Talking to actual people is the magical, indispendible ingredient you need in your analysis.
Of course they're quite often biased. That's not the point. But at least you get a human perspective. And even when they're biased, it turns out that "actual people" are usually far less intellectualized and rigidly ideological than the stuff you'll read online. And when they do have an axe to grind, usually there's at least a story behind it ("my family went through this") that you can at least incorporate into your dataset.
And the more actual people you talk to, the more you get to triangulate. Plus, they sometimes invite you into their house for really awesome food.
But if all one does is soak up online narratives and analysis (and worst off all, the crap that thinktanks typically put out, even when they're mostly on the right side) -- one basically becomes a language model.
Talking to actual people is the magical, indispendible ingredient you need in your analysis.
Of course they're quite often biased. That's not the point. But at least you get a human perspective. And even when they're biased, it turns out that "actual people" are usually far less intellectualized and rigidly ideological than the stuff you'll read online. And when they do have an axe to grind, usually there's at least a story behind it ("my family went through this") that you can at least incorporate into your dataset.
And the more actual people you talk to, the more you get to triangulate. Plus, they sometimes invite you into their house for really awesome food.
But if all one does is soak up online narratives and analysis (and worst off all, the crap that thinktanks typically put out, even when they're mostly on the right side) -- one basically becomes a language model.
Full of facts, but knowing nothing.