NY Times Disputes Validity of Pollster Showing Close Mass. Senate Race, Yet...

January 8th, 2010 2:42 PM

On Friday, New York Times reporter Abby Goodnough described the surprise struggles of Mass. Democrat Martha Coakley, once considered a shoo-in to fill the Senate seat of the late Ted Kennedy, but now facing a strong challenge from Republican Scott Brown: "In Massachusetts, Surprise Anxiety for Favored Democrats."

Goodnough's hook was a Rasmussen Reports poll showing Brown within nine points of Coakley. But she emphasized that "many news organizations dispute its methodology." Yet Rasmussen called the 2008 election with far greater accuracy than did the Times.

(Goodnough also authored an admiring December 10 profile of Coakley after her win in the Democratic primary.)

From her Friday story:

Martha M. Coakley, the Democrat running for Senator Edward M. Kennedy's seat in Massachusetts, had seemed so certain of winning the special election on Jan. 19 that she barely campaigned last month.

But the dynamic has changed in recent days. The news that two senior Democratic senators will retire this year in the face of bleak re-election prospects has created anxiety and, even in this bluest of states, a sense that the balance of power has shifted dramatically from just a year ago.

With the holidays over and public attention refocused on the race, Ms. Coakley's insistence on debating her Republican opponent, Scott P. Brown, only with a third-party candidate present has drawn mounting criticism.

And a new poll that showed a competitive race between Ms. Coakley and Mr. Brown has generated buzz on conservative blogs and energized the Brown campaign -- though many news organizations dispute its methodology.

In a sudden flurry of activity, the Coakley campaign released its first television advertisement on Thursday and accepted the endorsement of Mr. Kennedy's widow, Victoria Reggie Kennedy, at a splashy event outside Boston.
A Brown win remains improbable, given that Democrats outnumber Republicans by 3 to 1 in the state and that Ms. Coakley, the state's attorney general, has far more name recognition, money and organizational support.

Goodnough later sniffed once more at the methodology of the Rasmussen poll, which fails to meet Times standards. (Hating Rasmussen polls is a cottage industry on liberal blogs, since they often show reasons for Republican optimism.)

The poll that suggested Ms. Coakley's lead was narrowing, which was conducted by Rasmussen Reports and does not meet the polling standards of The New York Times because it relied on automated telephone calls, suggested Mr. Brown had strikingly strong support among independent voters. But most of them are unlikely to come out for a special election at an odd time of year, Ms. Marsh said.

So how did the Times's vaunted polling standards (which have been faulted by Times Watch and others for over-sampling Democrats) match up with Rasmussen's when it came to, say, predicting the last presidential election? The poll clearinghouse at Real Clear Politics provided the answer.

The last Rasmussen poll before the 2008 election (released the day before) had Barack Obama beating John McCain by a six-point margin, 52%-46%. That was very close to Obama's actual margin of 53%-46%, meaning Rasmussen came within a single point of the final result, allegedly dubious standards and all.

By contrast, the last CBS-New York Times poll before the election made the front page on Friday October 31, four days before the vote, and had Obama winning by 11 points, 51%-40%. The Times overestimated Obama's win by four percentage points. This analysis by Fordham University ranks Rasmussen first and the Times/CBS poll next to last in accuracy. Too bad Rasmussen doesn't meet the Times's "polling standards."