Apr 06, 2009
My favorite Bumps chart in the New York Times ...
For the purist, this is the original rank-based version.
With judicious use of color and background/foreground, this makes for a good story.
The color scheme here, however, is a bit bland. Green for improvement, blue for decline and orange for USA.
Note, for example, New Zealand and England both suffered similar drastic drops as the US.
It would be better to (for example) split out the large improvements and large declines, or to split out the developed world versus the developing world.
This chart is created like this probably because the accompanying piece makes only passing reference to this chart so there is not a clear message to the creator what to do with the data.
Interestingly, there were no ties in 1960 but quite a few ties in 2004. I wonder why. I'd shift the dot to the mid-point between ranks rather than move them up to the higher rank.
All in all, a much more engaging way to present this data than the reams of table found in say the UN World Development Report.
Reference: "Vital Statistics: U.S. Still Struggling With Infant Mortality", New York Times, April 6 2009.
Is it just me, or are rankings of countries regarding infant mortality rates irrelevant?
This chart doesn't tell me the overall shift in rates, nor the difference between the top- and bottom-ranked countries.
Posted by: James Pearce | Apr 07, 2009 at 12:35 AM
I found it annoying it says "Lines show change in rank" then it labels the lines as an increase or decrease in the _rate_. Those aren't the same thing. It's possible for a country's rate to go down, but everyone else decreased more so their rank could rise.
The issue of ties is strange...
Overall, this blog still has me thinking more about using bump charts.
Posted by: 9.2.5 | Apr 07, 2009 at 01:19 AM
I was also confused about the mix up between the change in rank and the change in rates. That is a pretty stupid mistake. But also the place where the labels are set is very poorly chosen. The label "lower infant death rates" is on the line for Sweden. It made me believe Sweden has now a lower infant death rate, meaning it started at the highest possible death rate. At the bottom, Chile has a line marked "higher infant death rates", and the line is going up. This matches that first wrong impression I got, and it was giving me data that didn't at all match my preconceptions about health care in these countries...
Posted by: Cris | Apr 07, 2009 at 03:35 AM
I agree with the above - the ranking is not so interesting if there has been a major compression on the rates, it just means that from one year to another the rankings will change a lot. Also Cris's point is a very good one.
Posted by: Matthew | Apr 07, 2009 at 11:32 AM
"Interestingly, there were no ties in 1960 but quite a few ties in 2004. I wonder why."
I would hazard a guess that it's because the distance between to highest and the lowest is smaller so there's a greater chance of collision.
As the previous commenters have pointed out the chart doesn't say anything about actual rates; it's entirely possible that all the countries in the 2004 list have lower infant mortality than Sweden did in 1960 but there's no way to know.
I suggest that a simple table of the 2004 list with colour coded +/- values would be a more space efficient and generally accesible way to show the key trends in the piece.
I think your love of bump charts might be clouding your usually excellent judgment here.
Posted by: Tom P | Apr 07, 2009 at 11:41 AM
The chart the NY Times labeled "Infant Mortality Rates World Wide" seems to be missing China, India, most of South America, and all of Africa.
The chart itself masks that infant mortality has dropped in all these nations.
For scaled look at this data go to Gapminder.com and click Gapminder World link.
On the chart's vertical axis select "Infant Mortality Rate (per 1,000 births)" and after the data loads, chance the scale from logarithmic ("log") to linear ("lin").
For the horizontal axis select "For advanced users"->"Years (use as color)"->"Year categorization (1950-)"
In the list of countries check Chile, Romania, Singapore, and Sweden (the top and bottom ranks ranks in the NY Times graph) to show a value trail.
(or just goto http://tinyurl.com/GapminderSample)
Press play to see the data history.
Posted by: Dave | Apr 07, 2009 at 05:06 PM
Great comments as usual.
The rate versus rank issue is definitely important, and the graph is mislabeled. We can solve this issue if we use a rate scale instead of the rank scale. That would be a bumps chart with a continuous scale.
I would argue on behalf of the graph designer that ranks do provide some useful information. Consider the Netherlands. It is quite possible that the actual mortality rate decreased from 1960 to 2004, and it is quite possible that the difference in rates between it and the countries ranked above it is small. However, the fact that about 10 countries overtook it is still salient.
Incidentally the Bumps boat race is all about relative ranks, not so much speeds so the Bumps chart is perfect.
Tom P: Good observation about why there are more ties in 2004. I'd add that if true, it would be an artifact of the rounding of the decimals.
Posted by: Kaiser | Apr 07, 2009 at 10:55 PM
An artifact or maybe an intentional decision on a difference of no practical importance maybe.
Posted by: James Pearce | Apr 08, 2009 at 09:21 PM
I made rates chart, not rank compare.
Posted by: muE | Apr 08, 2009 at 11:19 PM
Despite the obvious flaws in the chart, I note one interesting correlation: during the period covered by the chart (1960 - 2004) the percentage of population housed in "social housing" (or what we call "public housing" in the U.S.) in the two countries that ended up topping the chart in 2004 went from roughly 0% in 1960 to about 84% (Singapore) and about 60% (Hong Kong). Interesting.
Posted by: Brodie Hefner | Apr 29, 2009 at 04:50 PM