Looking Back at Daily Kos, Research 2000, and Bill Halter

If you haven't heard, the Daily Kos/Research 2000 polls, which played a significant role in the election-polling landscape for the past year and a half, have been tentatively discredited (at least, I suppose, until we find out more through court documents); Daily Kos thinks the polls were "likely bunk" and is suing Research 2000 over it.

These polls weren't conducted for every race, but they were done heavily for those races found to be interesting/important by Daily Kos, and, by extension, the progressive online community in general. One such race was the Arkansas Democratic Senate primary between Sen. Blanche Lincoln and Lt. Gov Bill Halter, the liberal (for Arkansas) challenger who mounted a serious challenge with the help of national labor unions and the millions of dollars they poured into the race.

I may be late to the game in pointing out things like this, but the Daily Kos/Research 2000 polls gave Halter a considerable boost, showing him ahead (though not quite comfortably) of Lincoln as the two entered the home stretch toward their June 8 runoff. After Halter forced said runoff in the  May 18 primary, collecting almost as many votes as she did, the race drew a good deal of national media attention. Labor poured millions more into the race, and, as media outlets looked to see who might win, Daily Kos/Research 2000 polls were the only major polls available.

On May 26, R2K showed Halter leading by three percentage points; on May 4, it showed him leading by four.

As it went, Lincoln won the runoff 52% to 48% and retained the Democratic nomination for her Senate seat. The polls missed that result seven or eight percentage points.

I don't know if Daily Kos is right about Research 2000's polls. I am not a polling expert; I am not a statistician; I have not conducted any rigorous statistical analysis or research into Research 2000's stated methodology, any unpublished magic formula to determine who "likely voters" are, or whether it follows its stated methods. I am suspicious of all polling, and I have never fully grasped the concept of standard deviation. I'm just a guy who read the complaint where Daily Kos leveled the "bunk" accusation. I make no claims otherwise.

So, with that said, I think it's fair to point out that, if Daily Kos's lawsuit bears out its accusations--if Research 2000's polls are shown, after hard evidence is revealed, to in fact have been "bunk"--the Arkansas primary will be one of the bigger examples where we look back and scratch our heads, wondering if that polling possibly affected the race and made it into an even bigger national news story than it already was, shifting the tone of the cable, web, and print coverage, and influencing how a significant event was anticipated and discussed by the whole of the political commentariat.