Ordinary Squares: Thoughts after a Year of Stats Classes
I have a confession to make. As an undergrad, I failed statistics. So I guess I had something to prove by taking the entire course of quantitative methods at IR/PS. By the end of the third quarter, in the middle of QM3, deep in panel data and time series analysis, I came to the realization that we were at the edge of the epistemic enterprise. The view was expansive--and completely terrifying.
Luckily, Professor C McI, development econ guru and stats wizard, was there to provide reassurance and entertaining quotes, "At the end of the day, it's easier to dress up a pig, because we know the pig. The pig is OLS (Ordinary Least Squares)."
Craig is a rebel econometrician. He professes that the data should speak for themselves, regardless of some traditional theoretical underpinnings. "I'd probably get dragged through the streets by some of the professors in the political science department for saying this, but..." was the way he prefaced half of what we learned. Yet PoliSci PhD students flock to this class because it is about results, not just long procedures. (Not to say the procedures aren't long and tedious).
When some of us started to gripe that our data collection and analysis of some policy problem was taking too long and ruining our lives, (I spent 36 probably hours collecting the data alone) he said, "Well, if you wanted to save time, then you probably shouldn't be in this class."
Describing one encounter and difference of opinion with our QM2 professor, he said, "We both paled beneath our tans each others lack of rigor."
Sometimes the data looked "weird enough" to take a closer look at. "Like someone missing their six front teeth."
We knew what we shouldn't run regressions on. "Testing discrimination statistically is just a veil of tears." And we knew that the compass clues of old cross-panel (single time period) regressions were out the window. "Looking at R Squared to ascribe significance and determine causality is like holding hands with the devil."
Maybe I'm over ascribing causality here, but I don't think most of us would have lasted two days without our professor making mind-boggling concepts as easy as taking a random walk.
On the last day, knowing we were left with only our ox-stunning econometrics text, an inch of notes, and a dozen STATA .do files, he said, "We end the way we have lived. In a kind of headlong rush."
Luckily, Professor C McI, development econ guru and stats wizard, was there to provide reassurance and entertaining quotes, "At the end of the day, it's easier to dress up a pig, because we know the pig. The pig is OLS (Ordinary Least Squares)."
Craig is a rebel econometrician. He professes that the data should speak for themselves, regardless of some traditional theoretical underpinnings. "I'd probably get dragged through the streets by some of the professors in the political science department for saying this, but..." was the way he prefaced half of what we learned. Yet PoliSci PhD students flock to this class because it is about results, not just long procedures. (Not to say the procedures aren't long and tedious).
When some of us started to gripe that our data collection and analysis of some policy problem was taking too long and ruining our lives, (I spent 36 probably hours collecting the data alone) he said, "Well, if you wanted to save time, then you probably shouldn't be in this class."
Describing one encounter and difference of opinion with our QM2 professor, he said, "We both paled beneath our tans each others lack of rigor."
Sometimes the data looked "weird enough" to take a closer look at. "Like someone missing their six front teeth."
We knew what we shouldn't run regressions on. "Testing discrimination statistically is just a veil of tears." And we knew that the compass clues of old cross-panel (single time period) regressions were out the window. "Looking at R Squared to ascribe significance and determine causality is like holding hands with the devil."
Maybe I'm over ascribing causality here, but I don't think most of us would have lasted two days without our professor making mind-boggling concepts as easy as taking a random walk.
On the last day, knowing we were left with only our ox-stunning econometrics text, an inch of notes, and a dozen STATA .do files, he said, "We end the way we have lived. In a kind of headlong rush."
Labels: mcintosh, QM3, statistics, stats