Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

The Jeff-alytics Podcast

The Jeff-alytics Podcast artwork

How The Federal Data Landscape Is Changing With Denice Ross

Aug 26, 202641 minAudio

Every crime rate, every per capita statistic, every trend line in American public safety depends on something most people have never thought about: the data infrastructure underneath it. You only notice it when it starts to break down. Today’s guest, Denice Ross, has seen this breakdown firsthand and it has informed her career ever since. Denice has served as the Chief Data Scientist of the United States under the Biden Administration, built the Police Data Initiative - an open data effort for police departments to voluntarily publish data - under the Obama Administration, and now runs the Data Policy Institute at the Federation of American Scientists. In this conversation, we get into what data has actually changed under the current administration. We also talk about the Census as the denominator for everything, declining survey response rates, and how Denice got a group of police chiefs to open up their records over cupcakes at the White House. Denice Ross is the Director of the new Data Policy Institute at the Federation of American Scientists, and the former Chief Data Scientist of the United States. In the Obama Administration, she co-led the Police Data Initiative which catalyzed law enforcement agencies to be more accountable by opening data about things like police use of force and citizen complaints. Denice worked in New Orleans (where we first met) for many years, before and after Katrina, first for a nonprofit data intermediary and then serving in the Landrieu administration, where she established the city's open data initiative. For more of Denice's work, check out http://dataindex.us/ and http://essentialdata.us/

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Transcript

From the show. Tap a time to play from there.

SPEAKER_01I'm Jeff Asher, and this is the Jeffalytics Podcast. Every crime rate, every per capita statistic, every trend line in American public safety depends on something most people never think about. The data infrastructure underneath it. You only notice it when it starts to break down. Today's guest, Denise Ross, saw this firsthand when Hurricane Katrina hit New Orleans, and it has informed her career ever since. Denise has served as the chief data scientist of the United States under the Biden administration, built the Police Data Initiative, an amazing open data effort for police departments to voluntarily publish data under the Obama administration, and now runs the Data Policy Institute at the Federation of American Scientists.

SPEAKER_01In this conversation, we get into what data has actually changed under the current administration. We also talk about the census as the denominator for everything, declining survey response rates, and how Denise got a group of police chiefs to open up their records over cupcakes at the White House. Let's get into it. My guest today is Denise Ross. Denise, thank you so much for joining me.

SPEAKER_00It's great to be here, Jeff.

SPEAKER_01So, Denise, walk us through uh what it what is your background? How did you get here to be today?

SPEAKER_00Well, as you know, I spent many years in New Orleans and uh was organizing data into neighborhoods and working with the nonprofit sector and local community-based organizations to help them use data to drive their own destinies. And um then Katrina happened, and uh all those data became instantly historical. And I realized that I had been taking federal data for granted and vowed that I would never do that again. Um, so in the in uh in the years since, I most recently served as the chief data scientist of the United States in the previous administration.

SPEAKER_00And now I spend my time keeping track of what's happening with the federal data ecosystem, looking for opportunities where those of us who depend on data can give input to the federal government about how important they are and how they could be improved, and telling the story about why federal data matter to everyday Americans.

SPEAKER_01What is my first question? Is hopefully the biggest softball. What is the US chief data scientist? For those that don't know, what does that person do? Why is it essential that the country have a chief data scientist?

SPEAKER_00The chief data scientist role was established in the Obama administration, right at the beginning of the open government, like open data movement. And it's really a policy role. So it's not like a chief data officer who's sort of overseeing the data. It's more how to align federal data with administration policies. And my role was like really, it was so satisfying because my role was to help federal agencies build the capacity to use data disaggregated by race and geography and gender or military connectedness, rural versus urban and disability, however, you need to slice and dice it to better serve the American people.

SPEAKER_00So make sure no one's left behind.

SPEAKER_01Why should we care about federal data? I mean, what what what is the overarching advantage to having this good data collection at the federal level?

SPEAKER_00Federal data touch every corner of American lives and livelihoods, but it's sort of invisible infrastructure that we take for granted. And, you know, the the concern, right, is that it we've been taking it for granted and it might be gone before we ever really fully realize its potential. And so the the general story that that I hear when I talk to people who depend on federal data is that they couldn't do their work without it, and that it's also not totally meeting their needs, and and that it's also being underutilized.

SPEAKER_00So it's really it's a national treasure that needs more attention. And it's gotten a lot of news in the last 18 months as federal capacity has diminished around data production. And uh, and so now's a good time to show our federal data some some love and attention.

SPEAKER_01And what so you're working now with the Federation of American Scientists. What is this work that you're doing? Why is it so important that we catalog how things are changing? And what is sort of the output that you're trying to put out?

SPEAKER_00We just launched the Data Policy Institute. And it's a it's it's really an area to focus on the present, where we've been monitoring changes to federal data that mostly manifests at a data set we call dataindex.us, where we have a newsletter to let people know about changes and opportunities where where the federal agencies are asking for public input. Uh, we have a six-point data checkup to uh to give data users a sense of what the risks facing the data sets they depend on might be.

SPEAKER_00And and then we also have we host these rapid response webinars when an opportunity for public input comes along. I think you participated in the early informal version of this back when the National Crime Victimization Survey, there were some, there was some word that it might be at risk of being terminated. And so you, you know, you blogged about it, and there was a lot of advocacy around that time, and the NCVS still stands, missing a few elements, but but we have been finding that public input does help protect the flow of federal data.

SPEAKER_00And then the other part of the current work that we're doing is telling the story about why federal data matter to everyday American lives and livelihoods, and that's at a website called essentialdata.us. And there we have over a hundred one-sentence love letters to specific federal data sets that don't talk about who uses the data, but actually how everyday Americans benefit from the data themselves. And then the future-facing part of the work is through the monitoring and the storytelling, we are seeing so many opportunities to improve the public data landscape so that it's more efficient and it's better at protecting data privacy and security, and that we're able to produce data that can meet the needs of a rapidly changing society.

SPEAKER_00So we that's more future-facing. And we think that with the attention that's on federal data now, it's a really good time to start thinking about how do we build a data policy environment that will serve the American people.

SPEAKER_01Can you sort of sum up? I know this is a huge question, but how has data changed under the Trump administration from a public-facing standpoint? You know, what is not good, but also like what is what is sort of maybe overhyped or is has not been changed necessarily as badly as people maybe first feared it might be.

SPEAKER_00When we look at the patterns of changes to federal data over the last 18 months, it falls mostly into three buckets. The first was the highest profile, and that was the big pull down of data that happened at the end of January of 2025. And that was in order for agencies to comply with some day one executive orders on DEI and gender and climate. And most of those data sets actually went back up. It was just that the career staff couldn't go through the data sets fast enough to figure out where there might be elements that that did not align with administration priorities.

SPEAKER_00And so some very large numbers came out of that January 31st takedown when we were getting ready to release the next version of our dearly departed data analysis. And I'll t I'll give you the scoop on this. It really still numbers just a little over a couple of dozen data sets that have been fully terminated in this administration. Um and so it's the overall losses aren't as bad as were initially reported. But there have been some explicit losses in DEI and climate and around gender identity elements in particular.

SPEAKER_00And then more recently, we've started to see data sets get terminated because they don't align, they they might reflect poorly on administration prior uh policies. So, for example, the highest profile version of this is the current population survey's food security supplement that the USDA publishes. It was terminated just weeks before millions of Americans were about ready to lose their food assistance benefits. And then the last bucket of types of data that have been diminished is just the collateral damage.

SPEAKER_00You know, there have been such huge reductions in staff, terminations of contracts, cancellations of federal scientific advisory committees that have been really essential for advising on data products and how to keep them modern and relevant and high quality. And then also there's just so much more red tape. So what used to be a routine process as part of the data lifecycle now might require a bottlenecked approval in the political offices. So of those three different buckets, the the one that really is going to be hardest to recover from is the collateral damage because it's just across the board.

SPEAKER_00And it's also these harms to the data often are happening happening quietly and in slow motion.

SPEAKER_01Do we have any sense? Again, this is not a quiz show, but do you have any sense like how many data sets exist? Like what is the what is the universe of federal data sets? Is there a number we can put behind it?

SPEAKER_00It depends on how you define a data set, of course, but hundreds of thousands, like you know, maybe half a million, is sort of the the number that I use. And I I did want to give you if if your game a few examples that are more related to criminal justice. Yeah. Okay. So one data set that was entirely terminated that that falls in that collateral damage bucket is the drug abuse warning network data set. And it's a health surveillance network that monitors drug-related visits to emergency rooms.

SPEAKER_00And what's what was great about that data set is it gave early warning to public health officials and local law enforcement officials about new new strains of illicit drugs that are coming into communities and sending people to emergency rooms. So you'd want that early warning and not be surprised by new types of substances in the community. And then you have a rationale for that? The the rationale, there was no rationale for that, actually. It just said that it was terminated. Um, at least like so often what you'll see on these on these web pages is there's a rectangular box at the top saying that it's just no longer, you know, going to be collected or published.

SPEAKER_00Sometimes it cites a specific executive order. In the case of the drug abuse warning network, it did not.

unknownOkay.

SPEAKER_00In many of these offices, all of the staff have been are no no longer in their places and the the offices had been reorganized. So often it's a people pop people problem.

SPEAKER_01That makes sense.

SPEAKER_00Yeah. There's also some other areas where elements had been removed from criminal justice-related data. And those those were the elements that did not align with administration priorities. The the one that I noticed first was the Bureau of Prisons, their inmate statistics data. They used to publish data on gender identity, and they don't do that anymore. Um, so those gender identity elements were removed. And those data are important because they allow the Bureau of Prisons to uphold its constitutional duty, right, to protect all federal inmates from violence.

SPEAKER_00And this, those data informed federal prison housing policies. And then another example, which I know you have followed closely, is the National Crime Victimization Survey. They lost three questions on gender identity. There were four total, and the one related to the hate crime specifically was retained. But that entire survey was at risk last year because of Doja's cost-cutting efforts. And it's it's worth noting that that data set is a collaboration actually between the Bureau of Justice Statistics and the Census Bureau.

SPEAKER_00So the Census Bureau's diminished capacity, because they've lost over a thousand staff themselves, puts the surveys that it collaborates with other agencies on at risk. And of course, the NCVS is important because like one good outcome of it is that it gives police chiefs a better, better visibility into the shortcomings of the data on reported crimes that they collect collect, especially on underreported crimes like sexual assault and domestic violence.

SPEAKER_00And it allows them to adjust their programming and uh protocols accordingly.

SPEAKER_01And you'd think for a survey designed at least partially to measure what doesn't get reported to police, you would want to explicitly measure crimes against populations that may be less likely to report them to police.

SPEAKER_00Totally. Yes. Um which brings up actually one of the biggest issues that we have with any data collection that comes through people, like forms or surveys. And that is survey response rates have been dropping, and that's a problem across the statistical system. But in but declining trust in government is just exacerbating that that challenge. So it's gonna have a pretty broad-ranging impact on some of the criminal justice related statistics that we depend on.

SPEAKER_01Is that a new trend or is that something that sort of social scientists have been measuring for a decade or more?

SPEAKER_00Aaron Powell It's it's it's a pre-existing trend that's been going on for several decades, but but it's gonna be, I think, particularly stark in areas where people already feel vulnerable, like the National Crime Victimization Survey, or even you know, around the decennial census.

SPEAKER_01Aaron Powell Why do you think, sort of changing gears just a little bit, like the crime data is sort of in the the officially reported crime data is at a in a golden age, essentially. Like we're getting monthly crime data from the FBI, states are publishing data, agencies have NIVERS uptake is through the roof, like unimaginably high compared to where it was five years ago. And yet there's all of these enormous data challenges elsewhere. The government shut shuts down and BLS doesn't report the labor data, but we're getting the crime monthly data now in a government shutdown.

SPEAKER_01Like, why are why is crime data doing so well when some other so many other places that are so important are not?

SPEAKER_00I'm I'm so glad you asked that because I often talk about crime data as sort of an exception to the greater trends. But you know, to your point though, like even the data that are being collected now are not haven't been sufficient to inform action on crime and improve the criminal justice system. Like they haven't been timely enough or detailed enough. And it does take a while for the standards to catch up to what the policy questions are. But I I you know I'm reflecting on when I served in the White House that September of 2024, was it, when you launched the crime index?

SPEAKER_00Um Yeah. So, you know, the the the FBI at that point was publishing annual, maybe quarterly crime crime data. Um, and that just wasn't fast enough to inform the national dialogue. And the minute you started harvesting the data from local jurisdictions and publishing it in the crime index, it like it changed the way everybody in the building consumed their data about crime. It was really wild. Like it was, it was totally, it was just transformative to have this peek into, you know, the the early monthly numbers.

SPEAKER_00And it was so complementary to the official numbers that the FBI was producing. And I think that is a real marker of resilience in a data ecosystem where you've got these official numbers that we can't live without. Um, but but there's always a trade-off between timeliness and detail and quality. And so, like, you know, the FBI should keep doing its thing, and then scrappy innovators like you can harvest the data at the source and start to make meaning of it so that we can maintain that that dialogue.

SPEAKER_00But you need both in order to have a healthy ecosystem. The other thing about this, the reporting of crime data is that it's coming from law enforcement agencies and the the technical infrastructure is very robust there. And the FBI set up standards and it's it's slow moving to add new elements to the standards because it takes time for all 18,000 agencies to catch up. But once they're on board, the data can start flowing. And and so I I and I the other thing I'll point out that this is sort of wonky, but it's worth noting that one of the reasons why the adoption of new standards is so slow is because it has to go through a formal advisory um committee process.

SPEAKER_00And I have you ever been to an advisory policy board meeting?

SPEAKER_01I can't say that I have.

SPEAKER_00Well, the Federal Advisory Committee Act basically says that if you're gonna be in, you know, getting input from the same group of people over and over again, that um you need to host these meetings in a formal public way. And that advisory policy board is just so incredible how it engages law enforcement at all levels of the United States and sort of brings them along to help advise on what the changes to standards should be. It's it's this really incredible, incredible democratic decision-making process.

SPEAKER_00And it's slow, but once the changes happen, you know, they're official and people are bought into them. So last year, uh non-fatal shooting data was added to NIBERS. It was a long time coming and fills a real gap in visibility on gun violence. But you know, that slow, slowly but surely those data got added to the official standards.

SPEAKER_01Yeah, and to your earlier point, I don't think that any agencies, I haven't seen any agencies reporting that via NIBERS yet, but it's like the type of thing that it gets into the standard and then all of a sudden it's it, yeah, you slowly get uptake and then much faster uptake figures. Right.

SPEAKER_00So the federal government will always be slower on these types of things, but you know, it's like a like um an aircraft carrier that, you know, but it has incredible capacity to to get things done and move things where they need to be, but it can't turn on a dime. And so innovators like you can do the turning on a dime that's necessary to squeeze value out of the data in a more timely way.

SPEAKER_01Aaron Powell Do you have any sense of what the future of I'm I'm certainly interested in criminal justice data, but also the whole federal ecosystem of data might look like.

SPEAKER_00That's a really good question. So one challenge has been that we don't have a complete mental model for what federal data are. Um we often think about it as just being statistical data or administrative data. Or if you're a scientist, you probably are just thinking about the scientific data. But there are data sets that implicate every domain that we might work in that are federal and shouldn't be taken for granted. So I would say the first step is sort of stepping back and widening the aperture so that we are taking a full accounting of the data that we have been taking for granted that the government produces.

SPEAKER_00That's the reason why my colleague Chris Markham and I, he's uh former OMB and Office of Science and Technology Policy, we created the Federal Data Field Guide. It's at federaldatafieldguide.us, and it's a really easy to consume field guide for the eight different types of federal data. And it's a useful framework for thinking about like if I really care about criminal justice, where are there types of federal data that that impact my work that I might not be keeping an eye on right now or that you might be underutilizing?

SPEAKER_01So you you very kindly called me an innovator, and I want to talk about one of your innovations because some of these I don't know the answer to, and I'm genuinely curious. But also I think it's like everything that I feel like I've built upon comes on the back of the police data initiative. And I talk about it all the time. Um, and it it sort of quasi-exists as a website, but it's it's like it lives in the spirit of the police departments that participated. Can you talk about like where what the genesis of that was and you know what what the idea is and what was actually built with this project?

SPEAKER_01Because I don't know how much people know about the police data initiative.

SPEAKER_00Yeah, thanks for asking. So you you know, I'd coming out of New Orleans and Post Katrina, New Orleans, one pattern that I saw when I was in the Land Rue administration was that we didn't have a common base of information to have discussions about what we needed to do to have a full and equitable recovery. And that manifested most visibly in arguments about blighted properties. And and there was such asymmetry of information.

SPEAKER_00The the you know, city government had access to information about code enforcement and legal proceedings and whatnot, that it was really hard for members of the public to access. And so we we got together with Code for America and they they came in and helped us build a tool that integrated all of the data that a property touches when it's in a code enforcement process. And so it gave like a single point of perspective into any given property.

SPEAKER_00And it was really transformative. And and I remember my first city council meeting after this tool went live, and someone It was it the topic was blight, so it was like standing room only. And this woman came up to the podium to talk for three minutes, and she had a printout from the website. And she was there to talk about, you know, her address at 123 Main Street. And the house next door was like falling on her house and full of vermin and crime and whatnot. And she gave the address, and then all the city council staffers started typing it into their computers.

SPEAKER_00And the city's blight czar, Jeff Abear, he brought it up on his iPad. And for the first time since the storm, we all had access to the same information. And the discussion was still hard about what we should do, but we were no longer arguing about the facts. And that really stuck with me about the power of a common base of information to inform a national dialogue, to inform a dialogue. And so after the police shooting of Michael Brown and Ferguson, Missouri, that's what came to mind that there were all these discussions happening nationwide about police violence and is it getting better?

SPEAKER_00Is it getting worse? Is it disproportionately affecting different populations? And we did not have the data to answer those basic questions. And so Clarence Wardell and I were both presidential innovation fellows. And we just we wrote one paragraph and we pitched it to the US chief technology officer and said, hey, what if we just brought together police chiefs and SAFA might release this data on use of force and citizen complaints in an open and transparent way? And that's where my naivety probably helped.

SPEAKER_00So, you know, so we actually we reached out to DOJ at that time and we're like, hey, we're thinking about bringing together some police chiefs and their data people to talk about opening up data on use of force. And they rightfully said, we've got a process, right? They have a beautiful, like well-architected process for changing standards for data collection. But we needed to move quicker than that. So, and and in retrospect, we did absolutely the right thing where we brought together police chiefs and their plus ones were their city data people.

SPEAKER_00Some in some cases they weren't really working together, but a trip to the White House sometimes brings people together. And and we we workshopped what the what were the important data sets that they wanted to publish in order to build trust and accountability with the public. And so, of course, you know, citizen complaints and use of force was on the list, but so were 911 calls for service, because that gives you the context of police interactions. So were data on assaults on officers, which also gives you a different type of context.

SPEAKER_00And then, and and you know, police force demographics, and so they came up with a dozen data sets that they thought would be the priority. And then we had these great cupcakes with the presidential seal on them. And we had, we closed with a session we called Cupcakes and Commitments. And so we had this big board, and police chiefs are are competitive and aspirational by nature. And and we had the data sets that they had raised as being important, and then we had the names of the the 13, I think, jurisdictions who were there that day.

SPEAKER_00And they each would put a post-it note on whichever data set they voluntarily committed to opening while they were eating their cupcakes.

unknownOf course.

SPEAKER_00And so you could see them in front of Liz chart, and and Art Acevedo from Austin put, you know, checked off every single data set that he was going to open. And uh so that's raised everybody's ambition. And um, and then and then we just started the the ball rolling. And as they started publishing these data, it reduced the risk for other law enforcement agencies to do the same. And the law enforcement leadership were finding that by proactively releasing data, they were building trust with their community.

SPEAKER_00They were making it more predictable for their forces about what transparency would transpire, you know, when there's use of force. And it just it just made it easier for them to operate. And then a few years later, the voluntary use of force collection started at the DOJ. But it was definitely informed by that local innovation in figuring out how to best produce these data in order to build trust with communities.

SPEAKER_01Well, in the I mean, I'll just say if the calls for service data I've built for New Orleans, having worked in the as a public safety consultant for the city council, although I'll say my council meetings, they were usually empty behind me. So, you know, we're we're talking about the crime data trends from the data person, they don't get quite the draw. But I mean that the calls for service data is essential. And we built dashboards that hooked up to it. We identified response times as enormous issues, identified crime, you know, skyrocketing and now plunging and all of these discussions about violent crime in New Orleans are because the city switched to NIBERS and isn't publishing crime statistics, we're able to track crime using just the calls for service.

SPEAKER_01And it's not perfect, but no crime data is perfect, so we accept that. And then, you know, my my latest piece on Substack or late a recent piece on Substack was all about measuring rape offenses in New York City and uh comparing to something that the mayor had said, and and that's entirely using public data that exists only because of the police data initiative. And it's like it it's part of their these agencies' DNA, even if probably agencies don't realize that. So quite the accomplishment, I'll say. Is there were there lessons that you learned from this that like speak to you today or what you what what you think about how we should be approaching public data today?

SPEAKER_00One lesson I learned, which I had figured out in the Landrew administration, but it became really clear as you know, we were bringing on over a hundred law enforcement agencies around the country. And that is that at the beginning of any administration, that is your magic window to start committing to more data transparency. Once you've been in place for a while, it becomes harder because the data are going to reflect the performance of your administration. As soon as you get in, like that's when the best time to lock in an open data policy.

SPEAKER_00And, you know, Art Acevedo, who committed to all the data sets, I really appreciated his leadership in like saying out loud that even when it looked when when Urban Institute and UCLA did an analysis of his use of force data and found that there was indeed bias in it, he owned it. And he said, uh, you know, as a police chief, I'd rather enter into this eyes wide open so that I can fix the disparities rather than sticking my head in the sand like an ostrich.

SPEAKER_00And that type of bold positioning, I think, is more important than ever as public trust in government is declining.

SPEAKER_01How do you get, how do you talk to policymakers and get them to accept like the the data showing there's a problem is not your enemy. It's your friend. It's telling you there's a problem that needs to be fixed. And you shouldn't come after the, you know, don't shoot the messenger, but acknowledge that this is an issue to fix. How do you get that, get that message through to policymakers?

SPEAKER_00I learned pretty quickly that the best way to do that is with stories and with examples. So if you can just get a few key leaders to go through this whole cycle where they are hesitant to release the data and then they do, and then the sky doesn't fall, and in fact, good things happen and it becomes this virtuous cycle. And telling the story about that makes it so much easier for other leaders to make the jump.

SPEAKER_01So if you were doing the police data initiative all over again, you know, you're in the you're in back in the White House, and all of these lessons that you've learned, are there things that you would do differently? What would you what would you change? What would you do the same?

SPEAKER_00Aaron Powell So what we would be what we've been doing now is looking at how our federal data capacity is sort of crumbling. And part of the problem that's causing that is that it's a patchwork quilt of policies that they govern different types of data and data collections. And so it's what I would do this time around is really take a more entrepreneurial view of like, you know, what are the data policies that need to change so that we can have, for example, less burden on law enforcement agencies to report data to different federal departments for different programs.

SPEAKER_00Like there's we we've got to reduce the friction for that reporting, especially if many of these collections are voluntary. And so so looking at looking at the policy landscape and also there's been some changes in laws. For example, the Evidence Act was a real game changer. The Open Date Government Data Act that was signed alongside that by President Trump in his first term requires that agencies create an inventory of all the data they have and that they actively engage with public stakeholders, including state and local officials, about how the data are used and how to improve the utility of the data.

SPEAKER_00And that type of like really robust engagement, I think is crucial for creating high-quality data. One thing that sticks in my mind was a a police chief describing how he had the data he reported to the federal government, and then he had over here the data he actually used. And that was double work for him, and that's not right. Like, let's figure out how can we make that collection also extremely useful for the jurisdictions that have to do the footwork.

SPEAKER_01So the the podcast, again, the whole idea is is talking about data, but also like how do we communicate data? And so from all of your experience, this is the the question where I ask guests to do my job for me. How would you improve communication of of crime trends to a public that doesn't usually or always follow it?

SPEAKER_00Those of us who work in data, we know these data sets like they're old friends, right? We use nicknames for them. Yeah, you know, we know their strengths and their weaknesses. We forgive them when they're, you know, a few weeks late, like and and we have a very different relationship with data than the general public does. But the general public it like directly benefits from these data. And it's technically their data because they they paid for it. And a lot of times it actually comes from activities that they do, like filling out the census.

SPEAKER_00And what we've been trying to do with the central data and in particular getting on social media and talking about federal data is America understand that this is like invisible infrastructure that we take for granted that needs to be nurtured and funded. It's like the internet or cell service or bridges, just, you know, we need data. Data is the nervous system of government at all levels.

SPEAKER_00And without it, we're not going to be able to take the actions or make the decisions that we need to. So just fundamentally sort of helping, helping to tell the story about why these data matter to everyday Americans, I think is a nice complement to then talking about what the data are showing and and connecting that to their experience in the real world.

SPEAKER_01And my last question is how does AI factor into any of your work? And how do you think it could potentially factor into better policymaking into the future?

SPEAKER_00AI factors into everything. In the case of federal data in particular, making federal data AI ready is has been a priority for now two administrations at least. And one of the most important things about data that will be useful for AI is that it's detailed and accurate and unbiased, as unbiased as it can be. And for me, that is an opportunity to invest in our core, our basic data that we need to run a modern society, so that any quality issues with data don't magnify as we go to scale with AI solutions.

SPEAKER_00And, you know, I think I'm particularly concerned right now about the nation's denominator, the census. It's it's certainly a data set that's that is you used broadly in analyses and AI applications because it's the the fabric of you know who who lives in America. And it's the denominator for all of our crime rates and like basically every is the denominator for everything. And I don't know if you remember after Katrina how the census estimates weren't able to keep up with population changes.

SPEAKER_01The FBI has the craziest population number for 2006, I think. It's like the same, and then it plunges in 2007.

SPEAKER_00It's right, exactly. And the and the downstream consequences of that were huge because if you have too high of a number that hasn't been updated, then it looks like you have no crime at all because there's a smaller population actually committing crimes. And then if you can't keep up with the repopulation, it looks like you've got sky high rates, and that might actually impede people feeling safe to come back to the city. And so the nonprofit I was with at the time, we found that data on the from the junk mail company, uh that yeah, that send you junk mail, that that was the best proxy for repopulation.

SPEAKER_00And so we were able, they donated it to us. So we were able to publish uh percent households repopulated by block every month. And there was a point, I think it was like 2008, 2009, where the police chief was using the junk mail data as his denominator.

SPEAKER_01I I have I have used the data centers addresses, getting data to calculate a consistent rate many times. So yes, it's a very useful data set still.

SPEAKER_00Yeah, absolutely. But it just, you know, that for me, like that denominator is so important. It's so important for AI, and it's so important for actually understanding the context of crime in America. And there's multiple threats. Uh, the census is always, you know, sort of at risk for different reasons, but the declining response rates that I mentioned earlier, coinciding with a declining trust in government. There's active talk now about adding a citizenship question to the census, which would have a chilling effect on response rates and probably a significant undercount for especially mixed population, mixed immigrant status households.

SPEAKER_00And the Census Department of Commerce recently issued an English-only order for all official business. So in the last decennial, something like 3 million people used the non had non-English responses. And so that could definitely cause response rates to fall. And we're also rolling back some modernizations on how we ask questions about race and ethnicity. And that was real progress because Middle East and North African folks have been statistically invisible.

SPEAKER_00And the two-part question about Hispanic was generating low quality data. Um, so that's concerning. We will, you we could end up with lower quality data about race. And then what I think especially impacts crime and AI and the need for detailed data in order to train models is this new Department of Commerce policy that's that prohibits the standard privacy protection methodology of noise infusion. And I know this sounds wonky, but the implications, there's two possible ways this could go.

SPEAKER_00One is that either small areas or small populations, small areas like neighborhoods or small populations, that those data might need to be suppressed. So you wouldn't be able to see them at all. And the other option is that they just publish the data anyways, and privacy are breached, which would result in even more distrust and really sort of a death spiral for participation in in forms and surveys like the census.

SPEAKER_01So, Denise, what what is next for you?

SPEAKER_00I'm having so much fun telling the story about why federal data matter for the American people. And what I tell everyone is most of my friends are data users. Most of the people I interact with are data geeks. But if you use data or you benefit from federal data, now is now's the time to talk about why it matters. It's just it's a part of being part of this democracy, participating in decisions about our data policies and how we keep them keep data safe and how we make data more useful to serve the American people.

SPEAKER_01Excellent. And where can people find you? Can you r remind us one more time?

SPEAKER_00Easy to find. Denise with a C, W Ross, on Instagram and LinkedIn. And I'm doing some fun videos about data where I oh Jeff, you'll appreciate this. I I wear a costume. So that's my thing. I will not have any costume ideas. I try to like match the costume with the data set. And so I'm bringing a little bit of New Orleans to the DC area. And so, you know, so we're posting videos there and other campaigns about specific data sets. And so if you go to essentialdata.us, that's where we're telling the stories.

SPEAKER_00If you want to host a like a sh workshop around a specific type of federal data that you particularly care for and want to make sure that we're telling the story about why it matters, we have this really fun, easy way where we can bring five to 20 people together on a Zoom call and just whip out a bunch of use cases about why the why the data matter. Um, and then on dataindex.us, if you sign up for our newsletter, you'll be first to hear when there are proposed changes to surveys like the National Crime Victimization Survey.

SPEAKER_01Awesome. Denise, thank you so much for joining me. I've learned a lot. I know everybody else will too. So I really appreciate it and keep up the great work.

SPEAKER_00Thanks, Jeff.

SPEAKER_01Thanks for listening to the Jeffalytics Podcast. Be sure to subscribe and to learn more, head on over to ahdatalytics.com for more information and previous episodes. If you like what you heard, please leave a glowing review, which will help others to discover the show. Until next time, I'm Jeff Asher.

Open the transcript file

Hosts and Guests

Nobody is credited on this episode yet.

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. Why Fewer Murders Get Solved These Days With Phil Cook

    Nov 18, 202634 minAudio

    Police departments said they solved 92 percent of homicides nationwide in the early 1960s. Today that figure hovers around 60 percent most years. Phil Cook has spent the last half century figuring out what that enormous...

    Credit the people on this episode

  2. How Gun Violence Makes America Sick with Dan Semenza

    Nov 11, 202638 minAudio

    Dan Semenza is a criminologist and director of research at the New Jersey Gun Violence Research Center at Rutgers University’s School of Public Health - a position that puts him deliberately at the intersection of two...

    Credit the people on this episode

  3. How States Can Address Gun Violence with Lt. Governor Austin Davis

    Nov 4, 202623 minAudio

    Gun violence is not a one-size-fits-all concern. It can be a different issue depending on where you live and whether it is an urban or rural community. And that makes the problem genuinely hard to address at scale....

    Credit the people on this episode

  4. Policing the Big Easy with New Orleans Police Chief Anne Kirkpatrick

    Oct 28, 202634 minAudio

    It takes a certain kind of leader to see a city deemed the country's "murder capital" and actually want to take on that problem. Chief Ann Kirkpatrick has spent 44 years working in some of the most challenging...

    Credit the people on this episode

  5. How Cities Succeed and Fail in Public Safety Policy with Gloria Gong

    Oct 21, 202633 minAudio

    One of the hardest questions facing any mayor - and a question that most mayor’s struggle to answer fully - is how you actually know if what you’re doing is working. Gloria Gong co-founded the Government Performance Lab...

    Credit the people on this episode