English in America

EnglishLike many Americans, I’m a little hazy on my family tree. However, I’ve always thought of myself as predominately British, and assumed that as a former British colony, English descendents would be heavily represented when I mapped them.

This was not the case.

While America has plenty of English-Americans, they’re much more sparse than, say, German-Americans. This also brings up an interesting question – how much of these census results is attributed to actual lineage, and how much is due to variances in self-reporting? In other words, when you can count German, English, Welsh and Scottish relatives as part of your family tree, how do you choose to describe yourself? Perhaps claiming English descent seems too pedestrian, and loses out to the more “exciting” choices of, say, Scottish. To get true data, you would probably have to turn to genetic testing, and data from genetic testers that release demographic information can’t be trusted as being statistically representative of everyone – getting tested costs money, and requires an interest in such things. That’s going to skew results.

For now, the census is as good as it gets, and I was quite surprised by this one. Here’s hoping I’m not alone – expected results can be awfully boring.

Link to my Tableau tool:




Continuing my look at American immigration, today I’ve put together a map showing the distribution of zip codes with concentrations of residents of Danish descent. I’ve also made changes to the Tableau tool so that you can now mouse over zip codes and get information on the nearest city, the county, and the state where that zip code is located (the graphic overlay makes it impossible to read state borders and such in Tableau).

Calling America a melting pot is a cliche – but looking at the ingredients is fascinating. I’m currently working on getting racial data pulled in, because the ancestry data in these tables is curiously lacking of such obvious and necessary things as native Americans, Mexican-Americans and African-Americans. The census can be maddening to work with.

Here’s a link to the Tableau, so you can zoom in, get county info, etc:


Technicolor Hyper-kinetic Campaign Finance Explosion!


The Louisiana Senate race draws on money from across the nation. The animation above is looking at individual donations – donations that can be tied directly to a particular human (rather than some committee) and shows how, over time, both Landrieu and Cassidy have seen their in-state share of donations shrink. Cassidy ends up with 58% of his donations coming from Louisiana, but if we look at breakdowns by quarter, we find that in the most recent quarter, Cassidy received more out-of-state dollars (as a percent of his total) than Landrieu did.

Click on any graphic to get it full-screen — they look *much* better that way.

Percent from Louisiana Line Chart

Before you become completely offended by so much outside money flowing into Louisiana, it’s important to have a benchmark. Hence, we take a look at Kentucky, where *both* Senate candidates are running at roughly 20% in-state donor levels. Let me know which design you prefer, by the way – I like the Louisiana animation better. It’s more recent, and I think I improved it with time. I’d like to know what everyone else thinks.kengif2







An overall comparison of district 6 congressional candidates — this is pure horse-race coverage stuff, trying to show lead changes in fundraising over time:


Now, let’s kick up the epilepsy-inducing intensity! Here are animations of District 6 house race candidates, and their in-state percentages of donations over time:

edwin-edwards-animated-gif-FIXED       new-animated-gif-graves

new-animated-gif-felder   new-animated-gif-trey-thomas

new animated gif claitor




And, finally, we look at the same candidates, only *this* time we examine their in-district versus out-of-district fundraising. Be sure and check out Edwin Edwards and how he changes once you look at district rather than state.

district-animated-gif---charles-thomas  district-animated-gif---edwin-edwards


A Viz about Vizzes

I’ve committed to Tableau as my primary tool, and I’d like to think I’ve learned how to use it, but just how much progress have I objectively made?

I figured I’d use Tableau and find out.


To me, the thing that jumps out is the audience lag. My “production mountain” shows all the work I’ve done, in terms of the cumulative number of worksheets and dashboards I’ve designed in Tableau over time. Cumulative views, on the other hand, shows how many views I’ve gotten based on the date I published various visualizations. I hit 100 total sheets in October of 2013, yet I didn’t hit my first chunk of views until late January, 2014 – approximately a 3 month delay. Apparently, I needed practice before I properly figured out how to publish tools worth eyeballs.

I also have a bubble map showing my most popular visualizations.This one is colored by time, using a heat map scheme – so the most recent visualizations will have a more brick-red color with older visualizations fading out toward blue.

I already mentioned my two area charts – Production Mountain and Cumulative Views Over Time. These are both running totals, one showing how many sheets I’ve designed in Tableau (individual sheets inside projects I’ve published), and the other showing how many views my visualizations have garnered.

The Monarchist of LSU

One hears about liberal bias in education on a regular basis – but is it true? Just how liberal *is* LSU?

ImageNot very, as it turns out.

After running every name from the Reveille’s database of faculty and staff to retrieve their political party affiliations, I found that the LSU faculty is actually about 10 percentage points less Democrat-leaning than the surrounding Parish. The data says that LSU is no hotbed of liberalism – and it also showed we had a Monarchist. That one came as a surprise.

Working with Fernanda Zamudio-Suarez, my favorite reporter at the Reveille, we put together a great package showcasing the results and the implications of those results. Fern even knew the Monarchist – Faculty Senate President Kevin Cope – and got him to open up about his party affiliation.


The data was a blast, and it showed all sorts of fascinating things. The most liberal department on campus, unsurprisingly, was Music & Dramatic Arts, where 71.79 percent of the faculty & staff are registered Democrats (of those I was able to retrieve party affiliation for). The least liberal department, was UC Advising & Counseling, where 20 percent of the staff are Democrats, and 46.67 percent are Republicans.


To me, this is the perfect example of how data journalism should be done. We took a dataset we owned – the salary database for the university – and leveraged it into a brand new story, and a brand new insight, simply by running it through a new source of information. Data is all about recombinant information – the mutational evolution of your understanding of the world.

Plus, we found a Monarchist. I don’t think that’s ever going to get old.

Now, I just need to find a new, fun data project to feed Fern – I’m sure she’s disappointed I haven’t kicked anything good her way in quite some time.

One final addendum – after publishing, a friend of a friend asked about what would happen if we had zeroed in on the faculty alone – rather than faculty & staff together. I can’t do that breakdown perfectly, but I can approach it, and when I do it shifts the numbers by a bit less than 3% toward Democrat affiliation. The new numbers, if I’m trying to just grab them for faculty without staff, run 40.77% Democrat, 25.47% Republican, and 33.03% Independent. Don’t let the apparent precision of those percentages fool you — they’re my best attempt at faculty isolation, but they’re not perfect. Even these isolated numbers, however, fail to achieve a strong liberal bias. If you want strong bias, your best bet remains looking at different departments. Music & Dramatic arts, I’m looking at  you.

Special thanks to professor Rosanne Scholl for pointing out that whatever my personal feelings on the matter, the major political parties remain proper nouns and deserving of capitalization. Thanks as well go out to Barbara Clark, who asked the question I failed to consider: How do the numbers change if you look at faculty separate from staff?

Turkeys Over the Years

turkeyAs we sit down to feast and avoid picking a fight with Uncle Pete, why not take a moment to reflect on all the great turkeys to go before? America hasn’t always eaten so much turkey, but after a mad dash to the top, the bird has apparently hit some sort of glass ceiling… Or glass coop, at least. Domestic consumption has leveled off, and only increases in exports appear to show much hope for turkey sales to really take flight.



College Football Ranking Sentiment Map

Knowing where your team ranks each week is only half the battle – the other half is knowing *why* they ranked the way they did.

Map of rankings AP pollsters assigned Missouri, broken down by where pollsters live

Map of rankings AP pollsters assigned Missouri, broken down by where pollsters live

My new tool for the Reveille shows rankings, who voted for each team, and even puts together a national sentiment map to show where the voters live who love your team, as well as where on the map you’re not getting any love at all.

Here’s a link to my tool: http://public.tableausoftware.com/views/ReveilleWeek9APRankings-Narrow_0/DetailedAPCFBPollResults#1

Here’s the story on the Reveille: http://www.lsureveille.com/sports/football/interactive-see-where-ap-voters-ranked-lsu/article_af139c7c-39e2-11e3-82b4-001a4bcf6878.html

Not sure why my earlier post went away, but WordPress seems to enjoy doing that to me. Here’s hoping this post stays intact, so everyone can enjoy, share, and explore.

Oh – and be on the watch for a similar tool come basketball season!

Using Tableau With Porn


Data is beautiful, in that it doesn’t care about its surroundings. Data can be gathered by cloistered nuns, and it can be gathered by pornographers. Massive online porn destination, Pornhub, recently demonstrated this when it released an interactive showing the top three porn search terms, as well as the average length of stay, broken down by each state.

I will include a link to Pornhub’s viz, but first I’d like to put up a link to the “corrected” version I made. I found the original difficult to navigate when you wanted to see a particular state’s results. I fixed this by making the map a control feature. Now, when you click on a state, you only see that state’s information elsewhere on the page. Here’s the link:




It defaults to Louisiana. Simply click on our state to see everyone, or click on another state if you’d just like to jump to their results.

I don’t anticipate working with pornography data again anytime soon. That being said, I would definitely like to get my hands on some of the raw figures involved. I’d just be sure and wash my hands really well after.

Here’s Pornhub’s original visualization, without my snazzy map cotrol:




And at the moment, they have it saved with a state selected. Not sure what’s up with that. But I applaud their use of the technology. I would love to see more corporations follow their lead. Internal data holds amazing stories – share them with the world!