Data journalist Mona Chalabi is on a mission to, as she puts it, "take the numb out of numbers." In her illustrations, animations and articles for the Guardian US and publications like Fivethirtyeight and The New York Times, she explores data sets from the timely (voting trends, race, politics) to the offbeat (popular dog names in New York City) to the eye-opening (how many Americans eat pizza for breakfast).
Chalabi will be one of this year's keynote speakers at the DRIVE/ conference, being held in Baltimore, Md., March 11-13. Below, she gives a preview of her keynote and chats with CASE about what makes a great visualization, how to spot a bad statistic and the big mistake too many people make when working with numbers.
CASE: You studied international security in school. How did you find your way to data journalism?
Mona Chalabi: My first big job was doing monitoring and evaluation for the International Organisation for Migration's Iraq office, which at the time was based in Jordan. The work involved analyzing data and then writing up reports about what refugees and internally displaced people needed. But I quickly felt a bit disillusioned about the way that such important numbers were being communicated so I did a one-day workshop in data journalism in London. It was taught by Simon Rogers who was the data editor at the Guardian at the time. I asked if I could do work experience at the Guardian and the rest is history!
How do you decide what data to dig into for your illustrations? What intrigues you?
I often keep a close eye on the news and try to create work that helps people make sense of the information that's coming at them. I also check academic blogs for new studies but very often, the ideas are suggestions that from very smart readers who get in touch to ask me something that's been bugging them!
What tips do you have for making data digestible?
Keep it as simple as you can. Ask people who don't know anything about data what they understand from the sentences/images you have created. Did they understand it correctly?
What makes a data visualization good or helpful?
First and foremost, the data behind the visualization should be reliable. Then the visualization itself should clearly communicate those numbers. If it's a memorable image, that's great. If it's beautiful, that's even better.
What's a big mistake a lot of people make when writing about/using data?
People often overstate the precision of the numbers. Don't put a decimal place on something like polling numbers: that suggests that you have a high degree of precision in your methodology, which just isn't true.
In your TED Talk, you say people need to be better at spotting bad statistics. What are a few ways we can do that?
Stay skeptical. Ask yourself who collected the data, when they did it and if they have any incentives for reaching certain conclusions.
What makes working with education data—student debt, student diversity, graduation rates—challenging?
I've done a few pieces on educational data. I've looked at the impact of affirmative action and how unemployment rates vary by educational status. I think one of the challenges is that in a country as huge as the United States, national figures really don't necessarily help people understand the state of education where they live. But it can be hard to show statistics for each of the states and break it down by things like race or gender.
What metrics matter most?
I care about fairness. That's why I first became fascinated by data, because it's a useful way to show patterns of systematic injustice. Those are the metrics that matter most to me: the ones that show whether the U.S. educational system is producing outcomes that are fair.
What do you think is the future of data journalism? What should we expect to see in the next few years?
I hope there will be much more of an emphasis on creating more accessible data visualizations. This year, I want to think about ways to create work that is fully accessible to people with disabilities, including the blind and visually impaired.