Actually, other than Northwestern, it is important for almost every school.
Last year, I just did a simple regression analysis for my writing sample, even the competition is not as strong as this year, I was either rejected or waitlisted by every school I applied.
So this year, I forced myself to practice a lot of codings, and self-learnt knowledge of geo-spatial analysis and deep learning, and wrote a very quan writing sample, using thousands lines of python codes and a lot of frontier algorithms, politely asked lots of previous PHD students in computer science, economics, and geography department to point out the mathematic misunderstandings I may initially have in WS, and then self-learnt new knowledge, refine the method and writing, keep doing this for a few months. I have to say it helped a lot, either for a much more thorough understanding of Quan methods, or the final admission results.