Yesterday I looked at the aboriginal Canadian identity infographic and wondered if bubbles in a bubble suffice for understanding size and relationship. Today we look at an interactive graphic from the Los Angeles Times where I do not think the bubbles suffice.
California Budget 2013–4
In this graphic, I cannot say the bubbles work. Besides the usual difficulty in comparing the sizes of bubbles, too many of the bubbles are spaced too far apart. These white gaps make it even more difficult to compare the bubbles. Furthermore, as you will see in a moment, it is difficult to see which programmes receive more than others because there is no ranking order to the bubbles.
Below is a quick data sketch of the state funds only data for 2013 and 2012.
California Budget 2013–14
While I did not spend a lot of time on it, you can clearly see how simply switching to a bar chart allows the user to see the rank of programmes by state funding. It is not a stretch to add some kind of toggle function as in the original. One of the tricky parts is the percent growth. You will note above that my screenshot highlights high speed rail; the growth was over 3000%. That is far too much to include in my graphic, so I compared the actuals instead. That is one of the tradeoffs, but in my mind it is an acceptable one.
Credit for the original goes to Paige St. John and Armand Emamdjomeh.
Recently the National Post looked at the results of a Canadian census that identified significant growth in people identifying with the aboriginal populations of Canada. As an American, I am not terribly familiar with Canadian native populations, but if I recall, they are broken into the three groups examined in the infographic: First Nations, Inuit, and Metis. The First Nations are the original tribes of Canada, the Inuit are the natives from northern Canada, and the Metis are the mixed-race persons of native and early European colonisation.
Aboriginal Canada
I find interesting the National Post’s use of network diagrams (the bubbles with lines) to show how the subcomponents form the whole. This as opposed to perhaps a more common form of a tree map or bubbles within a bubble. I would be curious to see or learn about which is the most effective at showing the relationship both in terms of structure (hierarchy) and size (without the datapoints included as labels).
Credit for the piece goes to Andrew Barr, Mike Faille, and Richard Johnson.
Today’s post comes from xkcd. It looks at how much time can be spent improving efficiency before you become an inefficient efficiency person. It is important to note that this is over a five year span. And while I do not know about my readers, I can barely stick doing one thing for more than a year.
Yesterday both the New York Times and the Washington Post published fascinating pieces looking at the difference in the cost of medical procedures. But each took a different approach.
I want to start with the New York Times, which focused at the hospital level because the data is available at that level of granularity. They created a geo-tagged map where hospitals were colour-coded by whether their bills were below, slightly above, or significantly above the US average.
Hospitals across the United States
The ability to search for a specific town allows people to search for their hometown, state, country and then compare that to everyone else. My hometown of West Chester, Pennsylvania is fortunate—or perhaps not—to have several hospitals in the area that charge at different rates. That makes for an interesting story. But I am from the densely populated East Coast and someone from say rural Montana might not have the same sort of interesting view.
Hospitals near West Chester, Pennsylvania
Regardless of the potential for uninteresting small-area comparisons, once you find your hospital, you can click it to bring up detailed statistics for procedures, costs, and comparisons to the average.
Brandywine Hospital's data
All of this makes for a very granular and very detailed breakdown of hospital versus hospital coverage. But what if you want something broader? What good is comparing Brandywine Hospital to some medical centre in Chicago? Neither is reflective of the healthcare industry in the Philadelphia area or the Chicago area, let alone Pennsylvania or Illinois. The Washington Post tackles this broader comparison.
The Post leads off with a hospital-level example from Miami. Two hospitals on one street have vastly different prices. If we knew about this in Miami we could surely find that in the New York Times map. Instead, the Post guides us to that kind of example.
Comparing two hospitals in Miami
But the broader view is the centre of the piece. Using dot plots and filters, the user can compare the state averages for 10 different medical procedures. Fixed to the plot are the minimum and maximum averages along with the national average. And given the Post’s smaller circulation area—the New York Times is national, the Post is less so—there are quick links to states of particular interest: DC, Maryland, and Virginia.
Pennsylvan's averages
The ability to pick different states from the drop down menu allows the user to quickly see differences between states. What is lacking is perhaps a quick view of where all the states are visible so that the user does not have to click through each individual state.
California's averages
Both pieces are very successful at their narrowly-focused aims. Neither tries to do everything all at once, but nor would their designs allow for it. Plotting and filtering all the hospitals could be done in the Post’s style, but it would be messy. The state averages could all be made to colour state shape files, but you would lose the inter-procedure differences, the minimums, maximums, and the averages. In short the two pieces from the two teams complement each other very well, but a weird and hybrid-y cross of the two would be large, cumbersome, and potentially difficult to use without spending a lot of time to design and develop the solution. (Which I imagine they did not have.)
Credit for the piece from the New York Times goes to Matthew Bloch, Amanda Cox, Jo Craven McGinty, and Matthew Ericson.
Credit for the piece from the Washington Post goes to Wilson Andrews, Darla Cameron, and Dan Keating.
Well, actually, your author is driving back from Ganister today. Unfortunately, while on holiday I was not working (nor was I planning to.) So while I could of run silent today, I wanted to share with all of you again a project I created last year about my return drive from Ganister. For all of who familiar with the piece, I apologise for my re-posting of previous work. For those of you unfamiliar with the work or with Ganister and its distance/remoteness, enjoy. (It’s full-size, so no click-for-higher-resolution.)
Earlier this year, the mobile phone (or cell phone for many Americans) turned 40. Today’s infographic comes from the National Post and looks at the history and the near future of the mobile phone market, mobile phones, and related technologies. A nice touch is a actual-scale drawing (best seen in print) comparing a modern iPhone to an “old school” mobile phone, as shown in the cropping of the original below.
The history of the mobile phone
Credit for the piece goes to Mike Faille and Kristopher Morrison.
Of 2048. Well, kind of. Lately the country has been talking a lot about immigration and its impacts because of this bipartisan desire to achieve some kind of result on an immigration bill working its way through the Senate. One of the common thoughts is that if we legalise a whole bunch of illegals or document most of the undocumented (I’ll leave the language for you to decide), the new American citizens will overwhelmingly vote Democratic and there goes the Republic(an Party).
Nate Silver—yes, that Nate Silver who accurately predicted the presidential results and a whole bunch of other stuff too—looked at a more complex and more nuanced set of demographic variables and found that the aforementioned argument greatly oversimplifies the results. The problem is not entirely the entry of newly documented or illegal workers. Instead, there are systemic demographic issues.
So here comes the New York Times with an excellent data explorer and forecast modeller. You can set the year to examine and then set the results of the immigration debate with how many immigrants are made legal/documented and then how many of them vote. After that you can begin to adjust population growth, voting patterns, &c. to see how those affect the elections. (The obvious caveats of acts of god, party platforms, candidates, &c. all hold.)
2048 Results
The fascinating bit is that if you keep the demographic patterns as they are currently, adjusting the immigration factors at the outset have very little impact on the results. The country is moving towards the current Democratic platform. Even if 0% of the undocumented/illegal immigrants become documented/legal, and if 0% of 0% vote, the result is still a landslide for the Democrats. The real changes begin to happen if you adjust the population growth rates of the legal/documented citizens and voters. But those patterns/behaviours are a lot more difficult to adjust since you can’t legislate people to have more babies.
All in all a fascinating piece from the New York Times. The controls are fairly intuitive, drag sliders to adjust percentages. The sliders have clear labels. And the results on the map are instantaneous. Perhaps the only quirk is that the ranges of the colours are not detailed. But that might be a function of forecasting the data so far into the future and having growing ranges of certainty.
Credit for the piece goes to Matthew Bloch, Josh Keller, and Nate Silver.
Today I have more immigration-related information graphics and data visualisation for you. Earlier this week the New York Times looked at immigration to California, but this time the focus was on Asian population growth and not Hispanic. The graphic here supports an article looking at where the growth has been focused in California. And given that emphasis, the map accompanying the article makes sense. And as the reader can clearly see, much of that growth has been centred in the San Gabriel Valley and Orange County.
Following on last week’s posts onimmigration comes today’s post on how that might impact Republican politics. Well I say might but pretty much mean definitely. The graphic comes from the Wall Street Journal and it takes a look at the demographic makeup of states, House congressional districts and then survey data on immigration broken into Republicans vs. Democrats.
The GOP's Tricky Terrain
I think the piece is a good start, but at the end of the introductory paragraph is the most salient point about the piece. And unfortunately the graphic does not wholly embody that part. Of course within limited time and with limited resources, achieving that sort of completeness is not always possible. That said I think overall the piece is successful, it just lacks that finishing graphical point.
Credit for the piece goes to Dante Chinni and Randy Yeip.
Earlier this week, Wolfram Alpha released some findings from its analytics project on Facebook. While the results offer quite a bit to digest, the use of some data visualisation makes it a little bit easier. And a lot more interesting.
The results offer quite a bit of detail on interests, relationship statuses, geographic locations, and ages. Below is just one of the small multiple sets, this one looks at the number of friends of different ages for people of different ages. Basically, how many young or old people are friends of young people? Friends of old people?
Friends of Friends for the Ages
But I was most interested in the analysis of social networks. The mosaic below is indicative of the sheer size of the survey, but also begins to hint at the variance in the social structures of the data donors.
Just Some of the Networks
While these views are all neat, where it begins to get really interesting is Wolfram Alpha’s work on classifying the different types of social networks. By aggregating and averaging out clusters, simple forms begin to emerge. And after those forms emerged, they were quantified and the results are a simple bar chart showing the distribution of the different types of networks.
Simplified Cluster Distribution
Overall, some very interesting work. But one might naturally wonder how their own networks are structured. Or just be curious to look at the data visualisation of their own Facebook profile. Or maybe only some of us would. Fortunately, you still can link your account to a Wolfram Alpha account (you have to pay for advanced features, however) and get a report. Below is the result of my network, for those who know me semi-well I have labelled the different clusters to show just how the clustering works.
