Learning from YOU - Test-driving Census Reporter BETA at ONA
By Sara Schnadt
The Census Reporter team was present in force at the Online News Association conference two weeks ago in Atlanta. We all came out to show off the newly launched beta site and talk to attendees one-on-one about Census Reporter. We were there as part of the Knight Village area at the Midway, which featured some of the most exciting new technology projects in journalism in a hands-on exploratory setting. In addition to demo-ing the site during office hours, doing a lightning talk, and presenting on a civic data panel, we spent most of our days holding down the fort at the Knight Village talking to journalists in-depth about their work, their use of Census data, and watching them try out our new site.
Project Lead Joe Germuska talking to Claudia Núñez about Census Reporter at the ONA13 conferenceThis process was invaluable, and put our work of the past several months into rich perspective. We have approached the site build process iteratively, based on extensive requirements development from interviews with journalists, and developing in the open. And, we have been working intentionally towards ONA as a half-way mark, creating only the core site functionality so that we could take the opportunity to plan the fully-fleshed-out site based on feedback from journalists responding to our work.
Community Liaison/Designer Sara Schnadt and Back-End Developer Ian Dees ready to demo the site.
So, having everyone on our team there in person at ONA to hear first-hand how the site is helpful, where the interface might need more work, and what else journalists would like to see there, was very informative to how we are conceiving of and building the site.
To make the most of being together in person (we are usually distributed around the country), Joe Germuska, John Keefe, Ryan Pitts, Ian Dees and met the day after the conference to make plans for the second half our build based on the feedback we had all heard. We were pleased to learn from all the folks we talked to that we are generally on the right track, there are no major features we are missing, and that this tool will make their work genuinely so much easier than it is now. We heard this consistently from shoe-leather reporters, app developers, data journalists, journo nerds, people with small regional beats and people with national beats.
After three days at the Midway at ONA, we had two main takeaways. The first was that the Profile a Place page is great, but we should create three pivot points for it: data over time; this same data for multiple custom places; and more in-depth and interactive data on this place. The second takeaway was that the Compare Places tool needs a new interface.
Other good ideas came out of the demo process as well, some confirming our own plans. These include: make the maps clickable as a way to navigate to data about this place or another place; include AP-style text about each statistic and visualization on a page to make the data more real and relatable (and easy to drop into a story); make all visualizations everywhere embeddable; put small contextual ‘spark line’ style visualizations next to key data points rather than national percentile #’s; always show margin of error on numbers; generate interesting ranked lists on topics and places; make something resembling a google search for the site for both finding a place and a topic; and, recommend interesting data on places that are outliers on a topic for lead generation purposes.
As we begin to build new features and improvements for the Census Reporter site, we encourage you to tell is what you think. All the ideas mentioned in this post are included in our new Feedback Forums (where you can comment and vote on them). You can also suggest new features, and give your feedback on the current site.
DETAILS ON WHAT WE LEARNED FROM JOURNALISTS
Profile of a Place View
Top area of Census Reporter ‘Profile a Place’ page with data for Los Angeles
Everyone we talked to found our profile page to be the most helpful and user-friendly part of the site so far. It features topical top-level information and visualizations for any geography in the US. We also got some great suggestions for how to make it better. With few exceptions, journalists wanted to be able to compare trends over time for the statistics and visualizations on Profile of a Place. Most journalists also expressed an interest in easily comparing one place to another, either a neighboring place, a larger geography (their city plus the whole state) or another place in the country (e.g. New Yorkers comparing their place to other large metro areas).
Another section of ‘Profile a Place’, with data for Los Angeles.
There was also an interesting trend that one insightful journalist pointed out to me - she learned this from her readers. When you live in a smaller place, you are more likely to want to compare your home to neighboring towns and the immediate region. But if you live in a larger place, you set your sights on a larger geographic area (other cities like yours around the country) for contextual statistics for your place. I also learned that readers in smaller places are frequently looking to localize the kinds of statistics they hear about in the national news, to bring them home and understand how they impact their community. Building a very intuitive geographic comparison tool as part of the Profile a Place page will make it possible to get all these kinds of contexts.
Other journalists who are more data-analysis inclined expressed an interest in digging deeper into the topics on the Profile Page, getting more details and more data points on the main topics (Demographics, Economics, Families, Housing, and Social) as well as cross-referencing and sub-dividing the information. For example, one journalist wanted to see how the educational attainment or poverty level visualizations would break down by race.
The number of topics and themes running through census data are quite vast. We curated them for this page by surveying journalists, calculating the topics that have highest frequency in the data, looking at the Census QuickFacts site, and by reviewing the kinds of issues that frequently come up in articles written from census data. While people we spoke to in person were generally happy with this curation, many wanted to see more detail and more data points on each main topic. We would also love to hear your thoughts on this.
Compare Places View
Current ‘Compare Places’ interface for picking census topical tables
The other main feature we built to try out on journalists at ONA was the Compare Places tool, designed to help journalists identify a topical table from the nearly 1500 tables available in the Census’ American Community Survey. As it is now, Compare Places helps the user filter through all of these tables and narrow down to a browsable quantity to choose from by combining a keyword search and topical filters. In addition to making the process of getting at a specific table for download significantly easier than it is with any existing census tools (we have received some rave reviews so far from app developers - and new apps have already been made), this tool lets you explore the data as well.
You can view the data as a table or as a choropleth map that lets you pick any row in the table and view it geographically. Many journalists I spoke to were excited about this view for its lead generation potential, and spent a lot of time playing with it to see nuanced and contrasting dynamics in various data (such as children in poverty or mobility of immigrant populations) related to their beat or state. Some patterns visible in the data confirmed what they already knew and some were surprising, speaking to the exploratory value of this kind of visualization.
A choropleth visualization of a selected topic table in the ‘Compare Places’ tool
There is also a distribution chart view into each table which allows you to see which geographies in your selected area are more and less similar to each other on each data point or detailed topic in the table. Mostly we got a simple ‘wow’, and ‘fascinating’ response for this visualization so far. But it can be quite useful for more in-depth analysis.
While Compare Places works well functionally as it is, and is extremely powerful, pretty much every journalist we showed it to found it difficult to use. This is essentially because its search logic is built to reflect how the data is structured rather than how journalists think about beginning to research a census topic.
Chris Amico, known for his work on Homicidewatch.org, stopped by our table and talked with Ryan and me. He broke it down for us like this: when he wants to write a story or build a visualization about Census Data, he starts with either one of two simple questions - what place do I want to find out about, or what topic? And so our interface should start out as simply as this. He even white-boarded with us:
Idea for primary site interface created in conversation with app developer/journalist Chris Amico during site demos at ONA
This conversation confirmed the disconnect we were getting from journalists. When we ‘drove’ for them, we were able to get the answers to some of their more complex questions, but when they tried to navigate this interface themselves, even the more technical journalists would get puzzled and stuck quite quickly. And so we are taking the user-facing piece of this tool back to the drawing board to re-imagine it.
More Ideas
Other good ideas came out of the demo process as well, some confirming our own plans. These include: make the maps clickable as a way to navigate to data about this place or another place; include AP-style text about each statistic and visualization on a page to make the data more real and relatable (and easy to drop into a story); make all visualizations everywhere embeddable; put small contextual ‘spark line’ style visualizations next to key data points rather than national percentile #’s; always show margin of error on numbers; generate interesting ranked lists on topics and places; make something resembling a google search for the site for both finding a place and a topic; and, recommend interesting data on places that are outliers on a topic for lead generation purposes.
Next Steps
Another blog post is coming soon with details on the new build plans we have made. Our next goal for the project is to finish the Profile a Place section by the end of the year, and then to complete all of the main site features by the NICAR conference in February of 2014. So now is the time to give us your thoughts. We are building this site for you!
