2013 Budget Proposal

Normally, I look forward to the release of the president’s budget proposed budget—fully understanding that it will never pass as proposed. We get to see lots of visualisations trying to show that we really do spend quite a lot on defence. And an awful lot on Social Security, Medicare, and Medicaid. And a little bit on a lot of other varied programs and departments.

Last year was a very nice tree map by the New York Times, see my post about it here to refresh your memory. This year’s, well, frankly, is not so nice. To be fair, the piece is aesthetically pleasing and well designed; the transitions and interactions are all spot on.

What is not so much is the use of circles and bubbles.

In the tree map of last year, all the various leaves fit nicely against each other inside branches as part of the tree. See the below screenshot for a reminder. There were small spaces between the branches and leaves, but no more than necessary. Does the overall shape or size of the tree map represent anything? No, but note how the leaves are grouped by branches. And how, in a pinch, you can compare vertical and horizontal axes of each cell against is neighbours to gain a size comparison.

The 2012 budget proposal tree map
The 2012 budget proposal tree map

This year’s overall spending graphic shows large gaps between some circles and overlap of others. It is difficult to compare circle to circle and thus gain any true meaning of the size differences between programmes. Furthermore, the spaces do not group like with like, in fact every time I reload that view, the circles are in a new arrangement, making it difficult to return to the programme I had just been viewing. Compare that to the tree map where everything is ordered by department and, because all the changes and filtering happen within the view, the cells remain in place.

All spending in 2013 proposal
All spending in 2013 proposal

This year’s budget proposal has an additional three views presented: types of spending, changes, and department totals. The first moves the circles into two camps: discretionary and mandatory spending. But, the areas of the circles are hard to compare against each other, and the placement of the circles seems arbitrary. Compare that to last year’s which highlighted the types of spending within the tree map and blanking out the other. The cells remained in place and by their positioning against each other, a more accurate sense of scale and relationship was created.

Changes sorts the circles into department, though that part is not entirely clear at first glance. Otherwise, this view makes sense, though I wonder if a more clean scatter plot could not be more useful in plotting size and growth on the x and y axes with colour remaining the change from the previous year. Though one loses the grouping by government department, such a grouping seems less important throughout the 2013 piece except in the by department view.

Sorting by change in spending
Sorting by change in spending

That view resorts the circles into a matrix with each department receiving a square-like cell into which its circles are dropped. This was handled much more adeptly and clearly by the tree map of last year.

I appreciate the need to create new and more interesting visualisations every year. But, whereas last year’s was a solid piece, this is a shaky step backward. I would have liked to see a more nuanced and featured improvement to last year’s tree map instead of throwing it out.

Credit for the piece goes to Shan Carter.

Adding Guided Context to Maps

Maps are cool. They show the geographic distribution of data. And that is fantastic if there is a story in said distribution. But even if there is a story, sometimes given both the scale of the map and the amount of data encoded in the map, how could you possibly expect to find the story? Which little region of the map do you search to find the interesting nuggets?

On Sunday, the New York Times published an interesting solution to that very quandary. The context is an article looking at the anger and resentment felt by some towards government assistance via the social safety net, and yet how these very same people depend upon that safety net through programmes like Social Security, Medicare, Medicaid, &c. The map, a choropleth, examines several different metrics that comprise government assistance, e.g. Medicaid payments as a percentage of income.

Government benefits overview
Government benefits overview

One can easily toggle through the various metrics at the scale of the entire United States. This is a rather standard feature for such maps. However, in the upper-left corner, the designers placed a ‘guide’ that provides context and stories for each metric. But, not only does the guide provide text to support the map, but it zooms in on specific areas and regions that then support the text and best exemplify the point.

Here we see the map of the whole US for Medicaid, which appears to be scattered pockets of higher percentages. Interesting perhaps, but the user likely has few ideas as to what that visualisation actually means.

Medicaid's broad overview
Medicaid's broad overview

Compare that to the guide’s view of the map, which focuses on the large cities on the East Coast.

Medicaid's eastern core context
Medicaid's eastern core context

Providing context and guiding a reader/user through the stories contained in the map, or at least those deemed interesting by the designers and editors, is an interesting solution to the problem of finding the story in maps such as these. However, by moving away from a strict visualisation of the data, the New York Times and others that try similar avenues introduce human biases in the story-telling that may otherwise be unwanted or distracting.

Credit for the piece goes to Jeremy White, Robert Gebeloff, Ford Fessenden, Archie Tse and Alan McLean.

Circle Charts ca. 1937

Another image from my 1930s algebra book is on pie charts, or what was then called circle charts. And while the utility of such a chart form has not changed, especially in these examples, the circle chart of the 1930s had one particular good use for students. Constructing it.

Circle chart construction
Circle chart construction

Today a student plugs in numbers into a spreadsheet in Google Docs, Excel, or Open Office. He or she presses a button and the circle chart is done. Back in the 1930s, students needed to convert absolutes to percentages and then use protractors to draw the slices on pieces of paper. Fancy that, students having to do math to make a chart.

How to Use Good Data Visualisation in Your Private Life

Often we think of graphs, charts, and other forms of data visualisation as a means to exploring the economic growth of so and so, or visualising traffic patterns, of explaining the complexities of science, or the reporting of yesterday’s news. But, we can all use data visualisation in our own lives to help make better decisions.

While I normally opt not to post links to other data visualisation blogs—I figure most people are also already checking those out—Nathan Yau posted about why he wants to cut the cable, i.e. lose his cable television subscription. He has two separate charts that are simple but effective in driving home the point that he really ought to think about cutting cable out.

Cutting cable
Cutting cable

The article, while a bit longer than usual, is well worth the read. The charts with the explanation make for a compelling argument.

Comparing the (Display of) Results in Colorado

Mitt Romney lost badly last night. No way around that. But as I watched the results come in through various sources, I noticed two interesting design decisions that made me think; one from the Guardian (the British perspective), and the other from the New York Times.

Using only Colorado as an example, here is the map of county results by the Guardian.

Results map from the Guardian
Results map from the Guardian

Note how the map is presented in 3-D that therefore allows the use of height as another encoded variable, in this case the size of the lead. Now compare that to the map used by the New York Times.

Results map from the New York Times
Results map from the New York Times

Note how this map is flat. So much less cool, right? But try to compare the results in Denver County. When I look at the Times’ map, I see blue; Mitt Romney won. When I look at the Guardian’s map, I see…actually, I can’t. That label is in the way. And then even when I begin to interact with the map, Denver County is hidden by the height of Arapahoe County.

But what about the size of the lead? I cannot see that encoded in the New York Times map. No, one cannot. However, they added a toggle function to change the data displayed on the map—though the utility of that view can be left for another discussion.

And now to a minor point about comparing the totals.

Again, a look at the Guardian’s presentation.

Results table from the Guardian
Results table from the Guardian

And now the New York Times. Numbers are numbers and faces are faces. But look at the graphic element representing the percentage. With the Guardian, I can just barely discern that the size of the circles for Santorum and Romney are not the same. And the same goes for Gingrich and Paul. But when I look at the Times’ presentation, I see a simple bar chart that more clearly shows the relationships between the results.

Results table from the New York Times
Results table from the New York Times

So interesting design decisions lead to one view that I find far more successful in showing the data: the New York Times.

Housing Prices Fall Some More

Houses are meant to be lived in. Which is good to know if you’re a real estate investor because the housing market in the US is still not so good. According to an article in the New York Times, we’re back to 2003 levels (on average of course) for single-family homes.

Accompanying the article is an interactive chart that lets users view the full breadth of the survey while highlighting specific markets of interest and showing actual values along the length of the chart.

The Case-Shiller Index for Housing Prices
The Case-Shiller Index for Housing Prices

Credit for the piece goes to Kevin Quealy and Jeremy White.

Florida Primary

The Republican primaries…they’re still going on…on the long inevitable road to Romney’s coronation. Next up is Florida, always an interesting state to watch. There are a lot of people there with a whole host of interesting demographic slices. Perhaps one of the most interesting ones, at least to the media, is the Hispanic vote. Other things to look at in Florida include the burst housing bubble and rather high unemployment.

The New York Times published a graphic with a few maps and charts trying to paint the landscape of the Florida primary battle. These two selections below show which Republican primary candidates won which counties in 2008 as well as the size of the Hispanic population registered Republican.

Florida primary landscape
Florida primary landscape

Credit for the piece goes to Haeyoun Park.

Show Europe the Money

The Guardian has an interactive piece that details payments to and from European Union member states to institutions, determining whether each state is a giver or receiver.

Comparing payments and receipts in EU member states
Comparing payments and receipts in EU member states

The concept sounds all well and good. However, the piece itself feels clumsy with too much scrolling and whipping about to pan across the whole EU. The charts look a tad heavy—which could have been remedied for a more concise piece—and the callouts beg for a level of interactivity that is otherwise lacking.

Lastly, I have concerns about the list of countries at the top, although these may stem only from the point of view of an American not too familiar with Europe. Flags are not circles, they are, in most cases, rectangular in shape. Does cropping a symbol or icon of a country make it more or less useful of a symbol or icon? Furthermore, do the British recognise the flags of their fellow EU member states?

The country icons/flags call for some type of sorting function, to compare payments and receipts and their balance. But, instead, they sit there in unalterable silence, providing only an economic overview when clicked. An overview that through its staid design feels more like an afterthought.

The Finances of Social Security

Simple graphs can tell great stories with little annotations. This graphic by the New York Times illustrates that point well with a stacked line chart set behind a line on the same scale. The two should match, or at least the red should be beneath the greys. When they don’t, you have a story and the Times calls it out.

Social Security Revenue vs. Expenditure
Social Security Revenue vs. Expenditure

Killed in Action in Iraq, State by State

The Iraq War is over. And now it is time to reflect on what we have gained and what we have lost. This map by the Guardian details the number of soldiers killed in action in Iraq. (Other options include total wounded, killed by non-hostile, &c.)

Killed in Action in Iraq by State
Killed in Action in Iraq by State

Unfortunately, I call it a ‘no kidding’ type of map. The data, accessible via the Guardian here, corresponds nicely with a list of states by total population. Of the top ten countries in KIA, only Virginia is not among the top ten in population; it is 12th. The country thus not in the top ten in KIA, but in population is North Carolina. It’s rank in terms of KIA? 11th.

The data is interesting and worth depicting if we are to reflect. But, perhaps a more suitable visualisation could have been chosen.

On a personal note, these Google Maps overlays are annoying when, in the cases of, e.g., Wisconsin and Massachusetts, the shapes are incorrect. Perhaps coastlines are not as easy as states with ‘straight lines’ for boders, but we would do well to try and make irregular coasts at least somewhat correct.