A linguistic turn

Greetings patrons,

It’s time for another research update. In today’s update, four things:

1. Back into lockdown

2. Analyzing words

3. Probability and intuition

4. Post frequency

Back into lockdown

Like many places in the world, Toronto is struggling to contain its Covid numbers. Midway through our daughter’s winter break, we found out that for the first week of January (probably more) she’s not doing in-person classes . That means she’s home with us all day … theoretically doing school.

To stay productive, I’ll be returning to my 5 am writing schedule. It’s something I started last March when we first went into lockdown. I get up at 5 am and do about 3 hours of writing before my daughter wakes up. This early-bird schedule was more pleasant in the summer (when there was 5 am sun). But what I lack for sunlight I supplement with coffee.

If you’re home with your kids, I hope you have a similar sanity outlet.

Analyzing word frequency

Last October, I took a foray into linguistics. In Deconstructing Econospeak, I measured how the frequency of words in economics textbooks differs from word frequency in mainstream English. The results were fascinating, mostly for what economists neglect. (Words like feminism, imperialism, and censorship are conspicuously under-used.).

Since publishing Deconstructing Econospeak, I’ve been contacted by a variety of researchers seeking to use/extend my methods. So I now find myself devoting a significant chunk of time to corpus linguistics. One of the projects I’m involved with is going to analyze word frequency in the field of ‘urban studies’. As part of this work, I’ve scraped about 3000 books from Library Genesis which we’ll eventually analyze. Getting/manipulating the data has been a major challenge. Fortunately I like coding, so I’ve enjoyed the process. If you’re interesting, I’ve got a bunch of new text-analysis code on Github.

On another word-frequency note, I’ve recently been analyzing trends in econospeak. The idea was suggested to me by The Mint Magazine publisher Henry Leveson-Gower. Like many people (myself included) Henry suspected that economics jargon was becoming more common in mainstream English. I’ve now crunched the numbers and the results are surprising. According to my data, over the last three decades economics jargon has become less frequent in mainstream English. This result is not what I expected, and I’m still trying to understand what’s going on. I hope to write a post on it in the next few weeks.

Probability and intuition

One of the things you learn in school is that our intuition about probability is bad. As an example, suppose you flip a coin 5 times and get all heads. Are you more likely to get tails on the next flip? Most of us have a strong intuition that this is so — that tails becomes more likely after a run of heads.

In high school math you learn that this intuition is wrong. That’s because each coin toss is independent from the last. So the fact that you’ve flipped heads many times previously has no bearing on the next flip. The chance of tails is still 50-50. I learned this mathematical fact in high school and since then thought nothing of it.

Then I read an article by Jason Collins called Why Behavioral Economics is Itself Biased. (Hat tip to Evonomics publisher Steve Roth for giving me a heads up about this piece.) Collins shows something astonishing. If we restrict the length of the coin sequence, then tails is more likely after a string of heads, . If you flip a coin three times, for instance, you actually have a 58% chance of getting tails after a head. (If you don’t believe me, read Collins’ article.)

This is such a fascinating result that I had to take a deeper look. I’ve run some simulations and found that up to very long sequences (thousands of flips), tails remains more likely after a string of heads.

Why is this important? Because in the real world, we always look at fairly short sequences of coin tosses (or short sequences of any random event). Over these short sequences, our intuition is actually correct! Tails is more likely after a string of heads. I’m still coming to grips with this result. But I’m convinced that it has important implications for how we understand our mathematical intuition. Stay tuned for a detailed post in the next month.

Post frequency

I try to post on Economics from the Top Down about once a week. But I’ve realized that this schedule isn’t always feasible.

The main issue is that I’m writing about my own research, and the pace of this work is unpredictable. Often what seems like a days worth of empirical work turns into several weeks worth of digging. If I was good at multitasking, I’d be able to write up old research at the same time I did new research. Alas, my brain doesn’t work this way. I can only do one thing at once. That means that when a research project takes longer than expected, I write less frequently.

I’m telling you this because it’s been on the back of my mind since joining Patreon. The quid pro quo of Patreon is that you fund me and I write. So when I don’t write for a period of time, I feel guilty. If I was doing nothing, the guilt would be well earned. But my writing breaks are always to dive into new research. So the next time I don’t post for a few weeks, know that I’m in the middle of a research dive.

Until next time

That’s it for this research update. Let’s hope 2021 is better than 2020.

Cheers,

Blair