent among likely voters this time and 33 percent among registrants in July, but if Quinnipiac hadn't switched the polling pool, the Big Story last week would've been Ferrer's decline.

No one's benefited more, however, from these polling illusions than Fields. Her 17-point showing among likely voters in August still left her tied for second with Miller, a single point higher than in July. But among registered Dems, she plummeted to 13 percent, a full 10-point collapse since her high of 23 in May. Unlike the three others, she has been in consistent free fall the last four months.

What it all adds up to is that Democrats are running away from this primary field, with an additional 8 percent of registrants indicating that they won't vote or will vote for someone other than the Four Dems, a doubling of that rejection rate since March. What it also says is that we'll have to wait until the next round of polls, when we can compare one set of likelies and leaners with another, to see if any of the candidates is really gaining. Here are a few of the other secrets of Quinnipiac's poll:

Ferrer is actually more popular among whites now than he is among blacks, a shocking reversal of his 2001 performance. Only 21 percent of likely black voters view him favorably, while 34 percent of whites do. His 27 percent unfavorables among likely black voters is by far the highest of the four. Thirty percent of likely white voters, as opposed to 24 percent of blacks, say they will vote for him in the primary. Ferrer actually leads Miller and Weiner among white voters, while Fields trounces him with black voters, getting 37 percent. This contrasts with the July poll of registered Democrats, when Fields and Ferrer split the black vote almost evenly, 34 to 33 percent. While these are not comparables, the 9-point drop suggests that Ferrer's slide with the black base that supported him four years ago is continuing.

Black voters who are leaving Ferrer are not, however, going to Fields. While only 15 percent of black registrants pulled the lever for Undecided in July, 31 percent did in August, meaning that blacks, the key swing vote in November, are moving away from all of the Four Dems faster than anyone else (the white percent went from 22 to 29). Even within the much smaller pool of likely voters, blacks are more up in the air than whites, 21 to 14 percent, with only 10 percent of Hispanics undecided. In fact, the prototypical undecided yet likely voter is a black woman from Brooklyn and Queens. Only 2 percent of the Bronx's likely voters are undecided, compared to 21 percent of Queens'.

Next Page »