4 Replies to “[2020 Fall] POLS GA 3200 Quant Field Methods”

  1. Here is my first cut at doing the design effect exercise. One thing that I want to highlight is that I am implementing estimators using the “estimatr” package functions, which are what you would want to use in actual applications so as to easily get the right standard error estimates: [.html]

    I am actually getting substantial gains from the blocking, so I need to see what things are different here as compared to what the two groups obtained. Maybe there is a bug — let me check. Nonetheless, have a look at least to see how to implement the estimators.

  2. Here is a link to an older version of the syllabus, which used to ask you to do more readings. I’ve found that coding delivers more insight than reading for this stuff, but it may still be nice to have these to get the ideas another way. Focus on the readings for the first three weeks: [.pdf]

  3. There is a really dumb coding mistake in our code 🙁

    When generating the randomized y, we are using

    y <- zy1 + (1-z)y0

    instead of

    y <- zdt_cluster$y1 + (1-z)dt_cluster$y0.

    Since we changed the order of the data frame at the very beginning (dt_cluster % arrange(x)), the treatment is not assigned correctly.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.