Main | August 2005 »

In Praise of the Bumps Chart II

Bumpschart2005On the left is my beloved Bumps chart (Cambridge 2005 May Bumps). It has a perfect union of function and form.  Here are some salient features:

  • The horizontal axis records time: the first and second columns of text display the starting and ending orders of the college boats.  The zigzagging lines delineate each boat's movement over the four days of the race.
  • The vertical axis serves dual functions: it both gives the current ranking and maps to the physical location of the boats along the river.
  • What we care about is the movement of a boat over the four days; what we really care about are boats that have moved a lot, either up or down. The chart manages to highlight precisely what we want to see: the larger the movement, the steeper the line, the more attention it gets from our eyes.
  • Focusing on #10 and #11: the criss-crossing lines tell a rich story of tit-for-tat over four days, in which the boats exchanged bumps during the first three days, with the Jesus boat leading after day 4.
  • The story at #1 (Caius) was altogether different: as "Head of the Cam", this strong boat eluded the chasing fleet all four days.
  • My alma mater started and ended at #3 (Trinity Hall)

A truly spectacular chart can be produced by placing all the historical 4-day charts side-by-side, painting a rich history of the rise and fall of different boat clubs over decades.  If anyone has seen such a chart, please send it my way!


In Praise of the Bumps Chart I

In my opinion, the Bumps chart takes the crown for perfect alignment of function and form. In this post, I describe the Bumps boat race and in the next post, I will explain why I like the Bumps chart so very much.

Unlike American colleges, where athletics claim a prominent role and where athletes achieve iconic status, at British universities pretty much the only sport that matters is rowing (called crew in the U.S.).  In particular, the annual Boat Race between Cambridge and Oxford can be compared to the Game between Harvard and Yale. Now, Cambridge and Oxford both consist of more than a dozen colleges, which are autonomous academic and residential units.  Each college has their own boat club and their boaties (rowers) face off at races known as "Bumps".

Ccat_bumps_2The Boat Race is a standard side-by-side race: the first boat to cross the finish line wins.  By contrast, at the Bumps, the competing boats start from a single line; each boat chases the boat ahead by "bumping" or touching before it gets "bumped" by the boat behind.  When a bump occurs, both boats exit the race.

The Bumps happen over four days, with one race on each day.  Thus, after four days, normally a boat can move up or down by up to four slots.   On rare occasions, a boat can "over-bump": this happens if the two boats in front bump, and the said boat then catches up with the boat that has started three slots in front.  Triple over-bumping has also occurred, which means a boat has moved up five slots in one day.

The starting order of one year is the ending order of the previous year, ever since the 1820s (at Cambridge).

Reference: First and Third Boat Club, Trinity College, University of Cambridge

Even a simple chart can be recycled art

Last month, Mayor Bloomberg's office was ecstatic to report a sharp rise in student test scores in NYC.  It didn't take long for the NYT to publish a skeptical piece, the gist of which I have summarized here:

1) Citywide, the gains on this year's standardized reading and math tests were so out-sized - particularly among fifth graders, who improved 19.5 percentage points in reading and 15.2 percentage points in math - that they left some education experts, not to mention Mayor Michael R. Bloomberg's political opponents skeptical. [my italics]

2) But in interviews at P.S. 45 and other schools [...] principals, parents, superintendents, teachers and students offered this most basic explanation: They worked hard.

3) But even with this year's unprecedented increases, only half the city's students were proficient in reading and math.


The NYT article came with a simple chart, shown on the left together with my junkchart version. What did I change?




  • The use of vertical bars obscures the main point of the chart, which is to show the jump in test scores from 2004-2005.  Switching to a line chart remedies that.
  • Foregoing the full vertical scale, I only included the relevant percentages, which again highlights the jump in scores.  Tufte calls the excessive use of grid-lines "chartjunk".
  • Labeling the second series "Reading" instead of the cryptic "English Language Arts" brings it in alignment with the text.
  • Using lines also allows me to stack the math and reading data on top of each other, making clear that students have always scored better on reading than on math.
  • The new subtitle tells the reader how to read the chart, instead of a wordy description of the nature of the data.

Another possibility is to plot the change in percentages from year to year, instead of the absolute percentages, which would serve the main point best.  However, when differences are used, some data are lost, specifically the absolute percentages meeting or exceeding standards.

Reference: "What Lifted Fifth-Grade Scores? Schools Say Lots of Hard Work", New York Times, 6/12/2005.