Immigration Data Literacy Skills to Help You Survive a New Wave of ICE Confusion

In this issue: ICE's latest detention data contains errors, growth of non-criminal detainees continues, and DHS Secretary Kristi Noem's inadequate claims about ICE arrests.

In the past 24 hours, the Department of Homeland Security has made several blunders related to immigration data. From ICE releasing faulty detention data to DHS Director Kristi Noem issuing a perplexing claim about ICE arrests, the urgent need for immigration data literacy remains high.

Today's post addresses this serious topic, but it is not without its joys of discovery and opportunities for education along the way. In doing so, what I hope to do with this newsletter is not just give you solutions to puzzles, but help you become better puzzle-solvers. And immigration data—like the immigration system itself—is nothing if not a puzzle!

Today's post will show you why ICE's most recent detention data is wrong, justify my inference into what I think the real numbers are, address ICE's inadequate press release that includes a confusing statement from Kristi Noem, and confirm my previous hypothesis for why ICE stopped posting daily arrest data.

Identifying & Understanding ICE's Detention Data Error

I want to walk through the issue with ICE's detention data. It's a simple mistake, not a sophisticated one, but it is worth discussing for two reasons. First, the mistake is sloppy, but not obvious. A casual observer who is not familiar with ICE's detention data might easily overlook ICE's mistake and wrongly report current detention numbers. Second, once you see the mistake you can't unsee it, which might prompt you to wonder how ICE's own staff (whose job it is to work with this data) didn't catch the mistake themselves.

A quick refresher on ICE’s detention data. Congress requires ICE to regularly produce a report on immigrants held in detention. This data is published as an Excel spreadsheet and updated on ICE's Detention Management website roughly every two weeks.

ICE's detention data is the most up-to-date publicly-available data on the outcomes of ICE's purported enforcement surge under President Trump. And unlike the scattering of unverified (and unverifiable) statements made by DHS officials online and to reporters, this data set has been around for awhile which contributes to its credibility and objectivity (even if it is far from perfect).

The spreadsheet contains a number of tabs and tables. For our purposes, we are only going to look at the table that shows total immigrants in detention broken down by criminal category (i.e., conviction, charge, or immigration-only violations) and the agency that initially encountered the individual (i.e., CBP or ICE).

The distinction between CBP and ICE is relevant because the enforcement contexts and immigration populations encountered by each agency shows some divergence, even if they both end up in ICE's custody.

Immigrants arrested by CBP tend to be encountered closer to the border and they tend to have been in the United States for much shorter periods of time. Therefore, fewer of them have lived in the country long enough to accumulate criminal charges for the same reason that the percent of the population that has a traffic ticket at 30 years old is higher than at 20 years old. By contrast, because ICE tends to operate throughout the interior of United States (i.e., "interior enforcement"), immigrants arrested and detained by ICE are more likely to have been in the country much longer and have higher rates of criminal history.

Furthermore, ICE enforcement practices (and therefore their total numbers) tend to be more influenced by the political priorities of the current administration, since ICE arrests require the additional step going out into the community or to a jail to take someone into custody. While policy certainly shapes CBP's numbers, the nature of border policing is more reactive. CBP's total arrests depends in part on the total population attempting to cross the border, which is not something that CBP can unilaterally control.

As a result of these dynamics, if ICE's total detained population increases and if CBP's arrests are driving this increase, then the percent of detained immigrants with criminal convictions is likely to decrease. Here's an example of these dynamics. During the pandemic, when Title 42 prevented migrants from crossing the border and lower priority detainees were released from detention or not booked in in the first place, the total detained population was at 20-year lows, but the percent of detained immigrants with criminal convictions was at 20-year highs. Remember that immigrant detention is civil, not criminal, so people are not required to have criminal histories to be detained in these jails and jail-like facilities.

Let's apply that understanding to ICE's detention data. The table below is from ICE's detention data published two weeks ago (around February 14, 2025). The data confirms the explanation above that most of the immigrants with criminal histories were initially arrested by ICE while the majority of immigrants with only immigration violations were arrested by CBP. This table (and a collection of these historical tables that I save over time) is what enabled me to produce the data I published in my previous post.

Now let's look at the table from the detention spreadsheet that ICE posted yesterday on February 28, 2025. If you only look at the total number of detainees recorded and the totals from each of the two arresting agencies, it might look normal. Numbers went up across the board. But if you look more closely, you'll see that the breakdown among the three categories of criminal history are off in a strange way. ICE's breakdown looks more like CBP's and vice versa. See if you can spot it yourself.

This is an obvious mistake if you know what you're looking for but you have to know what you're looking for (which not everyone does) and you have to have copies of ICE's older detention tables (which not everyone has).

This isn't the first time that ICE has released bad data to the public, but in the past the mistakes tended to show up on another tab in their spreadsheet for "Alternatives to Detention," which relies in part on data provided to the agency by an outside contractor. This is the first time I remember seeing such a glaring error in ICE's own detention data, which makes me worried that this is just the beginning of more errors.

How and why is ICE's staff allowing sloppy mistakes like this to make it out the door? I do not believe this mistake is malicious or intentional; I believe it is primarily driven by institutional indifference, but that's not an excuse. We should remember that indifference, especially indifference to the truth, is a political choice--not a mere accident that we can simply ignore.

The final question for ICE is, how will they handle this error? Will they issue a correction or let this misrepresentation persist until the next time they release data, which may contain yet new and unforeseen data errors that may be of more consequence and harder to identify? How the agency responds to this mistake will say as much about their commitment to transparency and accuracy as the mistake itself.

Now that we understand ICE's error, what can we do? Fortunately, based on what we know, I believe a reasonable inference can be made that the data in the two columns were simply transposed. I will proceed in the next section as if this was the case and treat the following table as the true numbers.

As you can see, data validation is really important! If you want to learn more about the importance of data validation to good reporting, watch or listen to my panel discussion "Covering immigration stories in 2025: Expert insights and reporting advice for Harvard's The Journalist's Resource with Linda Dakin-Grimm, Caitlin Dickerson, and Carmen Nobel.

Immigrants with Only Immigration Violations Continues to See Largest Growth

ICE's updated detention data shows a continuation in the trend that I've noted over the past month: as the total number of detainees arrested by ICE continues to increase, the group that is increasing the most are immigrants without criminal histories. Hopefully the discussion in the previous section illustrates why I am only focusing on ICE.

The graph and table below shows the change in the total number of immigrants in detention by criminal history from the end of the Biden administration to the most recent data released under the Trump administration. In terms of overall numbers, the majority of immigrants in detention have some criminal history, although this ranges from fairly minor to serious. At the same time, the biggest growth, and the growing share of immigrants with only immigration violations is seeing the largest growth.

The updated graph shows that since the start of the Biden administration, immigrants without criminal histories have seen an increase of 4.3x while immigrants with criminal convictions have seen the least relative growth at just 1.3x.

Based on previous data that I have examined over the past several years, I fully expect that the number of immigrants with no criminal histories will continue to increase and overtake the total number of immigrants with criminal convictions or charges as ICE's detention capacity expands. Despite what the administration says, the truth is that total population of immigrants with criminal convictions that end up in detention is far less elastic than the population of immigrants with no criminal histories.

A caveat for those of you who want to level up your critical thinking skills around detention data. These data are what we call "snapshot" data, so they only show total population on a given day. When looking at these data historically, one should keep in mind that different categories of detained populations move at different speeds through the system. Someone with no criminal history may be booked into detention, then released again fairly quickly, while someone with a serious criminal conviction might be in detention for a longer time. Therefore, we don't really know the total number of people that have passed through the system. Imagine if you took a photograph of a 5K race. If you took two pictures 60 seconds apart, you might be able to observe an increase or decrease in the total runners at one point in time, but you wouldn't be able to calculate how many total people are running in the race because people run at different speeds. That's the idea.

This simple factual observation undermines the administration's attempts to represent all immigrants as dangerous criminals, a claim that they have yet to substantiate even in the case of Guantánamo Bay detainees. Earlier this week, the New York Times reported that despite claims that everyone going to Guantánamo Bay were criminals associated with the Venezuelan Tren de Agua gang, State Department officials told Congressional staffers that the only criteria for sending people to the controversial island detention center was that individuals had final orders of removal.

If you want more context for immigration data and policy, listen to my discussion below with Emma Vigeland for the Majority Report.

Kristi Noem's ICE Press Release Claiming 627% Increase in ICE Arrests

A colleague shared the following press release issued by ICE with a quote from DHS Secretary Kristi Noem in which the agency claims that ICE arrests have increased 667% in the first month compared to the Biden administration. This misleading post is yet another great opportunity to think critically about how the Trump administration is using what is sometimes called “number laundering” or “math-washing” to misrepresent the totality of what the agency is actually doing. I may write a longer post on this topic, but for now I want to share a few conclusions after looking at the data on ICE arrests over the past few years.

The press release claims that ICE arrested 20,000 immigrants in the past month. That’s a lot, no question. Let’s take a look at what this short and unsubstantiated press release doesn’t say.

It is true that the number of what are known as “at large arrests” (arrests that take place out in the community) was down last year to 33,000, which is lower than it has been in a while. ICE reassigned a lot of staff to border enforcement, which did mean that the agency with its limited resources had to limit its efforts elsewhere. As many of us have been saying for a while, ICE does not have unlimited resources so it can’t do everything everywhere all at once—no matter who is president.

As a result, last year was not a particularly representative year. In fact, at large arrests were much higher in FY 2023 at over 90,000 (which was higher than any year of the Trump administration) and at large arrests were much higher during the Obama administration at around 300,000 for one of the highest years on record. ICE is selectively comparing only an out-of-the-ordinary subset of data to a selective time period of the past 30 days of supposedly super-aggressive enforcement.

The agency completely ignores two other kinds of arrests: “custodial arrests” (where ICE takes someone into custody from another law enforcement agency) and criminal arrests from Homeland Security Investigations (HSI)—both of which are major sources of arrests for ICE.

The reason this matters is, since the agency does not have unlimited resources, the agency is pulling efforts from custodial arrests and HSI criminal arrests to help with these “at large” arrests in the community. Therefore, even if at large arrests go up, it may not actually matter that much in the scheme of things if it contributes to a decline in other types of arrests. And that’s even assuming that the agency thinks it can keep this pace, which I don’t think it can.

Suppose that the 20,000 number is correct, let’s put this in perspective. We know based on the detention data that the agency is not keeping everyone they arrest in detention. Most people are being released one way or another. But let’s even say that this surge of enforcement is sustainable and let’s say that they deport every single person they arrest. They still wouldn’t hit a million people in the full four years of the Trump administration—and the administration seems to want to deport all 12 million or more people overnight. Don’t get me wrong, 20,000 people a month is a lot. I just want to highlight the systemic constraints to mass deportation.

A final point here. I previously wrote about ICE’s public data memes they were posting on Twitter, which they started and suddenly stopped in the first week of the administration. I conjectured that they stopped posting those memes because their initial high numbers were dropping off. Turns out that was correct. Again, if we assume that ICE is accurate about making 20,000 arrests in the first month, that works out to 667 arrests per day which is less than their initial posts were showing. That means, yes, the average daily arrests for the entire month is considerably lower than their initial reports of over 1,000 on busy days. My guess is, they were worried about their MAGA supporters online trolling the agency for not keeping up the pace that Trump promised. Here’s that previous graphic again.

ICE Corrects Immigration Data (Revised 3/5/2025)

Around noon on March 5, ICE re-released their detention spreadsheet with corrections to their table of immigrants who were detained in ICE custody as of February 23, 2025. My analysis was correct: the agency simply transposed the two fields of data for detained persons arrested by ICE and CBP. I appreciate that the agency corrected its mistake. The corrected table below is now consistent with the table I used in the analysis above, which means that, thankfully, I don’t have to repeat the analysis.

Consider Supporting Public Scholarship

Thank you for reading. Please subscribe to receive this newsletter in your inbox and share it online or with friends and colleagues. If you believe in this work, consider supporting it through a paid subscription. Learn more about the mission and the person behind this newsletter.

Photo by Jeff Golenski on Unsplash