lukestein’s Twitter Archive

…in reply to @sewenz
@sewenz @causalinf In a sense it’s all definitions, but here’s an example of what I had in mind: Suppose we wanted to understand the effect of being an older vs. younger sibling on earnings. There are lots of different “methods” we can apply (e.g., various means comparisons or coefficients from [1/
Permalink On twitter.com ♻️ 2 Retweets ❤️ 1 Favorite 2021 May 25 Mood 0

…in reply to @lukestein
@sewenz @causalinf various regressions. Any one of these methods can deliver a number as long as I can figure out how to (a) get data and (b) program my computer to work with that data. Either of (a) and/or (b) might be difficult to implement. [2/
Permalink On twitter.com ♻️ 2 Retweets ❤️ 1 Favorite 2021 May 25 Mood -1 🙁

…in reply to @lukestein
@sewenz @causalinf And we’ll have to argue or agree about what the estimated quantity says or means about the “effect” of being an older sibling is (and what that even means?!) which will usually require some set of assumptions. @lukestein/1397261994693218311
Permalink On twitter.com ♻️ 2 Retweets ❤️ 1 Favorite 2021 May 25 Mood +1 🙂

…in reply to @lukestein
@sewenz @causalinf And those assumptions are what might drive choices about things like how to handle middle children or only children, or how to “control” for age (correlated with older sibling status *and* earnings), or to look for data on unplanned pregnancies, or whatever 4/4
On twitter.com ♻️ 2 Retweets ❤️ 1 Favorite 2021 May 25 Mood +2 🙂