Monday, December 11, 2017

Don't refrain from using uncommon visualizations; explain them

I've heard from people who teach visualization that one of the questions they often get is why, when, and how to use uncommon graphic forms. Some designers prefer to stick to forms they think can be understood by most readers (bar graphs, time-series line graphs, pie charts,) and avoid unusual ones.

As I argued in The Truthful Art, I believe that this is a self-defeating strategy because (a) sometimes an unusual graphic form is a good way to convey a message, and (b) by using it and explaining how to read it, you'll expand your audience's visual grammar and vocabulary —their “graphicacy”. The caption of this 1849 line graph is an example, and so are the thorough annotations that pioneers such as William Playfair wrote to further clarify their graphics.

The latest example I've seen of this strategy is this story by Zeit online. It isn't just a nice combination of strip plots, scatter plots, and maps; the authors inserted an explainer of how to interpret a bivariate choropleth map in the story itself. This one (the color choice has been critiqued, by the way):


They also wrote explanations like the one below, describing a scatter plot. This may sound redundant to experts but I think that it can be very useful for many readers:
“The diagram shows the distribution of medical practices in relation to private patients by county. Each point represents a district. The further to the right the point lies, the more private patients there are. The higher the point, the more medical practices there are. The red line makes the connection clear: many doctors are where there are many private patients.”
Well done.

UPDATE: Jim Vallandingham has just published a good gallery of multivariate maps.


Friday, December 1, 2017

A makeover of the visualization of voting similarities in the Senate

A few days ago I praised this graphic of voting similarities in the Senate.


Andrew Gelman saw it too and offered some intriguing suggestions. (please read them!) However, he didn't do a makeover himself. I decided to use his ideas and design a very quick sketch to see how an alternative graphic may look like. You can see the mockup below. A few warnings:

1. I used the same similarity scale as Gramener. I asked them for the data.
2. Take the ideological score with a grain of salt. I got it from here but I didn't verify whether it's trustworthy. McCain is on the right position according to his ideological score (X-scale.)
3. Around one quarter of Senators are not shown because of missing data.
4. I have no particular preference. I like either of version for different reasons.



Tuesday, November 28, 2017

Visualization at Eurostat

Last week I was in Luxembourg, where I did the Visual Trumpery talk a couple of times. In between, I had time to chat with statisticians from both STATEC (Luxembourg's statistical institute) and Eurostat.

The Eurostat folks showed me some infographics and data visualizations and tools they've done recently. I'm fond of their Regions and Cities project and their Statistical Atlas. The former let's you visualize many variables in multiple ways: Maps, dot plots, tables, etc.:



And the latter is simply a great source for data to play with. You can select any variable, see it visualized on a map, and then download its data (scroll down on the Legend, on the bottom-right corner of each screen.) As far as I've seen —I downloaded a couple of csv files,— the data is already formatted tidily, which is nice. I wish more sources of would provide spreadsheets without useless headers, footnotes, comments, and in long format, rather than wide:



Monday, November 27, 2017

Visualizing visualization

Our latest Google News Lab's visualization (see the entire series and read about them) is pretty meta-referential: Designer Anna Vital's The Visualization Universe reveals the search interest for visualization books, tools, and charts.

FastCo Design has just published an article about the genesis of this project. As usual, I didn't do any hands-on work myself; I only provided some general art direction. We wanted the result to be pretty straightforward, so Anna used mostly stacked bar graphs and sparklines.

On the opening screen you can select one category and see a ranking that you can sort by name, overall search interest, or change in comparison to last year:


Clicking any icon will reveal more details:


If you notice that something is missing, you can contact Anna through this form or send her an e-mail.

Sunday, November 26, 2017

Visualizing voting similarities in the Senate

Gramener, a data science firm based in India, has just published a neat visualization of voting similarities in the Senate. As the intro to the project itself says “When senator 'X' votes a 'Yea' or 'Nay' what are the chances that senator 'Y' would do the same?” Just a few thoughts:

1. I like how simple but rich this graphic is. You can first select one Senator through the drop-down menu, and then click on any circle to get the similarity with another Senator.

2. I wish someone will do something similar with the House of Representatives, if only to reveal that some Congressmen who claim to be “moderates” (like my district's Carlos Curbelo) are actually hardcore party hacks who vote with their tribe nearly 90% of the time.

3. Numerical summaries and alternative ways of visualizing the data could help a lot. I wish I could make comparisons between each senator and each party's average or see some rankings. I got the feeling, for instance, that, overall, Republicans tend to vote more similarly to each other than Democrats, but I couldn't verify that hunch.

Here are some highlights:


Wednesday, November 15, 2017

Mapping HIV

I've been meaning to write this post for quite a long time. Months ago, I discovered AIDSVu.org, an initiative by Emory University’s Rollins School of Public Health in partnership with Gilead Sciences  that tracks HIV in the U.S. The data comes from CDC.

As I liked what I saw, I wrote some quick notes but, as it often happens, completely forgot about them. Then, three weeks ago I met some of its creators at the World Conference of Science Journalists, where I gave a talk (here's a summary of what I said.) I promised that I'd recover the notes and publish them, as this is a visualization project that is worth your time.

Once you enter the AIDSVu.org, it'll locate you —if you don't block location services— and display statistics that pertain to your area. After that, you can jump to a national map with tons of filters:



Or you can look for HIV testing and care sites, or learn how to use the maps and the underlying data for other purposes. You can also explore your own region. Here's my ZIP code:


The website also has “infographics”, which summarize certain portions of the data sets to address specific stories or audiences, a blog, a pretty active Twitter account, and city profiles. This is Miami.


Besides its obvious goal of letting people grasp the scope of the HIV, AIDSVu can be a useful resource for classes. It showcases almost no graphs, so this may be a great opportunity to imagine other ways to visualize its data besides the existing maps: Comparisons and rankings of states, cities, and ZIP codes, visualizations of relationships between variables, etc. AIDSVu promises to showcase your work, if you decide to share with them.

Monday, November 6, 2017

Metal and sunshine

One of the most celebrated examples in the Visual Trumpery lecture is this map of concentration of Heavy Metal bands per million people in Europe (source):



I use this map to discuss how to verify a source (spoiler: I love the map, and the data on it looks OK to me.) During the talk I often joke that this confirms something that, as a hard rock fan, I already guessed: overall, the less sunlight a country gets, the more metal bands it tends to have.

The audience loves this joke, but is it true? It turns out that, at least at a first glance, it may not be or, at least, it needs some caveats. Here's a quick ggplot2 of the relationship between the average yearly sunshine duration in the capitals of most of these countries and the concentration of metal bands (Note: I used the capital as a rough proxy, being very aware that many of these countries are big, so that the amount of sunlight within them varies a lot. In Spain, for instance, the South is very sunny, but Northern regions such as Galicia, where I was born, are cloudy, rainy, dark —and maybe very metal):



So my hunch is tentatively dubious, at least until we get more detailed data and think carefully about it: the relationship is clearly not linear and, even if that were the case, correlation isn't that strong. Besides, the graph reveals some intriguing features, such as at least three clusters of countries. The group on the left is largely made of Western, non-Mediterranean countries; the one in the middle is mostly Eastern and South-Eastern countries; and the last group, on the right, is Southern Mediterranean countries (Spain, Portugal, Greece, Turkey, and Italy.)

If we isolate the first group, with the exception of an outlier (Iceland) the relationship is positive: the more sunlight, the more metal bands. And if we ignore the four outliers on the Y-axis from the overall mix (Finland, Norway, Sweden, and Iceland,) there's barely any relationship. See graphs below.

What can we learn from all this? Not much, I guess, other than that even silly jokes ought to be verified and presented with plenty of caveats.





Wednesday, October 18, 2017

Uncertainty, graphicacy, and the power of statistics

Power from Statistics is an initiative by Eurostat and the European Political Strategy Centre. Their conference in Brussels begins today, so they've just launched their report. I attended one of the roundtables that led to this document, so they asked me whether I'd write an article for it.

The result is “Uncertainty and graphicacy: How should statisticians, journalists, and designers reveal uncertainty in graphics for public consumption?” (PDF), which consists of some miscellaneous thoughts about why the public —including journalists!— don't grasp uncertainty, how people even misread common charts, graphs, and maps, and what we could do about it. If you have attended any of the Visual Trumpery lectures, a few of these musings may sound familiar.

This is an essay in the literal sense: Writing intended to help myself think and play with some ideas, such as the timeline of “Golden Ages” and “Dark Ages” of visualization (below,) so take everything with a grain of salt. I'm more than willing to change my mind on anything I wrote. Enjoy.

(This same article will appear soon —abbreviated— in a book about visualization in the news. Consider ordering it. It looks fantastic.)


Tuesday, October 17, 2017

Interview at Stats & Stories

A few weeks ago I had a fun conversation with John Bailer, Richard Campbell, and Rosemary Pennington, the people behind the Stats and Stories podcast, also available at NPR.

We talked about graphicacy, Visual Trumpery, how numbers mislead, and many other topics. Listen to it if you have time.

Sunday, October 15, 2017

Ain't Data Truth?

My University of Miami colleague Hiram Henríquez has created an infographics non-profit organization called Ain't Data Truth, which he presented at the Malofiej Infographics Summit this year. Its main output is a series of enormous data posters about current events. Take a look at the “topics” menu on the upper-right corner of his website. You can also send him your feedback and suggestions.

Hiram worked in news media for many years (Miami Herald, National Geographic magazine, etc.,) and now, besides teaching at our School of Communication, keeps doing freelancing and side projects like this. Here's the latest poster he has published, about climate change (hi-res PDF here):


Tuesday, October 10, 2017

Visualizing gender and race inequality in newsrooms

Our latest project in the collaboration with Google News Lab —read about it here and see all projects here— is an exploration of gender and race in U.S news publications. It was designed by Polygraph based on data from the American Society of News Editors (ASNE,) which has also published an article about it.

One of the reasons I love this interactive visualization so much is the multiple ways in which the data is visualized —bubble plots, scatter plots, dot plots, tables, etc.— and also the animated transitions between them. Don't miss it.




Friday, September 29, 2017

An update on Visual Trumpery: New cities and dates

I've already delivered the Visual Trumpery talk in Mexico DF, Barcelona, and Atlanta. Speros Kokenes, who organized my visit to Atlanta, has written an article about it.

In October I'll be visiting Portland, Berkeley, Redlands, New York City (where nearly 300 people have already signed up!) and Ithaca (Cornell University.) The November schedule is also quite packed.

I'll soon add new places and dates confirmed for 2018, including Washington DC, Baltimore, Miami, Chicago, Columbus and Athens (OH), Syracuse (NY), Auckland (NZ), London (UK), and a few cities in Spain, Canada and Poland.

As I explained a while ago, this is what you need to bring Visual Trumpery to you:

1. A flight (coach is fine) from Miami and a place to stay (I'm not picky.)

2. I won't charge anything, but I'll need some minor expenses covered, such as taxis to and from the airport, meals, etc.

3. You must arrange a venue and announce broadly. I can help with promotion through social media, of course. The talk cannot be part of a paid-for event or conference. It must be open to the public.

4. I can present in Spanish, Portuguese, and English. If the audience speaks any other language, we may need an interpreter.

Wednesday, September 27, 2017

Low-tech visualization: How much space newspaper front pages used to cover hurricanes

Many of my students get a bit overwhelmed at the beginning of each semester by the amount and variety of tools we use in class. I decided to show them that sometimes you can create pretty neat visualizations with rather pedestrian techniques, such as drawing basic shapes in programs like InkScape or Adobe Illustrator.

I spent a couple of hours today designing the two graphics below. They depict the space devoted to the threat and consequences of hurricanes Harvey, Irma, and Maria in the past month on the front pages of The New York Times and The Washington Post. I first downloaded all cover images from both websites, and then drew the rectangles over them. Colors correspond to the region or regions that are most prominently mentioned on each story.

To see these in high resolution (ai, pdf, svg,) go to this folder. Feel free to use them.

UPDATE: Lynn Cherny and Moritz Stefaner have just told me that designer Krisztina Szűcs did something similar a while ago. It looks great. Check it out.


Monday, September 25, 2017

Updated Tutorials & Resources section

You're probably aware that this website has a Tutorials & Resources section where I post videos myself or other people have recorded. I've just updated that section with a short and quite informal tutorial about RAWGraphs, a great and tool by the Density Design lab.

These are materials I use in my classes at the journalism and interactive media Masters programs at the University of Miami. I'm a fan of “flipped classrooms,” so I don't devote precious class time to basic software training, just to answer questions about the tools, or to explain advanced techniques if students need them. Tutorials take care of software basics, so we can use class time for lectures, discussions, critiques, and feedback on exercises.

Here's a diagram of what we use each tool for:


Before you ask: Yes, Tableau and PowerBI are part of my classes, but later in the semester.

Tuesday, September 12, 2017

A conversation about designing better visualizations —and spotting misleading ones

I talked to the good folks at Discourse a while ago. If you enjoy in-depth journalism, you could consider following them. They've just published an edited version of our conversation, keeping just the good parts.

We discussed several takeaways from my Visual Trumpery lecture series and some of my to-go sources for great news visualization. Here's the most timely part, considering that ProPublica has just announced their partnership with several data scientists:


Thursday, September 7, 2017

In visualization, captions are as important as graphics themselves

(Updated on September 8 and 9. Go to the bottom of the post)

Data visualization isn't just about visualizing data, but also about writing headlines, intros, captions, explainers, and footnotes. I'm right now closely following the news about hurricane Irma —I live in Miami!— and feeling both amazed and terrified by the many great graphics news organizations and independent designers are publishing. As I've just tweeted, beauty is sometimes correlated with terror.

Anyway, I've just read a very good graphics-driven story in The Washington Post. This is its first map:



This is its caption:


I'm no expert in weather forecasting, but I believe that this is inaccurate. To learn why, go to minute 14:30 in my keynote at Microsoft's Data Insights summit. Here's some of what I said there:

Maps based on cones of uncertainty are quite problematic, as this article by Jen Christiansen, and this other by Robert Kosara explain. Among other reasons, some people don't see in that cone the possible range of paths the center of the hurricane can take, but the size of the hurricane itself.

This happens event to those who, like me, do know how to read this kind of map. I need to consciously struggle with my brain's inclination to see a physical object, and not a probability range. Why? I don't know for sure, but I'll make a conjecture: it's because the representation looks pictorial. The rounded shape of the tip of the cone roughly resembles the shape of a hurricane.

This map is made even more confusing if a black line is placed in the middle of the cone. Just read tweets like this. People may see that line not as a visual aid to emphasize the center of the cone (right), but as the most probable path (wrong).

Going back to the caption, the reason why it sounds wrong to me is related to something most of you probably aren't aware of: the cone of uncertainty doesn't represent the range of all possible paths the hurricane could follow, based on simulations. This excellent paper explains that the most common cone, the one by NHC, “accurately predicts the ultimate path of the tropical cyclone’s center about 2/3 of the time (J. Franklin 2005, personal communication). In other words, one out of three storm centers directly impact areas outside of the cone.” That's a 66%-33% chance.

Therefore, the caption could say something like this: “Based on predictive simulations of past hurricanes, there are 2 out of 3 chances that the path of the center of the hurricane could be anywhere within this cone, and a 1 out of 3 chance it will be outside of it.” This is longer and clunkier —I'm sure that any copy editor in the audience can improve it!— but truer to reality.

This other map shows the actual uncertainty of predictive simulations quite well; notice the faded lines, corresponding to less probable (but still possible) paths:


UPDATE: It seems that NOAA is listening. See the explanation that they have been tweeting. It ought to be published next to every single cone of uncertainty map out there:


UPDATE 2: The map below, by meteorologist Ryan Maue, is far better than any cone of uncertainty map if your goal is to inform the general public about the risks posed by wind. See it animated. The scale is predicted maximum wind speed in mph.



How to have fun with visualization

I'm often amazed by the power of visualization and infographics to unveil truths hiding behind complexity. Surprise and wonder are emotions a good graphic may cause. However, I don't remember myself laughing out loud when exploring a data visualization.

This changed a few months ago, when my friend Xaquín. G.V. (Twitter) began working on a project that's part of my ongoing collaboration with Google News Lab. As you may remember, I'm art-directing a series of experimental visualizations by top designers from all over the world. You can see them all here.

Xaquín has been head of visuals at The Guardian, and a graphics editor at National Geographic magazine, The New York Times, Newsweek, and El Mundo, in Spain, where we worked together between 2001 and 2005. His virtues are many, but the one that you'll first perceive if you ever meet him in person is his wild sense of humor. Xaquín is a genuinely funny fellow, and that shows in his style. Just take a look at his project for us, titled How to Fix a Toilet, and the article explaining how he made it.

And, yes, as Xaquín wrote in that article, this is my favorite animation by far:


Wednesday, September 6, 2017

A single data point is often meaningless without its context

(Chart updated with a suggestion by Andrew Losowsky)

In case you haven't heard, we're bracing for a monster hurricane down here in Miami. While praying to the gods of uncertainty and chance to push it a bit to the East, back into the Atlantic Ocean, I decided to relax for 15 minutes from installing shutters and getting supplies by designing a quick chart. I'm offering it for free to Breitbart News.

This morning, Breitbart published a story by reporter John Binder with this alarming headline “2,139 DACA Recipients Convicted or Accused of Crimes Against Americans.” This is its lede:
As Attorney General Jeff Sessions announced the end of the Obama-created Deferred Action for Childhood Arrivals (DACA), from which more than 800,000 un-vetted young illegal aliens have been given protected status and work permits, the number of them who are convicted criminals, gang members, or suspects in crimes remains staggering.
It's a staggering number indeed. Staggeringly low. If there are roughly 800,000 DACA recipients, 2,139 of them are just 0.27%, or roughly 3 out of 1,000. The numbers in the story suggest that undocumented youngsters protected by DACA commit proportionally far fewer crimes than American citizens do. This is, of course, if the data are reliable. Breitbart mentions its source, USCIS, but doesn't link to the specific report this comes from.

To put these data points in context I made the charts below; sources are thisthisthis, and this. If you have other figures that would make for a better comparison, let me know. For instance, to be more accurate we'd need to get the felony rate just of Americans who were under the age of 31 as of June 15, 2012. This was one of the requirements to apply for DACA. Also, it might be the case (I don't know) that a conviction doesn't immediately lead to your DACA status being revoked.

Finally, we'd need to consider the rate of Americans who have been convicted of “significant misdemeanors” or are affiliated with gangs, as the second graphic only plots felony —but not misdemeanor— convicts up to 2010. This might make the difference even larger:






Saturday, September 2, 2017

Improving Kid Rock's map t-shirt

A couple of days ago, on Thursday, I gave the first Visual Trumpery talk in the United States. The public in Atlanta was wonderful, and the comments during and after the lecture were very inspiring.

People seemed to have fun with the discussion about when to use state-level maps, county maps, or cartograms to show results of presidential elections. The most celebrated moment was the constructive critique of a t-shirt sold by Kid Rock. My suggestion was to use a county map, as the borders between the U.S. and Dumbfuckistan aren't accurately depicted at the state level. You can see the slide below.

(Read more about the Visual Trumpery tour here. Go beyond the talk title, and read its description, as the title is intended to trick you. If you want to sponsor a talk —I won't charge salary,— here's how.)



Tuesday, August 29, 2017

Visualizing the German elections

The latest project in our ongoing Google News Lab visualization series is based on searches for candidates in the German presidential election. This is the second time we've collaborated with Moritz Stefaner, after his very successful The Rhythm of Food and, as in that case, my role as art director of the series was limited to a few suggestions here and there.

This interview with Moritz explains how the project was done.

To learn more about the goals of this series of experimental visualizations, read this article at FastCo Design. You can see all our previous projects, articles, and hangouts here.