It depends on what questions you're trying to answer. My intuition tells me that predicting socioeconomic status with Facebook's data would probably be reasona

It depends on what questions you're trying to answer. My intuition tells me that predicting socioeconomic status with Facebook's data would probably be reasonably successful, and you could look at feature importance from that. If they're trying to test every theory that Munoz came up with in 4 years of brainstorming or w/e though, then I think they'll likely find it's an insufficiently complete dataset.