Diagrams and Foreign Languages

Quote

For an image to be useful, the relevant or key results should be immediately obvious, so that the viewer need not spend a great deal of time first trying to understand what might be important. Diagrams can themselves be complex, requiring some time of study before they’re understood, but the basic idea being expressed needs to be immediately apparent. It is similar to hearing someone speak in a foreign language, where even if it’s not possible to understand the language (the specifics of the diagram) it’s generally possible to tell whether they’re angry or happy (what parts of the diagram are relevant and require further inspection).

– Ben Fry, Dissertation: Computational Information Design, 2004

Pop Experiments

I drink a lot of soda. Maybe not as much as this woman (let it be a warning!!), but still, Diet Mountain Dew is a substantial part of my daily liquid intake. So of course I was intrigued when I ran across this dataset on soda consumption since the 1950′s (it actually inclues all beverages, but one track mind here). I decided to make an interactive chart out of this very significant information. I had just learned how to use FasterCSV and a little Ruby to turn a .csv file into a bar graph, so I used the same approach. Basically I ended up creating a bunch of little divs, their heights calculated as percentages of the total height of the container div. Also learned about jQuery .prepend method, which let me attach labels – the gallons of soda per person – to each bar. Here it is! 

Hey girl, it’s data

So last week at Science Online 2012, the amazing Ruth Spencer and I rambled a bit about data journalism. The main product – still evolving – is this tumblr blog, which is our attempt to gather our thoughts and lots of resources for people interested in getting started in collecting, interrogating and visualizing data. It could have been called “Everything I know about data journalism I learned from ProPublica and the NY Times.” But we decided to call it Speed Data-ing. Yes, it took us many iterations to come up with that one.

This was one alternate title:

But this was the REAL winner, which came to us via science writer extraordinaire Ed Yong a little too late to use it this year. We are assured that’s it’s gonna be an option for #scio13.

Experiments with Google Fusion Tables

To prep for our talk on data journalism at the Science Online conference coming up next week (!!), I figured it’d be good to actually review mapping some data with Google Fusion Tables. I chose a dataset released this October from the CDC on suicide from 2008-2009. Morbid, I know, but it’s is the first time this specific state-level data has been reported, and I wanted to actually visualize the differences between states rather than stare at numbers in a table.

The numbers originally came in a PDF that looked like this:

After thinking for a little while that I’d have to do something ninja-like to get them out of the PDF, I just selected all the numbers in the “Attempts” % column, and pasted them into a Google Spreadsheet. Not bad.

Then I switched over to Google Fusion Tables, and created a new table by importing that very simple Google Spreadsheet. Merging that with the pubic KML file for US states gave me this:

Next was going to Visualize > Map, fiddling around with Configure Styles and adjusting the HTML in the Configure info window to show just the state name and percentage.

Hardly beautiful (and I gotta work out much more coherent step-by-step instructions) but here it is:

Suicide Attempts by State (in % of population)

Now the REAL story here, which the map indicates but does not explain, is why Rhode Island in particular has such a high percentage of suicide attempts. The CDC report offers some speculation, but no definitive answers:

Several theories have been advanced that attempt to explain these regional differences. These theories attribute the differences to selective migration (i.e., populations with risk factors for suicidal behavior might migrate to the same geographic areas), to the sociodemographic composition of the population (i.e., certain regions have a greater percentage of persons who are members of sociodemographic groups that are at greater risk for suicidal behavior), or to the local social environment (e.g., unemployment levels, divorce rates, social support, or environmental factors such as increased access to lethal means). Such variations might point to areas of emphasis for planning and evaluation of prevention activities.

 

To be continued, then.

Which half? (on the importance of controls)

Quote

One day when I was a junior medical student, a very important Boston surgeon visited the school and delivered a great treatise on a large number of patients who had undergone successful operations for vascular reconstruction. At the end of the lecture, a young student at the back of the room timidly asked, “Do you have any controls?” Well, the great surgeon drew himself up to his full height, hit the desk, and said, “Do you mean did I not operate on half the patients?” The hall grew very quiet then. The voice at the back of the room very hesitantly replied, “Yes, that’s what I had in mind.” Then the visitor’s fist really came down as he thundered, “Of course not. That would have doomed half of them to their death.” God, it was quiet then, and one could scarcely hear the small voice ask, “Which half?”

—Dr. E. E. Peacock, Jr., University of Arizona College of Medicine; Medical World News (September 1, 1972), p. 45, as quoted in Tufte’s Data Analysis for Politics and Policy

Should I get a bone scan? The flow chart

I don’t love flowcharts. Most of the ones I’ve seen are totally useless or just silly. But I think the simple form of a flowchart can actually be pretty useful for explaining a complicated subject (that is, when that subject isn’t beyond all hope). This one on campaign finance, for example, is easy to follow and packs in a decent amount of info into a relatively small space.

Here’s one I did for a health reporting class on bone scans (and whether or not you need one). The current recommendations are tucked away in obscure reports, and according to the doctors I talked to, many people who should be getting bone scans aren’t. The chart’s not great, but it’s one attempt to lay out the information. Just for fun, I made it in the style of a typical Wired flowchart.

Here’s the bigger version.

 

When design goes bad

A long time ago I read a book that changed how I looked at, well, basically everything. That book was The Design of Everyday Things by Donald Norman. It’s packed with examples of bad (and some good) design – everything from objects to software to architecture. The basic idea is that the user has different interests than the designer, and when they come into conflict, you end up with unusable products and lots of errors.

Take the door. Seems simple enough. But how many times have you pushed a door when you needed to pull it open, tried to turn a door knob when you needed to slide it, or just plain walked into a glass door you didn’t see (ok, fine, that was me).  I’ll always remember one line from the book, which went something like: “When something as simple as a door has to come with an instruction manual – even a one-word manual – then it has failed.”

Along those lines, my latest rant about poor product design is over at Scientific American. This time, it’s about the hazards of the instant soup cup. The soup cup might not be quite as bad as the Coffeepot for Masochists on the cover of Norman’s book, but it certainly comes close.

IMAGE CREDIT

Memory on trial

The two-week countdown to graduation has begun (wha?!?) which means equal parts celebration and hyperventilation. Next up, I’m super excited to be starting an internship at ProPublica – a total shift from the psychology/neuroscience news realm I’ve been operating in the past few months. In particular, I’ll be writing more explanatory type pieces, like this one on the failures of eyewitness identification (I originally sent this piece in as a “sample” explainer; it’s now been updated slightly since October.)

And yes, OK, I clearly failed to pull myself entirely away from psychology (workin’ on it!) But the uncertainty of memory is so fascinating – and has so many legal and political consequences – that I couldn’t resist. The issue won’t be going away any time soon.

Here’s the story:

The trouble with eyewitness identification, explained

“I saw it with my own eyes.” There is little more convincing than the confident declaration of a witness, absolutely sure of her own memory. Eyewitness testimony has long enjoyed a lofty status as “the gold standard of the criminal justice system,” the real stuff of proof.  Yet it has also been decisively dethroned by modern science.

While scientists of the mind have long questioned the reliability of eyewitness testimony, it took the public spotlight on Troy Davis earlier this fall to cast more than a reasonable doubt in the minds of many. But how much do we really know about eyewitness identification, and what’s being done about it? Here we sort through the science, the law, and the future plans for this fundamental pillar of the criminal justice system.

What does the science say?

As TIME, the New York Times, and Slate note, there are all sorts of factors that affect how accurately witnesses can identify a suspect: how far they were from the scene, whether it was daytime or nighttime, whether they were under the influence of drugs or alcohol, whether the person they identified was of a different race. After the crime, other factors can skew a person’s memory during the identification process: how much time has elapsed since the crime, suggestions from the police officer, having seen photographs of the prime suspect, or even just wanting to believe a certain story. Over 30 years of peer-reviewed research on eyewitness memory has found that it’s pretty much inherently unreliable.

The problem is, we don’t believe it is. We tend to think that memory, especially our own, is just fine, thank you very much. As reported in Scientific American, 63 percent of US adults think that memory is about as good as a video camera. And according to the American Psychological Association, jurors tend to over believe eyewitness accounts. But the fact is, our memories are far from perfect.

How does this affect the criminal justice system?

In a big way. According to the Innocence Project, eyewitness misidentification is “the single greatest cause of wrongful convictions nationwide.” In 75 percent of cases where a suspect was found innocent because of DNA evidence, it was eyewitness testimony that put them behind bars in the first place. All sorts of problems can crop up, from discrepancies in a person’s description to problems in the line-up procedure.  The Virginia Law Library has a chart showing all the ways eyewitnesses have been mistaken in cases where the suspect was later exonerated.

And it’s not just that innocent people are convicted. As Santa Clara County Assistant District Attorney Karyn Sinunu-Towery tells TIME, “every time there’s a wrong identification, it means the real criminal is still out there.”

What are some easy ways to help fix the current system?

According to Nancy Skeblay, a psychologist at Ausburg College who researches eyewitness accuracy, we could see major improvements by simply switching the way a police lineup is done. Traditionally, witnesses are shown a row of photographs all at once and asked to pick out the culprit. But research suggests that when witnesses are shown each photo one at a time – a “sequential lineup” – they make fewer errors. It’s also important for the lineups to be double-blind, or administered by someone with no knowledge of the case. That way the officer won’t unintentionally sway the witness towards picking a particular suspect.

Finally, merely telling jurors about the well-known problems with eyewitness accounts may dampen their confidence in the testimony. The Innocence Project also has a list of other reforms that could improve the system.

So what’s being done about it?

Sadly, not too much, but that may change. Right now only ten jurisdictions across the US (New York is not one of them) use the recommended double-blind sequential line-up, and most don’t have strict regulations for how eyewitness accounts are handled in court. But this summer, the New Jersey Supreme Court set new guidelines for how its courts assess eyewitness identifications (here’s WNYC’s comprehensive podcast on the subject). New Jersey could serve as a model for other courts looking to make similar changes.

And for the first time in 33 years, the Supreme Court is hearing a major case about the place of eyewitness testimony in the criminal justice system. As this Slate piece makes clear, Perry v. New Hampshire may not be the ideal forum for a landmark decision (the high court looks like it will avoid confronting the issue head on). But given the explosion of commentary on the subject, it seems as though the failure of memory in the courtroom is finally getting the public attention it deserves.

 

Some updates

Well, to celebrate the revamped home of this blog (and also to catch up on a few months of sheer and utter neglect), here are a couple things I’ve been up to.

July: Wrote about open source intelligence, predicting war with physics, biological clocks, and educational mobile apps. Got to mess around with some data and use Illustrator in class (win!)

August: For some reason, this month was overrun with silly combat outfits and military uniforms. “Real journalism!” as my editor Noah Shachtman would say.

September: Probably the most difficult (and exciting) project I worked on all summer was a series of infographics about the transformation of the US military from 9/11 to today. (Lesson: dozens of calls to the Defense Department do eventually lead to something.) The numbers are sobering.

October: Switched over to Scientific American MIND, exchanging military technology for social behavior and fingerprint genetics.

November: Attended SfN’s Neuroscience 2011, just about the most overwhelming science conference you could imagine. 31,000 people. Neuroscientists swarming around everywhere. It was fantastic. The first morning, after about four and a half cups of coffee, I decided to pick a random topic (music! why not?) and try to see all the relevant talks/posters. I didn’t even get that far (who knew the auditory section of the poster hall would be a whole 6 rows?!) but learned some quirky facts about bilinguals and pitch.

Hopefully I’ll post again before another five months go by…

IMAGE CREDIT