Tuesday, February 4, 2014

Andrew Gelman opens a couple of cans of worms

Over at the Monkey Cage, Andrew Gelman has some kind words for West Coast Stat Views and our recent post on nepotism. I have to admit I was less pleased with this comment (though for a somewhat perverse reason):
Good question. As a political scientist, my question is: Can this be studied empirically in some way? I’m not quite sure how, but I think it could be worth looking into. It fits into some of our general questions about political discourse and economic perceptions.
It's not the topic I object to. Quite the opposite, this opens up a couple of big questions that I've been wanting to discuss in depth for a long time:

Are we adequately tracking the attitude and perceptions that need to be in our models?;

Are the standard technique of data collection and analysis up to the job?

For example, in the 2012 election, we spent a lot of time asking whether Obama had successfully painted Romney as a plutocrat but perhaps not enough asking how people felt about plutocrats and, possibly more importantly, how those attitudes have shifted over the years (and whether attitudes in the media have stayed in alignment).  We analytic types also relied heavily on polling data rather than other sources like text mining, non-verbal response data and implicit association tests, to name a few, Of course, the standard approaches proved more than adequate for the purpose of prediction but did they give us what we needed to understand the underlying dynamics? And will these standard approaches continue to provide reliable data in the future?

So what was my problem with Gelman's comment? It was the timing. I've set aside most of January and February for other projects and now I don't have the bandwidth to jump into this debate in a serious way.

So much to blog about, so little time.

No comments:

Post a Comment