Graffiti analysis part 5: University of Chicago

This is the last in a five-part series of posts describing the results of my analysis of my graffiti corpora. I strongly recommend you read “Prelude to a graffiti analysis” first to understand the methodology, data, and sampling. You might also be interested in part 1, Arizona State University; part 2, University of Colorado – Boulder, part 3, University of California at Berkeley, and part 4, Brown University.

FUNDespite the admission office’s recent attempts to make the University of Chicago more welcoming to the well-adjusted and increasingly selective, UChicago is still where fun comes to die. There are students who wouldn’t have it any other way, and others who hate every minute of it. Both groups, at some point or another, end up in the Regenstein Library, where the school motto (Crescat scientia, vita excolatur, ‘May knowledge grow from more to more and so be human life enriched’) is emblazoned near the entrance. The U.S. News & World Report states that the 6-year graduation rate is 91%, and their students’ incoming SAT scores, 25th-75th percentile, are 1370-1560.

This project started when I began documenting the graffiti in the stacks of the Regenstein Library in fall 2007. For two years, the only graffiti I cared about was what I found in “the Reg”, and that shaped my expectations about what university graffiti should look like. Unfortunately, I subsequently discovered that the UChicago corpus really is unique, leading to disappointment whenever I’ve sought out graffiti elsewhere. To put it in terms of an “interestingness” score, UChicago’s unweighted score is 1.8, with a weighted score of 1.85. Gilbert Ryled upTo be fair, this includes data from bathrooms– including the B-level (sub-basement) men’s bathroom, possibly the nerdiest place on earth. None of the other corpora include bathroom data, though I suspect this only helps their score, because bathrooms often drag interestingness scores down. If you exclude the bathrooms, the unweighted score is still 1.74, and the weighted score is 1.79– still .2 higher than the next-highest score, from Brown.

Most interesting categories

This ranking includes the graffiti from the bathrooms. Almost all the B-level men’s bathroom graffiti is a reference; the interestingness score for references, excluding the material from the bathroom, is 2.1


Category Score
Reference 2.45
Religion 2.17
Time 2.09
Meta 2.02
Orthography 1.93

Most common categories

Without the examples from the bathrooms, references make up 5.1% of the corpus.


Category % of graffiti
Quotes 10
Reference 9.5
Sex 4.7
Despair 4.2
Meta 3.7

Quotes and references

When I was only looking at UChicago graffiti, I undervalued graffiti quoting other sources, dismissing them as “typical”. But looking across corpora from different universities, quotes don’t seem to be typical at all– only Brown and UChicago have a high number of quotes; the other three schools rely more heavily on references. Where Brown has about twice as many quotes as references, at UChicago they’re about equal, if you include the B-level men’s bathroom. If you exclude that data, the pattern is more like Brown: about twice as many quotes as references.

The references to “intellectuals” are almost all from the B-level men’s bathroom. There are 20 examples of references to music, broken down into the following genres:

The “other” genres here are R&B, classical, country, hip-hop, reggae, metal, and funk. As for the sources of quotes:

“Scholarship” includes quotes from such works as “Sister Outside: Essays and Speeches” by Audre Lorde, “Science and the Modern World” by Alfred North Whitehead, and Nursing World, vol. 23-24 (1899). The last of these may just be a coincidence; the piece of graffiti in question reads “Nurse, pass the bread.” On the other hand, there might be a really great story behind it. Similar to Brown, about half the quotes are from songs, a total of 76 pieces in the following genres:

Again similar to Brown, rock of various flavors makes up about half the music quotes. At Brown indie is the major sub-category of rock, whereas at UChicago, alternative and punk rock are the largest sub-categories. Beyond that detail, the similarities are striking: rap, pop, punk, and folk are all about equally represented at Brown and UChicago.

Love vs. hate

The UChicago corpus is unique in how many things are hated, and how many things people have mixed feelings about. Like Arizona State, people both love and hate their school, but UChicago students are also conflicted about themselves and “it”. Nine people are loved by name, including Milton Friedman, and “you” appears as the object of love four times. There are fewer sexually-tinged objects of affection than in the Brown corpus, and more references to food. Gen chem is loved but chemistry and biochem are hated. Again conflictingly, both “graffiti” and “when they erase the graffiti” are hated.


I’ve previously written an in-depth examination of homophobia at UChicago, but to provide the most recent data in a form ready for easy comparison with the other corpora, there are 16 pieces of homophobic graffiti (1.1%), when counting each word only once if they’re used multiple times in a single piece. The UChicago corpus uses a greater variety of words than any of the other corpora:

Sexual vs. non-sexual

All in all, UChicago’s sexual word use is fairly middle-of-the-road. In regards to the word “fuck” (with 57 attestations), UChicago uses it more sexually than Brown (14% vs. 9%), but less so than Arizona State (20%) or University of Colorado (24%). For “suck”, UChicago largely aligns with Arizona State with 25% sexual usage– well more than Colorado (12%) and well less than Brown (45%). UChicago, like Brown, has a relatively large number of attestations of “ass” compared to Arizona State and Colorado, but “ass” is only used sexually 27% of the time, compared to Brown’s 53%.

Interestingness by location

Since I had location metadata for the UChicago graffiti, I decided to look at whether any locale had particularly interesting material. The B-level men’s room blew away all the other locations, with an average score of 2.64. Other than that, the other locations (A-level whiteboard, study carrels, walls in the stacks, study desks in the stacks, and other bathrooms) were more or less equal, with the study desks showing slightly lower scores.

Categories over time

One of the most popular metrics in my pseudo-scientific graffiti analysis was the time-based data. In the limited data set I examined, there were some very interesting correlations, where love and despair tended to pattern together, whereas sex reached its one peak in December and declined for the rest of the year. I can’t even recall how exactly I assigned things to “love”, “despair” or “sex” to create that graph, but I re-did it over a longer timespan (fall ’07 – fall ’10, rather than just the 2007-2008 school year), and using the same categorization I used for the rest of the analysis. I didn’t document any graffiti during winter 2009, although there wasn’t much graffiti to document as the walls had recently been painted over.

Sadly, the data is quite scattered, without any clear patterns falling out of it. Category-frequency-over-time analyses, I fear, may be a non-starter.

Interestingness over time / interestingness vs. sample size

Feeling nostalgic for classic pieces of graffiti like “I’m in love and it’s finals week”, I started this analysis convinced that the UChicago graffiti corpus was getting less interesting. However, when I looked at the data, I discovered that it wasn’t the case at all. If anything, the score has been more consistently high over the last school year. Note that the unweighted score is being used here:

One methodological note: there’s 106 pieces of graffiti from the study carrels that were written in either fall 2009 or winter 2010. I took the average of those 106 pieces, and added 53 pieces of graffiti with an average interestingness of 1.58 to the data for each quarter.

Between fall 2007 and fall 2010, the corpus of new graffiti each quarter has fluctuated wildly:

I calculated the Pearson coefficient for interestingness and corpus size, and the result was -.11 — indicating that there is no correlation between the size of the corpus for a given quarter, and how interesting it is. This fact leads me to not be too concerned about differences in sample size between the different university corpora, with the caveat that a minimum threshold (approximately 250 pieces) is met.

See for yourself

The spreadsheets I used to compile the data are available as a Google Doc. If you want to download the data for yourself, just go to File > Download and choose your favorite format. If you do something interesting with the data, I’d love to hear about it (quinn – at – crescatgraffiti – dot – com). There’s no single photo set for the UChicago graffiti on Flickr, but anything in the Crescat Graffiti collection that isn’t labeled with the name of another university or library is from the Reg.

In conclusion

I calculated the standard deviation for the quarter-by-quarter interestingness scores at UofC, with a result of 0.105. The overall unweighted score for Brown, at 1.56, is about 2.5 standard deviations below the UofC average. As far as I’ve seen, the graffiti in the Regenstein Library has no peer.

Happy New Year, everyone.

Add new comment