National Geographic Society
P.O. Box 98199
Washington, DC 20090-8199
800-647-5463
Lat/Lon: 38.90531943278526, -77.0376992225647
Inspiring people to care about the planet since 1888
RSS Syndication
Designing a graphic is like writing a story. You can't include all your material, nor can you present it with uniform emphasis. To engage readers, you have to selectively edit and then order your information into a narrative. In other words, what is most important for people to see and in what order?
Not every reader will agree with our choices. Our health care graphic from the January issue (left, click to enlarge) has provoked a healthy debate around the blogosphere. Some people love it; others loathe it. The issue isn't just premiums or public options. Many bloggers are talking about our process.
At least one blogger said we should have used a graph known as a scatter plot. A scatter plot lets you visualize the relationship between multiple variables. In the case of the health care graphic, our variables are spending, life expectancy, and doctor visits. The scatter-plot version (below) plots spending along the x axis and life expectancy along the y, and it uses dot size to show the doctor visits.
(Click graphic to enlarge it in a new window.)
There's nothing wrong with a scatter-plot solution. It just emphasizes different points—a different story. Here's why I didn't go this route.
1) Many people have difficulty reading scatter plots. When we produce graphics for our magazine, we consider a wide audience, many of whose members are not versed in visualization techniques. For most people, it's considerably easier to understand an upward or downward line than relative spatial positioning.
2) This graphic was designed for our vertical printed page (top). Scatter plots tend to work better as horizontals. Without this context, I can see why online viewers would question the graphic's shape.
3) Any data visualization is going to put certain information in the foreground and other info in the background. The scatter plot nicely organizes the countries into three clusters, but the varying dot size puts too much emphasis on the doctor-visit data, which should be a tertiary point. On the other hand, our parallel plot (which I anchored around the spending and life-expectancy averages) calls the reader's attention to discrepancies between spending and life expectancy—the story I wanted to tell.
I applaud people for questioning our approach; in fact, I love this kind of dialogue—especially about art and design. Whether we reach consensus, well ... that's a different story.
Subscribe to RSS feed
My Shot
Daily Dozen
Voting Machine
Puzzles
Desktop Wallpaper
Buy NG Photos
Special Issues
Comments
Jan 13, 2010 4PM #
Really interesting. The scatterplot made the spending:life expectancies more clear to me but the parallel plot made the spending per person jump out in an amazing way.
I didn't see Taiwan on this list of countries (?). I'm curious about Taiwan because they modified their system somewhat recently and they did it by analyzing the pro's/con's of established universal health systems, an approach that would have been useful for the US debate.
Jan 13, 2010 4PM #
Aside from the content of the article, which is such a hot issue, the question of graphical representation of data interests me very much, because, as a scientist, I have to do this on a daily basis.
I think your points are very good, and I thank you very much for taking the time to explain this is such detail, especially the part with the foreground and background information. I'll be a lot more careful with this issue from now on thanks to you.
Unfortunately, I have to admit that it took me a lot longer to find my way through the original plot--a few minutes, while a quick glance at the scatter plot gave me a clear picture. But I guess people are different. Maybe this could/should also be quantized somehow.
Very interesting subject.
Jan 13, 2010 4PM #
There is one major component that is not illustrated in both examples and that is lifestyle. The American lifestyle of over eating, poor diet and lack of exercise contributes to the decline in the average life span vis-à-vis obesity, diabeties, heart disease, ad nauseam.
Jan 13, 2010 4PM #
I would like to see the countries that have universal healthcare through private insurers (like Switzerland) differentated from those that have universal healthcare through public providers (UK, Canada). This could easily be done by using a third color, but I assume that it's more complicated due to some contries having systems that are a mix of public and private insurers.
Jan 13, 2010 4PM #
I was in the "loathe it" category, partially because of the design, which highlights so many of the flawed assumptions in the health care debate.
We should all beware of charts promising simple explanations for health care costs.
More here: http://bit.ly/info/8fdZfc
Cheers,
Evan Falchuk
Jan 13, 2010 4PM #
Potentially very misleading scatter chart. Are the costs only the out of pocket at the doctor office? Do they include the added tax burden for the universals? It's all in us dollars but how about as a percentage of average per capita income. Numbers and political topics are nothing but propaganda. Hopefully the advances in medicine over the next 20 years will make the lifespan numbers irrelevant.
Jan 13, 2010 4PM #
So far, everyone I know who looked at the graph thought it was obtuse. But I socialize with college educated, mostly in the technical fields.
It would be more interesting to me to see the size based on total number of people covered. The number of doctor visits seems irrelevant. Given the scatter plot, number of doctor visits doesn't correlate with cost or lifespan. The only thing it demonstrates is that the two non-universal coverage countries have less doctor visits. Of course that would be the case.
The US healthcare system despite having 50M people effectively uninsured has over 250M people insured, making a larger healthcare system than any other country, by far. I wonder if the spend correlates to quantity of people covered.
The problems of an effective system at such a scale may be unique, which may make effective solutions equally unique. Just because Australian style universal coverage works for a country of 21M may not mean it will work in a country of 315M.
(I have healthcare in the the US and AU; AU's is better and an order of magnitude cheaper, so i'm not arguing that the US is good. I'm just not happy with the data presentation. It seems weak.)
Jan 13, 2010 4PM #
I am glad this was redesigned. I though this graph was so bad that I posted it as a discussion topic for my graduate level Information Design class. Despite varying political beliefs, the class thought this graphic was executed very poorly. Here is a portion of the summary of the discussion.
Arrangement is confusing at best. Very good emphasis, graphic elements lead the audience’s eyes the exact place the author intended. This graphic has very poor clarity and very poor conciseness. The tone of this graph is meant to be one of fear and urgency, but it is not executed well. The confusing design and excess information detract from the tone. The ethos of this chart is the biggest issue. Because of manipulative information, and lack of quality cognate strategy (except for emphasis) this chart and the author develop little to no ethos.
Jan 13, 2010 4PM #
this is an interesting topic. A scatter plot is used when a variable exists that is under the control of the experimenter.
Jan 13, 2010 4PM #
I thought your Health-care graphic was terrific! I had to get on the WEB to see if I could get a copy so I could show all my friends how our US Health care costs are ridiculous; particularly in contrast to other countries. Of course it was clear what potential universal health care could do for the US.
Jan 13, 2010 4PM #
This article is interesting. Scatter plots are difficult for most of the people. This graphic designing is good!
Jan 13, 2010 4PM #
that's really a fantastic post ! added to my favourite blogs list.. I have been reading your blog last couple of weeks and enjoy every bit. Thanks.
Post a Comment