Econometrics Memes and Poems, 2020-2021

Non-credit assignments, Williams College, 2021

Memes and Poems by Williams College ECON-255 students, Academic Year 2020-2021

Meme by Margot Berman ‘22

econometrics character alignment chart

Haiku by Sara Sadri ‘23

Selection Bias: Studying a Lot and Getting a Bad Grade Doesn’t Mean that Studying is Correlated with Lower Grades.

Those who study more
might be self-selecting to,
because they know less.

Poem by Emery Zahner ‘22

A Ballad to the Multiple Regression Model

As if a single variable were not enough
Multiple regression came along to add more stuff
When used for causal inference
There is one big difference.

MRM hold all other regressors constant
Alas that there not be any conflict
We still seek the change in Y
Only now we have a new ally

Betas still have a meaning
Though now their interpretations need intervening
Independent variables are the best
Cambridge Econometrics operates in Budapest

The interpretation of the intercept is largely the same
Linear regressions are siblings with a like aim
It simply shows where the population line starts on the Y axis
So there is no need to enter anaphylaxis

X two, X three, X four and more,
Multiple Linear Regression is nothing to abhor!

Poem/adaptation by Tharini Prakash ‘23

The Non-linear Model Not Taken

Two models seemed to diverge in a problem set
And sorry I could not use both
And be one economist, I did not fret
Because when I predict values with the coefficients I get,
The logit and probit models show similar growth

Logit and Probit equally lay
In normal or logistic CDFs from zero to one
And while heavier the logit tail may weigh
I know neither model will lead me astray
So long as the data are non-linear, a reasonable prediction will be done

I shall be telling this without hesitation
Somewhere, after econometrics is done and I’ve done my due diligence:
Two models seemed to diverge in a problem set,
I took the probit model for its easier interpretation
And it has not made much difference

Memes by Jonah Garnick ‘23

econometrics students: ols giving an estimate of the causal effect of x on y. omitted variables bias...

do treatment, do treatment, always takers

running a linear probability model, my model is significant with a p-value of 0.0001, my model predicts negative probabilities for a quarter of my data

spiderman meme: regressing x on z and then regressing y on predicted values of x, dividing your beta from regressing y on z by your beta from regressing x on z