Misrepresenting with Maps

An Atlanta Braves map-product showing individual tickets sold in 2012. Shows concentration in north Atlanta and Northern Suburbs.

An Atlanta Braves map-product showing individual tickets sold in 2012.

The Atlanta Braves are moving to Cobb County, away from their long-time residence in Downtown Atlanta. There are many reasons given for this move, but one of them is shown in the picture which is the concentration of ticket sales to the north of Atlanta. The new stadium location is approximately central on this map, which their website describes as “…situated in the heart of Braves Country. Each red dot on this map represents a ticket sold to a Braves game in 2012.”

On the face of it, this seems like a reasonable conclusion. After all, just look at that map and all the red dots to the north of Atlanta. Seems obvious.

But is it?

There’s a book out there for anyone to read called “How to Lie with Maps.” It’s a very good primer on the ways geographic information can be misrepresented or represented poorly. I’m curious if the GIS (Geographical Information Systems) specialist who put this map together made any further inquiries into the data, or if this was the sum total of it. For example, I would ask these questions:

  • How many dots show repeat purchases from the same address? I’d like to see a heat map showing frequency of purchase and how those data are geographically located.1 Did someone in Dunwoody buy a dozen tickets to one game, but somebody down in Grant Park get twelve individual tickets to twelve games? Which one of those has a greater impact on the Braves’ bottom line?
  • How many dots are private residences vs. business?
  • How does the average value of a ticket get reflected geographically? How about some metric which looks at average household expenditure (8 tickets at field level vs. 1 ticket in the nosebleeds)?
  • The most important question that’s not addressed is comparing those dots to the density of the population in the area. For a good lampoon of the problem not looking at population density causes, see this xkcd comic.

That last point is the one I think is the most interesting. But there are also some artifacts in that graphic the Braves produced that are worth pointing out (Click through to the larger version for better clarity). Note how the map is broke out by zip code. It appears to me that the “address points” aren’t actual addresses, but points randomly placed within a zip code block. This is obvious on the southwest side where densely packed zip codes are immediately next to nearly empty ones. There are also locations where, if you happen to know the area, you know there aren’t any residences present because it’s undeveloped. A good example of this is the northwest portion of the block containing Peachtree City (Big Dense Red Blotch south of the city; zip code 30269).

What does all this mean? Maybe nothing. Maybe everything. The point is to watch out for these sorts of mapping shenanigans/errors/simplifications. I try to be an informed information consumer. It’s best if we’re all an informed populace. Even when it comes to sports teams moving to the money.

Answering these types of questions is the most important part of a GIS graphic. Blindly throwing up a frequency map doesn’t tell nearly the whole story. ((Note that I don’t necessarily think that producing map products to answer my questions would shift the new stadium centroid back toward downtown. In fact, I would bet money that if you did the metric of average expenditure by address, it would shift further to the north. But that’s just my butt talking. Somebody would have to run the numbers.)

  1. By the way. The name of that image on the Braves site is “heat-map.jpg” but it’s not a heat map. Just locating a bunch of dots is a frequency map. []
This entry was posted in Uncategorized. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>