Making of a double page spread containing statistics on art exhibitions

In February 2004 Roman Voronezhsky and Artem Gorbunov, designers of our studio, in cooperation with Mir magazine staff worked on a double page spread containing world statistics on attendance of art exhibitions in 2003.
Narrated by Artem Gorbunov.

Initial visitor figures for 233 exhibitions were provided by 96 museums located in 61 cities around the world, and published in The Art Newspaper.

This information was arranged in a table:

There also were exhibition top tens for the following genres: Old Masters, Contemporary, Decorative Arts, Modern and Antiquities.

The author accepts full responsibility for all assumptions and conclusions he built upon these data.

Hypothesis

Amusing statistics is the kind of information you would like to learn, analyze, and compare. Unfortunately, most designers who deal with statistics do not bother to draw the meaning from the figures they have to present. Instead, they get busy trying to make them look pretty.

In fact, the information is intrinsically attractive. The designer should determine what in particular may present interest among the data he has, and then each milligram of printing ink and each solid pixel should be devoted to this purpose. (Ed. note: See Edward Tufte’s works.)

So, our work on the double spread sheet began with data analysis. We transferred the initial table into Microsoft Excel format and applied some transformations to it. This software allows playing with data, grouping them, constructing aggregates, and visualizing relationships.

By the way, at this stage you can judge the quality of the initial data. For example, we were surprised to see that figures provided by the Hermitage were rounded to 1,000 or even to 10,000. All the rest museums counted every single visitor. Exhibition popularity rate expressed by daily averages was calculated on this basis.

In order to present this information in the best way, we needed some idea or hypothesis that could be somehow illustrated or tested. We could suggest, for instance, that there is correlation between exhibition duration and its popularity. Lets make a diagram based on the Excel table showing joint distribution of these two quantities.

Voronezhsky-Gorbunov hypothesis: Short exhibitions attract more visitors a day. Constant exhibitions are less popular.

The diagram proves right our humble guess and excellently demonstrates that the public flocks to occasional exhibitions with short duration and strong media support.

What’s more, we can clearly see the points (circled) that appear to be exceptions from the rule—they show high attendance figures for exhibitions several months long.

Firstly, there are three are Guggenheim Museum exhibitions in New York and Bilbao. The Guggenheims attract a lot of tourists because of their original architecture and extraordinary approach to display. Secondly, it’s the Michelangelo exhibition in Florence (and, as we’ll see further, old masters exhibitions enjoy the highest popularity).

Ranked first as to daily attendance is the exhibition called “Leonardo da Vinci, master draftsman” hosted by New York Metropolitan Museum of Art. “Peter the Great” of the Russian State Hermitage, an exhibition devoted to the 300th birthday of St. Petersburg, took third place.

Method

Default settings in Excel produce less-than-perfect diagrams that can only be used on early stages.

Lets use the above diagram as a sample to show how just about any chart can be improved in a few simple steps.

Final diagram. Outstanding exhibitions are signed.

Comparative analysis

Besides main list of the exhibitions we had five separate ratings—top tens by genres.

Publishing just five tables is not a good idea. It would be better to compare all five ratings on a single diagram:

Attendance figures by genres

Average daily attendance is shown on the left of the listed exhibitions. The vertical axes equals in length to the range of visitor figures in all five ratings. Marked in red is the St. Petersburg successful exhibition.

General population

From the very beginning we meant to publish a detailed table with names of all exhibitions arranged in descending order of daily attendance figures. It helps to increase credibility of the diagrams we publish—a reader can check everything for himself. When working with statistics, it is important that illustrations (and the conclusions drawn) are credible.

Table layout is another issue. With such big amount of data the structure has to be optimal. Below you can see how introducing a small change enhances readability of the table.

On the left: before, on the right: after (in Russian). Opening and closing dates are put in two separate columns. All exhibitions by defenition took place in 2003, so the dates do not show the year unless an exhibition opened the previous year

Model

We have prepared the diagrams and the table. Lets bring them together onto the double-page spread leaving some space for commentary.

The reader shall perceive information on this double page spread on macro and microlevels. From the macrolevel point of view he sees the whole thing at once with all parts and connections between them. This includes large headings and short notes explaining the information in the table and on the diagrams.

If the reader looks closer at separate lines in the table or points on the diagram, he moves to microlevel. Studying the two pages he is likely to keep switching from macro to microlevel and back. But since all the information is well before his eyes, it should present no difficulty.

Also, we’ve connected the points in the diagram to corresponding table lines with hairlines (0.15 pt) to reinforce the logical connection. The second diagram shows daily attendance figures, so finding any of the listed exhibitions over in the table won’t be a problem. These things can as well earn the reader’s confidence.