An Amazing Graphic on Box Office Receipts

It's a rare graphic that gives even Gawker pause. But one that ran in the NYT on Sunday visualizing movie box office receipts between 1986-2007 has created a buzz among the admittedly geeky world of graphic specialists. (See here, here, here, here, here, and here.)

The real visual strength of the graphic -- for me -- is the un-Tufteian use of curves, which give it a very organic feel and actually make you internalize the 'ebbs and flows' the graphic presents.

To find out more about how it came together, I asked Amanda Cox, the lead graphic editor on the piece, for a quick email Q&A. Unlike most graphics designers, Cox has a master's degree in statistics (from the University of Washington) and worked at the Federal Reserve for two years.

Q. Where did the idea come from? Did it go through other iterations that looked nothing like this?

I had a few ideas left over from Oscars 2007, when we ended up making a social-networky graphic. One of them is below. But I met Lee Byron - an undergraduate at Carnegie Mellon - a few weeks ago, and he showed me work that he had done using this form to view his Last.fm music listening history. It was clear then that this might be a fun way to go.

(Editor's Note: Here is a very early iteration of the box office graphic comparing Juno to other romantic comedies:

(click to enlarge)

and here is Lee's Last.fm graphic:

Q. How long did it take you guys to put that together? It seems like a Herculean task.

Except Lee, everyone involved has been busy working on politics, so it was a three-day project. I started the print version on Tuesday afternoon (By then, Lee had sent me code that was generating the chart). It was finished Friday morning (But I also made a completely unrelated print graphic for Thursday's paper). We enlisted Shan Carter and Mathew Bloch to help with the interactive version Wednesday night around 6 pm, once it was clear that it had potential. We posted that version Saturday and tweaked things like label placement on Sunday.

Q. Was there a big difference between doing a print graphic and the web graphic?

It's pretty much like starting over. Starting over with a good idea, but like starting over.

Q. What in my opinion really makes it jump off the page/screen is the curvature throughout the piece. Can you talk about some of the design choices?

I think that distributions are always more interesting that averages (that's really the only secret to anything interesting I ever do). But you're right that curves are what makes it pretty.

Q. The only criticism that I've heard about the graphic is that since the baseline is not tied to some fixed x-axis, it makes it a little difficult to compare some of the movies throughout time.

Matthew actually made a version for people who care deeply about a fixed x-axis. We rejected it because it didn't do a good job of answering some of the questions that I thought were the most interesting (about peaks and ebbs of individual films). As it is, it's a lot of fun to look at films like Top Gun and Ghost. In a more traditional chart (like a stacked bar graph), it probably wouldn't be, unless you were willing to lose the shape of every other film. Different forms do better jobs at answering different types of questions. This form attempts to distort the shape of each film as little as possible; it works well for some types of questions; for others, not so much.

(Editor's Note: Below is the fixed x-axis version. Click to enlarge.)

Comments

[wwwenger:]

Just to be sure I understand how this graphic works --
1) the height of the "mountain range" in the fixed x-axis version (and the total vertical spread in the other version) indicates the total size of the movie-going audience at a particular point in time? Is this only for first-run films, or for all movies in theaters across the country, including revival showings at repertory houses?
2) when a new film opens its graphic area starts at the apex (in the fixed x-axis version) of the areas already claimed by previously opened films? That is, we're not dealing with overlaps here -- this is a two-dimensional image, not an image in which "National Treasure" rears up majestically from the same baseline as all the other films and towers incredibly high above them. The suggestion of a mountain range and the existence of convex curves at the top of the range incline me to see the area for a film as continuing behind the films below it in the graph, but I assume this is just because of a bias toward interpreting images as representations of three-dimensional realities. Pretty, but confusing.
3) in the variable x-axis chart, why was the shape not symmetrical horizontally? Because of the introduction of some films above and some below the center (was this done on the basis of date, or of filling in where available space was growing most quickly, or what?), the tailing off of the films toward the right is not going to provide a fixed baseline in any case, so what are the reasons for the high and low points for a given date not being in symmetry? The fine effect of the graphic suggests that it was created this way for very definite reasons, but that leaves me feeling inadequate if I cannot determine what the reason was and therefore what the chart is intended to convey. Either that, or the practical purposes of the chart have been shunted aside in favor of its appealing style.
On the whole, though, I find this approach a positive development and will be interested in seeing how it's picked up and refined in the future.

2008 Feb 27 02:11 PM