Better decisions depend on data, but only if people can actually use it

Terry Gerton  We are launching a new monthly series. We’re going to call it The Data Dish, where we talk all about federal data. So Denice, you come from that background. When you look across government right now, what do you see as the state of federal data?

Denice Ross I see that federal data capacity has been incredibly weakened, and some of that’s intentional and some it is really just collateral damage from the loss of so many federal employees, the contracts that support data collection and publishing, and the termination of federal scientific advisory committees that are so crucial for keeping our data collection systems modernized. There also is additional red tape, so things that used to be sort of routine for data stewards inside of federal government to do now require more clearance from political leadership. So, although we’ve seen some pretty high-profile terminations of data sets, including the food security supplement from USDA that was terminated right as millions of Americans were about ready to lose their SNAP food assistance benefits, the larger threat right now is really just the loss of capacity.

Terry Gerton And Chris, that brings us to the Federal Data Field Guide. What problem is that trying to solve? What do people struggle with when they actually try to use the kind of data that Denice has just described?

Chris Marcum Yeah, thanks for that question. So the Federal Data Field Guide, which Denice and I published in May of this year, it solves this sort of problem of how people, really any wide set of stakeholders, could learn about federal data. And often federal data are taken for granted. They’re integrated into lots of all of our daily lives, like the weather data, for example, or the navigation data on your phone, your mapping apps use federal data. And the Federal Data Field Guide provides a really fun, whimsical way for anybody to get familiar with really eight different categories of federal data, and it’s a fun exercise. And I think part of what’s happened with the losses in federal data capacity that Denice had just talked about, we feel a sort of a sense of sort of grimness around federal data right now, and the Federal Data Field Guide fills a little bit of a gap, it adds a little of whimsy to the conversation.

Terry Gerton You mentioned eight categories of data. Can you just hit those really quickly?

Chris Marcum Yeah, so we’ve organized federal data really, I think, for the first time in a publication into an ontology, and those categories include statistical data, and that’s data that measure population-level characteristics. There’s administrative data; geospatial data; scientific; accountability; evaluation; navigation data, that’s really data that helps citizens access government services; and then reference data, and so it’s for those of us who love our libraries, but it provides information on standardized systems across the government.

Terry Gerton Most people probably have no idea that there are that many different kinds of data.

Chris Marcum Yeah, that’s right. And what’s really cool about the Federal Data Field Guide is that we recognize that not all data is going to fit into one of those categories. And in fact, many blend across the categories and each of them, the categories has a little mascot, little animal mascot that Denice and our assistant, Google Gemini, helped us create. And the overall mascot for the Federal Data Field Guide is the platypus because it is an animal of many different other parts.

Terry Gerton Love that. Denice Ross is the director of the Data Policy Institute at the Federation of American Scientists and the former chief data scientist of the United States. Dr. Chris Marcum is senior advisor for federal data policy at the Federation of American Scientists. Denice, let me come back to you. Walk us through an example of how someone might use the Federal Data Field Guide.

Denice Ross I use these categories when I’m thinking about an issue that I’m working on. For example, June is Alzheimer’s Awareness Month, and so if you think about where do federal data sets support our understanding of Alzheimer’s and help support Americans who are grappling with this disease? And when you do that, you go beyond the usual suspects. And that’s what, I’d love to pass it over to Chris to talk about the most obvious data that we talk about when we think about Alzheimer’s, and that would be data that look at Alzheimer’s prevalence.

Chris Marcum Yeah, so as Denice said, what’s really nice about the Federal Data Field Guide is that we can think about data sets that are sort of timely and relevant, like in the context of brain health and Alzheimer’s Awareness Month, we might think of something like the CDC’s, of course, the National Vital Statistics System tells us, like, how many people die from Alzheimer’s disease every year. And that’s really the sort of blended data, one of these platypus data, because it’s statistical data. We have a statistical release. There’s also administrative data because it’s pulled in from information we get from states on death data. There’s also this really, really important CDC dataset on prevalence called the National Health Interview Survey. And in that, we learned that about 4% of the population over 65 has been diagnosed with dementia. And what’s very sort of consistent with that is that another data set that’s like really, I think it’s just really cool, it’s part of the Veteran [Affairs]’s Million Veteran Program. They’ve had basically the same finding in the veteran population, and some of that data are controlled access, they’re not really available to the public, but they’re available for researchers at the [VA]. And it’s really provided key insights into the exposure of that. Maybe Denice has more information on data sets that provide exposure risk for Alzheimer’s.

Denice Ross Yeah, thanks. And another type of data set that informs our understanding of Alzheimer’s and actually helps caregivers and patients is the navigation data that helps Americans navigate federally related resources. For example, the Centers for Medicare and Medicaid Services, Nursing Home Compare Database. I like to name data sets by name, give them credit. And it helps individuals assess which nursing homes may be the best care for their loved ones. And this also is a platypus because it falls into the accountability category, because it includes quality metrics on these federally supported facilities. And another data set that helps patients navigate federal resources is the clinicaltrials.gov data set at NIH. And that allows people to find what trials might be relevant for the stage of dementia that they are in.

Terry Gerton So I can imagine that different constituencies are looking for different answers. If you’re a patient or a caregiver, you’re going to type into your search, where do I find help? If you are a researcher, you’re going to be looking at, what are the prevalence issues and the conditions associated with it? How does a user of this data, who doesn’t even know that they’re a user of this date, find it?

Denice Ross I’m glad that you mentioned that, because in order for these data to serve their purpose, sometimes you don’t even need to find it, really. So, for example, an important reference data set is from the Social Security Administration. They have a data set on medical conditions that qualify for what they call compassionate allowances, and that enables someone to fast-track their disability benefits. So if you’re a 45-year-old and you just got diagnosed with early-onset Alzheimer’s, and you apply for disability benefits with the Social Security Administration, you don’t need to know that this reference dataset exists, but because the young onset Alzheimer’s disease is listed in this dataset, you can get services very quickly rather than having to wait the normal amount of time to get your benefits.

Terry Gerton So Chris, what happens in practice when someone tries to combine or use these different data sets from the field guide? Do they create their own platypus or do they have some other animal that’s out there?

Chris Marcum No, that’s great, because in information policy and data policy in particular, we talk about something called fitness for purpose. And fitness for purpose is a really, really important concept because it describes when the data were originally approved for collection, what purpose did they fill? What role is this going to play in Americans’ lives? And what services is the government agency that’s collecting the data going to benefit from the data? Now, fitness for a purpose makes a lot of sense on the original collection, but data have a life of their own. And sometimes, actually, I think often, we find that we can’t anticipate what downstream uses public data are going to have. And yeah, in some cases, it means that you might combine different data sets, you can imagine. And sometimes agencies do this themselves. So there’s a really impactful tool from the EPA called their Smart [Location] Database. And what they are able to do is link different data sets, like the American Community Survey from the census, to their own environmental data at EPA and give information about, I’m sorry, it’s called Smart Location Database. It’s part of their Smart [Growth] program, and they’re able to provide information on all kinds of environmentally informed location information about how people are located close to transit, for example, if they need to get a job.

Terry Gerton And so Denice, we’ve talked about Alzheimer’s data here today, and it’s really important, but it’s just a teaser for this series. As we go along, we’re going to come back to different pieces of the data ecosystem. Where are you going to plan to take us next?

Denice Ross Well, last October for Halloween, we had a piece on dearly departed data sets. And although it’s not Halloween, we are going to be releasing a comprehensive update of data losses. So look forward to that. Also, coming out of New Orleans and Hurricane Katrina, I always like to keep an eye on my hurricane-related data. So as we head into peak hurricane season, we’ll be covering federal data sets that relate to disaster response and recovery. And then, of course, talking about the future of data policy, we’re encountering some real new frontiers around the complexity of data and what our society needs from federal data. And that’s the future of what we’d like to talk about.