“Visualizing data is as much an art as a science. And seemingly tiny design decisions — where to set a color threshold, how many thresholds to set, etc. — can radically alter how numbers are displayed and perceived by readers.”
Author Archives: Matt Stiles
I'm a data journalist at The Wall Street Journal. Let me know if you have ideas for future visualizations. Contact me: mattstiles at gmail.
This post has been updated. See correction at the bottom of the page.
To some Bernie Sanders supporters, the Democratic presidential race must seem close. Their candidate, after all, has essentially split victories with Hillary Clinton in the more than 30 election primaries and caucuses since the process began in February — including several in a row recently.
Clinton, their thinking goes, may have a lead in pledged delegates for the nomination, but her sizable and (for now) critical lead among party leaders known as “superdelegates” could crumble if the Vermont senator continues winning.
Anything’s possible. But Clinton has already secured many, many more votes than the Vermont senator and, party rules and delegate grappling aside, is absolutely dominating the race in terms of raw support. This trend is likely to continue with large, Clinton-friendly states coming up, and it could undercut Sanders supporters’ “will of the voters” hypothesis going forward.
Clinton has won roughly 9.4 million votes, compared to Sanders’ 7 million, according to returns from U.S. states. Along the way she’s posted huge victories. Her margin in Florida alone (530,000 votes) is about the same as Sanders’ margin in his victories combined.
Here are a few quick sketches (see correction below) that help illustrate this fact. First, let’s look at their states on a scatter plot to explore not just the number of victories for each candidate but the size of their respective states — and the relative victory margins:
Here’s a similar view in the form of a bubble chart (or, Alastair Dant would say, “BALLS!”):
And, finally, two maps. Again they show the large margins — and geographic differences — evident Clinton’s victories:
UPDATE: A Twitter user suggested I look at the data with time in mind. Not sure it proves his point, but here’s another sketch:
(Correction: Some caucus states — Washington, Nevada, Maine and Wyoming — reported candidates’ proportional number of delegates selected to state conventions, not actual votes. The charts below now reflect margin estimates extrapolated from the total Democratic caucus participation by delegate share, like this. The charts were updated to reflect these estimates. The central idea of the post, and the overall popular vote margin difference between the candidates, remains virtually unchanged, however.)
“[I]n some other parts of the country, adults with the lowest incomes die on average as young as people in much poorer nations like Rwanda, and their life spans are getting shorter.”
The GOP presidential candidates collectively have raised more than $300 million in this election cycle, according to Federal Election Commission data. Here’s a quick look at where the several of those candidates — Jeb Bush, Ben Carson, Chris Christie, Ted Cruz, John Kasich, Marco Rubio and Donald Trump — collected the money.
Each dot on the maps (see update below) below represents at least $5,000 raised by zip code. The dots in each place are assigned randomly within the zip code boundaries, creating a density map (think Verizon vs. AT&T) for campaign donations* through Feb. 29.
Here’s a (crowded) map with all the candidates. The dots were layered alphabetically, a method that unfortunately obscures some candidates at this scale, depending on the geography. Still regional differences are evident:
The former Florida governor raised large amounts of money, of course, from his home state, but also — as the perceived frontrunner for a time — from major population centers:
Carson raised a ton of money for someone who received relatively little love once voters started casting their ballots. His map shows he raised more from small locations throughout the country, not just the major population centers:
The New Jersey governor, understandably, raised a large proportion of his money from his home region:
The Texas senator raised money from all over the country, but he clearly has a financial base in Dallas-Fort Worth, Austin, San Antonio and Houston — the major population centers in his home state.
Like Christie, the Ohio governor tapped his donor base at home:
The Florida senator’s map looks much like that of his frenemy, Bush:
The billionaire business man, who has relied largely on personal funds for his campaign, has raised very little money in checks larger than $200, as reflected in his map:
* To keep the analysis consistent for all campaigns, the maps were created with a data set that excluded individual donations of less than $200. The data include only donations to the candidates’ campaign committees, not other political committees supporting them.
UPDATE (9:20 p.m. EST): Based on David’s suggestion in the comments, I’ve changed the dot color in the individual maps so it’s consistent. You can see the originals here: Bush | Carson | Christie | Cruz | Kasich | Rubio | Trump.
Last week we examined how the Democratic presidential campaigns have performed in the context of Barack Obama’s 2012 re-election.
That analysis grouped Obama’s vote share into categories, highlighting how the country’s reddest and bluest counties have voted for Hillary Clinton and Bernie Sanders so far in the Democratic primaries. Clinton, the clear frontrunner, performed best in areas where Obama was strongest against Mitt Romney. But Sanders did slightly better when majority black counties weren’t factored.
Here’s a look at the Democratic race (through the most-recent contests) in the context of voters’ educational attainment. Each candidate’s average vote share by county is grouped by the proportion of residents in those areas with at least a bachelor’s degree. Sanders doesn’t win among any group, but he generally performs best in places where voters have more education:
When we travel to and from South Korea, we’re often forced to trek out to Incheon International Airport (this country’s Dulles). The massive airport, located on an island about an hour west of Seoul, opened in 2001 and is now one of the largest and busiest in the world.
I had no idea, until today, how dramatically the shoreline around the airport has changed in the last three decades. Previously separated islands along the Yellow Sea coast were joined together as reclaimed land. The airport is now connected to the metropolis by the Incheon Grand Bridge, which opened in 2009.
For context, here’s a current Google Map showing the region:
Now check out how much the land area has changed and the urban growth has expanded, as depicted by these Landsat images acquired in 1981 and again in 2013. The Landsat archive contains 40 years of data, allowing users to see changes (like those in my former home, Washington, D.C.) to the Earth’s landscape. Check it out.
Here’s a before/after view of the airport:
Among the fascinating aspects of American politics are the various factors — demographic, financial, historical, etc. — that shape the geography of campaigns.
This election cycle is no different, with presidential candidates in both parties winning a seemingly random mix of specific counties and overall states since the primary season began in Iowa on Feb. 1.
On the Democratic side, one question has nagged me this week: How have Hillary Clinton and Bernie Sanders performed in the context of President Obama’s 2012 re-election campaign?
Inspired by a recent FiveThirtyEight story about the Republicans, I compared the Democratic candidates’ vote share in more than 2,050 counties thus far with how Obama did in those places four years ago.
Clinton has won about two thirds of the counties contested by the Democrats thus far that Obama carried in 2012. She has won about 5.8 million votes in those places, compared to roughly 4 million for Sanders.
But is Clinton’s support stronger in places where Obama was stronger? As a data exercise, I looked at Obama’s vote share by county and county equivalents and then grouped the locations based on his strength. (In some places contested by Clinton and Sanders, of course, Obama got clobbered by Romney. In others, he won by huge margins). I then calculated how Clinton and Sanders did respectively in each.
This first chart shows the results. It’s clear that Clinton does better in places that were the most Democratic in 2012. The more “blue” the county for Obama, the higher her vote share, on average:
See FiveThirtyEight’s take on Sanders’ five-state winning streak in caucuses this week:
Sanders’s strength in caucuses may also be, in part, coincidental. Every state that has held or will hold a Democratic caucus this year has a black population at or below 10 percent of the state’s total population, and black voters have been among Clinton’s strongest demographic groups. Without those black voters, Clinton just can’t match the enthusiasm of Sanders’s backers. (In Southern states, where Clinton romped, her voters were far more enthusiastic than Sanders’s supporters were.)
So I ran the numbers again, but this time removing the 93 counties where blacks are the majority, mostly in the Deep South. The picture is a bit different, showing how Sanders competes more strongly when the black vote is less of a factor:
(View interactive versions of the charts — and the data tables — here).
Sanders, of course, doesn’t get the pick the voters’ demographics.
Tomorrow we’ll look at how Clinton and Sanders performed when counties are grouped by their residents’ educational attainment. Hint: Sanders does better in places where a higher proportion of people 25 or older have at least a bachelor’s degree.