The liberal collaborative blog DailyKos yesterday renounced all work performed for it by the polling firm Research 2000, issuing a damning report that makes a convincing case that Research 2000’s numbers were either fabricated or tampered with.
As it turns out, the St. Louis Post-Dispatch used Research 2000 extensively in the 2004, 2006 and 2008 election cycles. It’s weird that the St. Louis Post-Dispatch would use the same pollster as Daily Kos. It’s disturbing that of the two, Daily Kos would be the ones to figure out that polling numbers were bad. It’s disgraceful that Daily Kos was the only one of the two to tell its readers about it.
It’s not just liberal blogsites that are beating the St. Louis Post-Dispatch on the whole integrity thing. Many former Research 2000 clients and groups that used their polling numbers are issuing retractions or pulling the data – including the Lexington Herald-Leader, Talking Points Memo and Greg Sargent from the Washington Post.
With Research 2000’s reputation in tatters, it’s worth revisiting some of the Post-Dispatch’s work with Research 2000. One article in 2008, titled “Poll Results are Clouded” might go down in history as the most unintentionally true article title the Post-Dispatch has ever posted.
Another article, trying to be insightful about how politicians treats its Research 2000 polls, writes that “Whether or not a politician agrees with the numbers, he or she makes sure to weigh in to put the best face on the results.” (Poll Numbers Are In, Now We Can Let The Spin Begin, 2006) It goes on to say “Those who dislike the poll's findings challenge some of the sampling techniques or questions, or say their own polling has shown their side's politician to be in slightly better shape.” The one possibility they didn’t consider was that four years later, the polling firm would be discredited for fabricating results.How many polls did the St. Louis Post-Dispatch conduct with Research 2000? When were they conducted? Is there any reason to believe that their work was compromised? These questions demand some sort of answer.