Discovering higher-level Patterns (E1) - Grimmer 2013

[HyunkuKwon]

Post questions about the following exemplary reading here:

Grimmer, Justin. 2013. “Appropriators not Position Takers: The Distorting Effects of Electoral Incentives on Congressional Representation.” American Journal of Political Science 57(3): 624-642.

[jacyanthis]

This model just attempts to determine causation with a regression, but what is typically the best approach to causal inference with topic models? Difference-in-difference with one model before and one after a shock? Regression discontinuity? Instrumental variables? Or is it broadly just the same tendencies as studies with non-text data?

I understand Egami et al 2018 make a case for splitting training/testing data in causal inference using latent variables in text, but they don't really cover best practices for the causal inference itself.

[theoevans1]

I’d be interested to know more about instances in which senators violate the patterns identified here, by taking positions misaligned with constituents for instance. How consistently does this have electoral consequences? Do senator positions affect constituent opinions?

[lilygrier]

Aside from talking about domestic vs. foreign policy issues, this article did not differentiate between policy topics. I'd be interested to see how specific policy issues come into play in regional contexts. Does being a "misaligned" politician affect likelihood of expressing views on certain policy domains more than others? For example, are Democrats in somewhere like Louisiana more likely to express their views on making improvements to the education system than on abortion (as the state has strong religious opposition to abortion)?