Part of my work I’m interested in the correlates of crime at very small places, particularly aspects of the built environment. Part of the difficulty of this work though is that some aspects of the built environment change very slowly. I often just anecdotally give bars as an example – when a bar goes under it often just gets replaced by another bar. So for example if I want to make an estimate of how much crime would decrease if you took a bar away, it is difficult looking at historical data because most of the time when a bar goes away it is just replaced by another in a short time span.

But admittedly this perception was just based on my anecdotal experiences. So when I saw some historical maps John Krygier posted of saloons I wanted to put a pretty strict test to my assertion. Here is a map of saloons in Buffalo (circa 1901 on John’s website):

I grabbed the current locations of places licensed to sell alcohol in New York State via the open data portal and geocoded those in Buffalo. (This includes things like grocery stores as well as bars.) I did a mediocre job trying to digitize the old map (here is the digitized image), and here we can see the overlap between the current and the historical locations. Zoom into the area with the blue icons to see the historical locations.

So we can see that my baseline of bars not changing is not accurate for this for this extreme comparison. If you zoom out you can see that there is a higher concentration of bars just to the west, so I wonder if over time there was a shift of these bar locations.

John has some more examples of historical saloon maps in Baltimore plus San Francisco and New York City (in the same post with Buffalo). I’d be interested to see those locations as well if someone takes the time to replicate this.

I may have to think more seriously about evaluating the effect of bars over time, and seeing if things like bars losing their licenses because of violations result in crime decreases.