U.S. police shooting data visualization: WaPo vs The Guardian

Perhaps the biggest challenge in evaluating the true nature of the problem of police shootings in the U.S. is the lack of national/official data on them. How do you honestly and legitimately evaluate something you aren’t measuring?

Thanks to Jeremy Singer-Vine’s Data Is Plural newsletter, I now know of two news organizations trying to remedy that data gap. The Washington Post and The Guardian have both published independent databases and data presentations. Looking at both of them, I can’t help but compare them and notice differences in the data presentation. While I’m very glad that both news orgs are collecting information on this important topic, it’s pretty clear to me that The Guardian’s presentation of the data is both more effective at conveying the insights that are in the data, and easier to use overall. Here’s what leapt out at me on first browse:

THE GUARDIAN

http://www.theguardian.com/us-news/ng-interactive/2015/jun/01/the-counted-police-killings-us-database

The Guardian makes a lot of excellent choices that sum up to a very effective presentation.

Good: Compact summary up top

The_Counted__people_killed_by_police_in_the_United_States_in_2015_–_interactive___US_news___The_Guardian

Pretty much everything I want to know at a state or national policy level is in that summary. As with all news, it’s great to start with the most important stuff.

Great: Emphasizes apples to apples comparison

I especially love seeing that when the interactive page loads, it’s set to show normalized rates rather than absolute numbers per state or race (via the “per capita” and “per million” buttons). One of my pet peeves is comparisons that don’t take population differences into account. We’d expect California, for example, to have more total shootings than Wyoming because it has a lot more people; but the rate of shootings per person is actually higher in Wyoming.

WY_CA_shootings

I also love the choice of a persistent color coding on the state tiles that uses a gradient to show relative rates. The difference between 28 shootings per 100,000 people and 31 shootings per 100,000 is basically noise – so the detail of absolute ranking is not as useful as a more general low-medium-high comparison like the gradient provides.

I love the way the buttons highlight the difference between oranges to apples and apples to apples comparison:

Shootings_race

While the emphasis in popular media has been on black vs. white victims, ala
http://mappingpoliceviolence.org/, it’s clear from the Guardian data depicted immediately above that Hispanics and Native Americans are also overrepresented among police shooting fatalities (though not nearly as dramatically as blacks).

Great: Presenting victims as both individuals and statistics

The_Counted_individuals

The combination of tiled image presentation with basic stats (name, age, state, manner of death) emphasizes both the victims’ individual humanity and their collective representation of terrible statistics. It suggests a calendar of violent deaths, a sort of shooting-a-day rhythm (a grim and effective visual concept).

THE WASHINGTON POST

https://www.washingtonpost.com/graphics/national/police-shootings/

There are several poor design choices in this presentation that make it substantially less effective than The Guardian’s.

Not so great: Emphasizing absolute numbers

WaPo leads with this:

Lede_Investigation__Police_shootings_-_Washington_Post

That’s a crap ton of space used to convey almost no information. Don’t get me wrong; I’m a fan of whitespace, and especially using it to emphasize something important. But that number doesn’t mean much without a lot more context. How bad is that? How does that number compare to other countries? Is that number rising year-to-year (not a lot of year-on-year data for this topic, unfortunately)? For those of us who actually live in the States and don’t think of the entire country as an undifferentitated blob, where are those deaths occuring? Are they close to me? How does my home state or city fare?

The Guardian’s summary does a much better job of conveying the shape of the problem.

Not so great: burying infographics in a horizontal slider with obscure icons

After wasting so much space up top, WaPo buries their most interesting graphics inside a horizontal slider.

Did you catch that? There’s a slider in the middle of the page. Maybe it’s more obvious in the mobile version where the arrows aren’t so much smaller than the content.

The Guardian uses tabs with text labels (“map”, “list”) to make it more obvious what the various visual options are. WaPo invented some icons to help you toggle between the different visuals:

WaPo_icons

Do you know what a smushed pixellated US map means? Nope, I don’t either. As a web development and design professional, I am a member of the “Look closely, click everything and find out what it means” club, but I’m pretty sure most web users aren’t that pokey. If your icons need a legend or if someone needs to click on them to deduce what they mean, you should probably just use text labels instead of icons.

Not so great: combining filters with data table

I’ll bet someone thought it would be clever and resourceful to make the data table double as a filter for everything else. Like many clever things, it turned out to be confusing, and it creates more problems than it solves. Now the filters are somehow both huge and obscure.

filter_dta

Did you even realize that clicking on the “female” data would filter to female? No? I rest my case.

Terrible: displaying monthly data in a way that makes it difficult to compare months

monthly_shootings

Why the hell would you do this? To save on vertical space? Is there an editor at the WaPo who still thinks people don’t scroll on the web? Display the months on individual lines so they’re easier to compare:

Chart5

It’s not really useful to show the number per month before you have at least two years’ worth of data – you can’t exactly infer seasonal trends from a single year. But it looks like WaPo built this thing to last and update over several years, so the monthly breakdown will become more meaningful over time.

Interesting: Squaring the states for an apples-to-apples comparison with geography in play

square_states

One problem with using a U.S. map to display rates as color-coded states is that it includes some information that’s not helpful and potentially distracting: land area. That is to say, if Texas and Rhode Island both have really high crime rates, Texas shows up as a big old lump of bad and Rhode Island barely shows up at all.

The Guardian avoids this issue entirely by displaying states only as tiles divorced from their geography. WaPo’s map of the states as equally sized squares attempts to address this issue and skirts the need to magnify Rhode Island and the other small states.

The one additional takeaway you really get from the squared states map is that the higher rates of gun violence are in the Southeast and West. There are some hitches to this ride, though. Something about the conversion to squares makes Kansas and Oklahoma look like part of the Southwest.

Thoughts?

The more important and controversial the topic, the more important it is to get the data and the visual data stories right.
What do you think? Who did it better, and why? Which choices work well and why or why not?

Leave a Reply