Thread by @NWPapageorge

My teacher expectations paper was released today as an @nberpubs working paper. Check it out!

nber.org/papers/w25255

It is joint with @SethGershenson and my PhD student Kyung Min Kang.

Before saying anything else: Kyung Min is on the market this year!

As the title implies, we show that teacher expectations matter. In particular, we show that higher expectations from teachers about a student's ultimate educational attainment causally affect students actual attainment. I'd like to highlight a few points @SethGershenson 1/N

There is a clear issue here about causality. If high expectations relate to attainment, it could be due to accurate forecasting (the teacher knows the student will attain a lot) or due to a causal effects, sort of like self-fulfilling prophecies. @SethGershenson 2/N

To estimate causal effects, we exploit how, in the data, 2 teachers evaluate the same student. We exploit teacher disagreements. The idea: one teacher's expectations "controls for" the "accurate forecasts" part and the differences allow us to estimate a causal effect. 3/N

We show evidence that higher teacher expectations in the 10th grade lead to a higher probability of a college degree.

We also study mechanisms. Higher 10th grade TE lead to (in the 12th grade)
1. Higher GPA
2. More homework time
3. Rises in students' own expectations. 4/N

This last point is key. It suggests that higher expectations affect students' outlooks about their own prospects. This supports a story that low educational investments are in part related to information, aspirations and expectations gaps. Terribly inefficient! 5/N

Earlier work shows: black students face an expectations gap. If a black and a white teacher report expectations about a white student, they tend to agree. If a black student has a black & a white teacher, the black teacher has far higher expectations sciencedirect.com/science/articl… 6/N

To think about this, We build a simple model that uses lessons from measurement error. Doing so, we can show the distribution of bias.
We find: teachers tend to be optimistic in general. This helps students! 7/N

BUT: white teachers facing black students are really "accurate", i.e., not optimistic. Here, accuracy, however, is selectively applied to black students in a way that puts them at a disadvantage. They do not enjoy the same optimism as white students. 8/N

We find: optimism comes in the form of "giving students the benefit of the doubt". So, a white student who is on the margin (50-50 shot at completing college) is more likely to face high expectations (he'll make it to college!) vs a similar black student. 9/N

In other words, we find that expectations matter and that teachers are on average optimistic, which helps students. However, black students facing white teachers do not get the same "benefit of the doubt" type of optimism, which exacerbates gaps. Fin.