Make open data actually work
- By Christine Carmichael
- Dec 03, 2014
The Digital Accountability and Transparency Act creates a monumental challenge for federal agencies. Government leaders, watchdog groups and citizens want to make sure tax dollars are being used in productive, efficient ways free from waste and fraud. They also want proof that publicly funded programs are performing well against stated missions and executing within budget.
The only way to know is by seeing the proof: the data itself. Government agencies can't just analyze and manage data. They must also share it in ways that are insightful and useful. The demand for government transparency and accountability is here to stay.
But open data should be viewed as an opportunity as much as a challenge -- an opportunity to encourage greater citizen participation and save precious budget dollars in the process. In fact, open data could be a huge money-saver. A McKinsey and Co. study has suggested that open data could allow government agencies to recover a combined $3 trillion a year or more. But those savings won't happen automatically.
The following suggestions, from Tableau Software's white paper "A New Era of Transparency: 5 Best Practices for Open Data in Government," can help any agency create data portals for consumers that are easy to use and transparent. Any agency that adopts the following guidelines will have a leg up in achieving true accountability and transparency.
1. Make everyone a data analyst. It's important to embrace data analysis beyond the IT department or data analysis experts. Thanks to increasingly user-friendly software, anyone can use data -- from executives on down. Such software is designed with the strengths and weaknesses of the human visual system in mind, unlike hard-to-read spreadsheets and other cumbersome tools. Such software makes data analysis democratic, and it brings increased speed and efficiency.
For example, at the Department of Veterans Affairs, data analysis tools have enabled employees to get a handle on data without waiting for cumbersome reports from the IT department. Tommy Sowers, former assistant secretary of public and intergovernmental affairs at VA, launched a national outreach program in 2013 that included a data analysis project to locate areas with large clusters of veterans.
Thanks to new business intelligence tools, everyone at the agency could access the data. Sowers said such software allows people to "dive into the data and really explore without interruption or the need to wait for another report to be run."
2. Showcase relevant, actionable information. "Open data" is a broad term, but it doesn't mean you can just dump all data on consumers willy-nilly. You must make sure your agency is using the most relevant data and, more important, providing ways to access data that any consumer or decision-maker could manage.
One way of gathering data is via a mashup. Mashups are a popular trend in books. Highlights include "Pride and Prejudice and Zombies," which merges a literary classic with the horror genre, and "Two Gentlemen of Lebowski," which retells "The Big Lebowski" film as a Shakespearean play.
Mashups are important to data as well. It is a powerful method that involves bringing more than one data source together. Data analysis software allows for many types of mashups, which diversifies the results and content of your data portal. The results aren't as weird or extreme as in literature, but they are extremely useful for federal agencies.
3. Use data visualization to tell stories. In a similar vein, agencies should make a habit of using dashboards to tell a story. People think in stories. We're used to hearing them from infancy, and we keep hearing, seeing and telling stories our entire lives. So it makes sense to find the story in your data and present it. A strong narrative -- which can be reinforced with a mix of text and images along with charts and graphics of various sorts -- can make data come alive for users. Storytelling is the key to communication. Don't neglect it when dealing with data.
For example, a data visualization of crimes in Hartford, Conn., doesn't simply tell people about crime, it shows exactly what types of crimes -- such as robbery, homicide and aggravated assault -- have occurred and where such crimes tend to cluster. People can also see what percentage of those crimes are open cases and which ones have led to arrests of adults or juveniles. The dashboard allows for a high degree of interactivity that could be implemented on the federal level, which leads to the next point.
4. Allow for interaction and collaboration. By suggesting you select the most relevant information and use that information to tell a story, it might sound as though data visualizations are one-way info dumps. They shouldn't be. Instead, you should incorporate options and choices for users so they can adjust information and follow the data down the path they choose. This allows the narrative to be influenced by the reader, much like a Choose Your Own Adventure story.
For example, if part of your data is geographical, a person should be able to access the data that most affects his or her location. Depending on the data, similar choices might involve someone's gender, age, income bracket or profession. If you think ahead to what people want, you can tailor the choices accordingly.
For example, Tantus Technologies works with the Federal Aviation Administration and other agencies to develop dashboards that analyze and visualize financial data. With the help of Tantus, agencies are better able to understand their financial picture, but the understanding isn't a one-way report. The tools allow for real collaboration rather than the real frustration of trying to manipulate and understand a spreadsheet. Any aspect of the FAA's finances can be explored by the agency's employees.
5. Make all of the above accessible. You could create the most useful, user-friendly, narrative, collaborative dashboards in the world, but if they're hard to find or get into, you have wasted time, money and energy. Make sure the fruits of your labor are easily accessible for consumers. Don't bury the data in a labyrinth of confusing links. Make it prominent on your homepage and as easy to find as possible.
Also, seek feedback on every aspect of how you're presenting the data. Is it as easy to access as you had hoped? Is crucial information missing? What could improve the experience for users?
Making great dashboards is rarely a one-and-done process, and feedback is essential. If you encourage -- and use -- feedback from the people who use your data, you'll multiply your chances of creating open, transparent and successful data portals.
Christine Carmichael is marketing segment manager for the Public Sector Group at Tableau Software.