CDC chart on covid hospitalizations

How data visualizations drive decisions

Since the COVID-19 pandemic began, data visualizations have been a key way for government agencies to communicate information on cases, deaths, unemployment rates and personal protective equipment shortages. This data has driven shelter-at-home orders and is now helping governors decide when and how to reopen states.

Not all data is presented equally, however.  We asked Amanda Makulec, senior data visualization capability lead at Excella, to explain what to look for – and what to look out for.

This interview has been edited for length and clarity.

GCN: Government agencies having been pushing for data-driven insights for some time, but it seems that the COVID-19 pandemic has made data a star. What have you seen in terms of data reliance?

Makulec: Data reliance … speaks to that need of “How do I use the data for decision-making?” I’ve found it almost ironic that COVID-19 has become the moment when we decided to rely so fully on this data and look to it both to inform what we do as organizations or in policy, but also in our own individual lives. The interest in data, the enthusiasm for data, is something I hope we can sustain.

What makes me nervous is that that hasn’t really come with an equal enthusiasm for understanding the nuances and challenges and, in many ways, uncertainty and incompleteness of the COVID case data. When we place something on a dashboard or on a chart or a graph and we publish it, we imply that it has a certain degree of certainty, “truthiness” and reliability … and there’s research showing that that causes people to believe it more than if they just read the same statement in text.

Information changes rapidly as the pandemic evolves. Are agencies at risk of providing misleading information in the name of speed?

I think we’ve been conditioned to think we can have this kind of instant access to information, but instant comes at some cost. It comes at the cost of data quality and the appropriate number of data quality checks. It comes at the cost of technology and what it takes to stand up the technology and the systems and the pipelines to do that. It really requires a cross-functional team -- not just your data visualization developer to also be the person to build your data pipeline and advanced machine learning -- but really thinking about the specialized skills [needed] and how you put a team together to build things that deliver closer to real-time information.

What are agencies getting right with their dashboards, and what tips can you offer for them to make sure that the information they’re disseminating is accurate and informative?

The things I really admire are when I see clear definitions of what they’re including in the numbers, so a confirmed case vs. a confirmed and probably-confirmed case count combined together. Having different definitions for metrics means that we can’t compare those metrics necessarily across different states or across even different countries. Having a clear date and time stamp when the data was updated so I understand and know how recent that data is. Finding ways to add context and try to smooth the data a bit. Early on you saw a lot of bar charts that were [showing how] every single day’s case count was going up and up and up. Actually adding something like a three- or seven-day rolling average can smooth out some of the noisiness that isn’t likely due to a big change in the burden of disease or infection, but rather due to just reporting variance. Help me understand not just the number of cases, but also what the outcomes of those cases were: whether the case resulted in a death, whether they recovered, whether we do not know – and do not know is an important category to think about.

What are the risks associated with getting public health data wrong?

There’s a lot more risk to putting out data and numbers that seem like they are facts when in all actuality they only represent part of the story. Inevitably as we move through the phase of being in this emergent pandemic and learn more about this disease and perhaps recommendations change, I think you really need to have faith that the government bodies advising you on the policies they’re making and the policies you should follow are government institutions that you trust. Having transparency around data is one way of imbuing that trust.

About the Author

Stephanie Kanowitz is a freelance writer based in northern Virginia.

Featured

  • automated processes (Nikolay Klimenko/Shutterstock.com)

    How the Army’s DORA bot cuts manual work for contracting professionals

    Thanks to robotic process automation, the time it takes Army contracting professionals to determine whether prospective vendors should receive a contract has been cut from an hour to just five minutes.

  • Russia prying into state, local networks

    A Russian state-sponsored advanced persistent threat actor targeting state, local, territorial and tribal government networks exfiltrated data from at least two victims.

Stay Connected