I shouldn’t be disappointed with MSNBC’s "Daily Rundown," one of the few reasonable political shows not airing on Sunday morning. But the show did a segment with Indiana University sociologist Fabio Rojas, who recently wrote a Washington Post opinion piece on how Twitter can predict an election.
Instead of featuring Rojas alone, substitute host Chris Cillizza could have aired a “Deep Dive" segment with Rojas and someone with a demonstrated background in election forecasting. Michael Lewis-Beck (University of Iowa), James Campbell (SUNY-Buffalo), former ABC News pollster Gary Langer (Langer Research Associates) or any of a number of other academics or professional pollsters could have raised important questions about the study and its conclusions.
Unfortunately, the IU sociologist was no more convincing on the "Daily Rundown" than he was in his Washington Post column.
Perhaps his strangest comments were once again in response to questions about why merely adding up all tweets about candidates, including very critical ones, can predict who will win an election.
“All publicity – or most publicity – is good,” Rojas said in response to a question about why negative tweets predict victory as much as positive ones. “The buzz is an indicator that you are picking up support. People don’t like to talk about losers,” he said in one of his more bizarre observations.
Pollsters, apparently, have been making a mistake for decades asking respondents whether they have positive or negative opinions of candidates. If Rojas’s findings and analysis are to be believed, it’s irrelevant how voters evaluate the candidates. (Presumably pollsters have also been wrong to limit their surveys to adults, registered voters and people who live in the relevant states and districts.)
Anyone truly interested in the subject of social media and public opinion might pull up Langer’s very accessible briefing paper on the subject.
It will tell you a lot more than you’ll learn by reading the Rojas opinion piece or watching "Rundown" segment.