If Twitter can predict elections, Stu Rothenberg of the Roll Call Rothenblog political blog has a bridge in Brooklyn he wants to sell you.
In response to a Washington Post article about Indiana University’s discovery of a correlation between Twitter mentions and success in Congressional races, Rothenberg doubts the relationship is a strong as researchers found it to be.
Rothenberg takes exception with the claim that IU researchers could predict 404 out of 409 competitive races by looking at “500,000 tweets that mentioned Republican and Democratic candidates for Congress in 2010.” He also doubts that the content of the tweet — either positive of negative — doesn’t matter.
The IU study, led by Indiana University Sociologist Fabio Rojas, studied the relationship of Twitter mentions between two competitors for House races. They found that the ratio of Twitter mentions between candidates became a potent predictor which one would win.
In his own assessment of House races, the Rothenberg Political Report rated “not safe” just a little over 100 contests where the results were in doubt. In the end, incumbents won the vast majority of House seats. In others, one could tell the winner by simply looking at the political flavor of the district.
In addition, Rothenberg dismisses claims that the Twitter content or whether a specific tweet originates from outside a district are not relevant, saying that it “seems to fly in the face of logic and everything political scientists believe.”
“Count me as skeptical about Rojas’ entire argument,” he adds.
Rothenberg holds significant disdain for the Post, who ran the story even though he thinks the general evidence was weak. The reason was that technology and new media are hot topics; the article was provocative and it boosts page traffic.
“Twitter can’t predict elections just yet,” Rothenberg concludes, “and it can’t cure cancer either.”