| Here's an outline of our method, which might help explain this (And please feel free criticize it): First, we look at news aggregators and twitter trends to see what the big stories of the day might be. Then we pull a bunch of tweets with our social media listening tool and try evaluate the story on three criteria: Momentum (is it trending and growing), Partisanship (are two sides talking about it), and Emotion (Is the story morally animating). Then we collect data by pulling tweets through our tool, and focusing on the most widely shared and engaged tweets, and sort them into pro/con (or whatever division fits the story. Then it's analysis. We identify the facts each side is referencing, the affected values and emotions from either side, and whatever the key source of division is (if there is one. Like any methodology, this has its limitations. It only looks at twitter, for example. If there is an argument or position out there that doesn't come through in the data, then we can't really assume it's there (although we do use broader ideas to contextualize and explain the data). |