DailyKos polls were "bunk"

Markos Moulitsas, the proprietor of the popular liberal blog DailyKos, just posted a report that he believes shows the site's pollster, Research 2000, was using "fraudulent" numbers.

As Josh Marshall writes, the report is "explosive"; if Kos and the three statistics experts who approached him are right, they've caught a well-known and oft-cited pollster in outright deception.

And it also caught our eye.

The Day has used Research 2000 on at least two occasions to poll Connecticut races, once in then-Rep. Rob Simmons' defeat of Democratic challenger Jim Sullivan in 2004, and again in the run-up to the 2006 Democratic Senate primary between Sen. Joseph Lieberman and challenger Ned Lamont.

In the Simmons race, Research 2000 and pollster Del Ali showed Sullivan within striking distance of Simmons in late October, just before election day. The cost of that poll was shared with the Manchester Journal-Inquirer. The story the next morning declared it a "virtual dead heat," with Simmons at 48 percent of the vote and Sullivan at 45 percent.

Simmons went on to win by eight percentage points, prompting our former colleague, columnist Steven Slosberg, to remark, "Some dead heat."

More from that column:

For this year's 2nd District race itself, pollsters like Del Ali of the Maryland-based Research 2000 pooped out. But what's a pollster, if not plucky."

On the morning after, Ali was all about spin, and the accuracy of his polling during the last week of October. Ali's company was commissioned by The Day, the Journal-Inquirer of Manchester and the Waterbury Republican-American to do polls of 2nd District voters. The most recent one, sampling 600 likely voters by telephone between Oct. 26 and Oct. 28, showed Simmons with a 3-point lead over Sullivan, 48 percent to 45 percent. That was up from 2 percent in the previous poll.

Allowing for a margin of error of plus or minus 4 percent, the poll's outcome, in the operative phrase of the moment, showed a virtual dead heat.

So what went wrong with the polling? Nothing, said Ali, in the face of Simmons losing only nine of 65 towns.

"I think the poll was right there at the time," said Ali by phone from Rockville, Md. "The turnout for the group between 18 to 29 was it. They just didn't show up. The poll showed the top issue was Iraq, Iraq, Iraq, especially with that 18 to 29 group. They didn't come out."

In 2006, Ali's poll showed what others had: Lamont, a virtual unknown, cruised out to a lead over Lieberman, and beat him in the primary. He would go on to lose the general election that November.

We have no idea if any similar shenanigans happened in the polls conducted for this paper by Research 2000. But if the polling budget is replenished anytime soon, I have a feeling we'll be shopping elsewhere.

Meanwhile, Moulitsas says his site will sue Research 2000, while Ali tells TPMmuckraker that his company "conducted EVERY poll properly for the Daily Kos."

Reader Comments