Our Associate Professor Dr. Douglas Webber recently has been getting attraction in his Twitter stream due to his Lego models showing econometric concepts. Kaitlin Flynn interviewed him about it. Enjoy reading the story:
One of his most popular tweets, receiving over 1000 likes, displays the Simpson’s Paradox. The Simpson’s Paradox, as described by Wikipedia is “a phenomenon in probability and statistics, in which a trend appears in several different groups of data but disappears or reverses when these groups are combined.” (“Simpson’s Paradox,” n.d.)
In the tweet shown below, Dr. Webber displays the Average vs. Local Treatment Effect. The purple pipe-cleaner on this model represents the whole population, while the yellow pipe-cleaner only represents a small portion of the population.
Dr. Webber shows “the importance of having a (good) comparison group,” with these models. This comparison is called Differences in Differences regression. A treatment group and a comparison group are observed, and this can be used to, for example, find the effect of a policy you want to implement. To clarify the image more, Olena Berchuk, a student of Dr. Webber, included an explanation:
“So looking in the picture, the blue set of Legos is a good comparison for the red set because the pipe cleaner slopes before the policy change (the vertical, blue Lego line) are the same. The orange pipe cleaner in that picture shows the difference in the outcome due to the policy change, while the green one is the overall difference (including the trend). But if you look at the last picture, the one with the yellow Legos as a comparison group, you can see that the slopes before the blue policy change line are not the same, which suggests that this is a really bad comparison group (and it is, because it actually causes us to underestimate the true effect of the policy — notice how the green pipe cleaner is really small as compared to the orange pipe cleaner in the previous photo with the other comparison group).”
Dr. Webber is also using these models in his classes for his students to get a better understanding for these econometric concepts.