Illegal Immigrants Are Far More Likely to Commit Serious Crimes Than the U.S. Public

Agresti, J. D. (2018, June 29). Illegal Immigrants Are Far More Likely to Commit Serious Crimes Than the U.S. Public. Retrieved from
Agresti, James D. “Illegal Immigrants Are Far More Likely to Commit Serious Crimes Than the U.S. Public.” Just Facts. 29 June 2018. Web. 19 June 2019.<>.
Chicago (for footnotes)
James D. Agresti, “Illegal Immigrants Are Far More Likely to Commit Serious Crimes Than the U.S. Public.” Just Facts. June 29, 2018.
Chicago (for bibliographies)
Agresti, James D. “Illegal Immigrants Are Far More Likely to Commit Serious Crimes Than the U.S. Public.” Just Facts. June 29, 2018.

By James D. Agresti
June 29, 2018

President Trump recently held a conference with family members of U.S. citizens killed by illegal immigrants. The parents of nine people slain by such immigrants spoke about their family’s experiences, and Trump presented an array of government data on criminal immigrants and stated:

I always hear that, “Oh, no, the population is safer than the people that live in the country.” You’ve heard that, fellas. Right? You’ve heard that. I hear it so much. And I say, “Is that possible?” The answer is it’s not true.

In response, the Associated Press published a “fact check“ claiming that illegal immigrants are more law-abiding than the general public. Various media outlets, such as the New York Times, Yahoo!, and a number of NBC affiliates published this article. The Washington Post ran a similar story, and other media outlets and so-called fact checkers have made similar claims in the past.

The truth, however, is that comprehensive, straightforward facts from primary sources—namely the Obama administration Census Bureau and Department of Justice—prove that illegal immigrants are far more likely to commit serious crimes than the U.S. population. Studies that claim otherwise typically suffer from fallacies condemned by academic publications about how to accurately analyze data.

The Most Concrete Facts

Data on illegal immigration and crime is often clouded, precisely because these are unlawful activities where perpetrators seek to hide their actions. Also, governments sometimes fail to record or release information that could be or has been obtained. The Obama administration, in particular, refused to release the names of convicted immigrant sex offenders and hid other details about crimes committed by immigrants.

Nonetheless, a combination of three material facts sheds enough light on this issue to draw some firm conclusions.

First, U.S. Census data from 2011 to 2015 shows that noncitizens are 7% more likely than the U.S. population to be incarcerated in adult correctional facilities. This alone debunks the common media narrative, but it only scratches the surface of serious criminality by illegal immigrants.

Second, Department of Justice data reveals that in the decade ending in 2015, the U.S. deported at least 1.5 million noncitizens who were convicted of committing crimes in the U.S. (Table 41). This amounts to 10 times the number of noncitizens in U.S. adult correctional facilities during 2015.

Third, Department of Justice data shows that convicts released from prison have an average of 3.9 prior convictions, not including convictions that led to their imprisonment (Table 5). This means that people in prison are often repeat offenders—but as shown by the previous fact, masses of convicted criminals have been deported, making it hard for them to reoffend and end up in a U.S. prison.

In other words, even after deporting 10 times more noncitizens convicted of crimes than are in U.S. prisons and jails, they are still 7% more likely to be incarcerated than the general public. This indicates a level of criminality that is multiplicatively higher than the U.S. population.

Furthermore, roughly half of noncitizens are in the U.S. legally, and legal immigrants rarely commit crimes. This is because U.S. immigration laws are designed to keep criminals out. Thus, the vast majority of incarcerated noncitizens are doubtlessly illegal immigrants. If legal immigrants were removed from the equation, the incarceration rate of illegal immigrants would probably be about twice as high as for all noncitizens.

On the other hand, there is uncertainty about the exact number of noncitizens in the U.S., and Census figures are almost surely low. Hence, the incarceration rate of illegal immigrants is likely not twice as high as the U.S. population. Nevertheless, this is only the tip of the iceberg, because the U.S. continually deports massive numbers of illegal immigrant convicts.

Data Misuse

According to the AP, one of the supposed reasons why Trump is wrong is that “Ruben Rumbaut, a University of California, Irvine sociology professor, co-authored a recent study that noted crime rates fell sharply from 1990 to 2015 at a time when illegal immigration spiked.” This study is a quintessential example of data misuse, because it confuses association with causation and cherry picks a timeframe that creates a misleading impression.

Per the study:

Between 1990 and 2013, the foreign-born share of the U.S. population grew from 7.9 percent to 13.1 percent and the number of unauthorized immigrants more than tripled from 3.5 million to 11.2 million.

During the same period, FBI data indicate that the violent crime rate declined 48 percent—which included falling rates of aggravated assault, robbery, rape, and murder.

Such data reveals nothing about the effects of illegal immigrants on crime, because as explained in a textbook about analyzing data:

Association is not the same as causation. This issue is a persistent problem in empirical analysis in the social sciences. Often the investigator will plot two variables and use the tight relationship obtained to draw absolutely ridiculous or completely erroneous conclusions. Because we so often confuse association and causation, it is extremely easy to be convinced that a tight relationship between two variables means that one is causing the other. This is simply not true.

Numerous other academic writings say the same, and this basic fact is taught in high school statistics. For example, the Common Core math standards require students to “distinguish between correlation and causation.” Yet, Rumbaut and two other PhDs authored this flawed study, and the AP and PolitiFact uncritically quoted it.

The Rumbaut study has another major problem, because it cherry picks a timeframe that diverges from the larger picture. Contrary to its narrative that growing numbers of immigrants during 1990–2013 caused crime to fall, the number of immigrants also grew during the 1970s and 1980s, but during this time, the homicide rate rose, fell, and rose again:

In the words of another academic book about data analysis:

One of the worst abuses of analytics is to cherry pick results. Cherry pickers tout analysis findings when the results serve the purpose at hand. But, they ignore the findings when the results conflict with the original plan.

The AP also cited another study that suffers from much the same shortcomings. Published in the journal Criminology, it examines state-level data from 1990–2014 and finds that “increases in the undocumented immigrant population within states are associated with significant decreases in the prevalence of violence.” Besides using a timeframe when crime fell sharply, these statistics are merely associations, and the authors admit “they are hardly conclusive.”

Thus, they take their study further by using a statistical technique called a “regression” to control for factors besides immigration that could affect crime rates. Based on this, they find that “undocumented immigration over this period is generally associated with decreasing violent crime.”

Again, this is just an association, and it does not show that illegal immigration reduced crime, because other factors are undoubtedly at play. As detailed in a book about regressions, they share “an additional problem with all methods of statistical control,” because “there’s no way that we can measure all the variables that might conceivably affect” an outcome.

In this case, the outcome is general crime levels, and their causes are notoriously difficult to identify. In the words of an Oxford University Press textbook on criminology, this is “because there are simply too many unknowns and unmeasured dark figures of crime and explanation to enable us to draw valid and reliable conclusions from research.”

Moreover, illegal immigrants comprise only 4% of the total U.S. population in this study. Yet, the study seeks to determine their criminality by measuring total crime rates for the entire populations of the states. That is beyond far-fetched, because even small changes in crime among the other 96% of the population could easily overwhelm any effects from illegal immigrants.

In sum, these studies don’t prove anything about the criminality of unauthorized immigrants. As obvious as it sounds, this needs to be said: To measure the criminality of illegal immigrants, one must actually measure their criminality or a valid proxy for it. These studies do not do that.


The AP also cited a study from the “libertarian think tank Cato Institute“ as evidence that “people here illegally are less likely to commit crime than U.S. citizens,” but the study is a classic example of one that misleads by ignoring relevant facts.

The study, conducted by Cato policy analyst Alex Nowrasteh, examines arrests and convictions of illegal immigrants in the state of Texas. Unlike sanctuary jurisdictions that refuse to record and/or publish such information, Texas collects and makes this data publicly available. By using this data and immigrant population estimates, the study finds that in 2015, the conviction rates of illegal immigrants in Texas for:

  • homicide were 25% below that of native-born Americans.
  • sexual assault and larceny were 11.5% below that of U.S. natives.
  • all criminal acts were 56% below that of U.S. natives.

In isolation, these statistics vastly understate the criminality of illegal immigrants—first and foremost—because they fail to account for the fact that hordes of criminal immigrants are constantly deported. This is especially relevant when it comes to crimes like murder, in which the vast bulk of perpetrators have previous criminal records. For example:

  • in New York City from 2003 to 2005, more than 90% of the known killers were people with criminal records.
  • in Baltimore during 2015, 77% of murder suspects had prior criminal records, and the average suspect had been previously arrested more than nine times.

These facts have even greater importance in a state like Texas, which readily cooperates with the Department of Homeland Security to deport criminal immigrants. As a participant in the Department’s Priority Enforcement Program, Texas investigates the immigration status of all arrestees and helps immigration authorities “take custody of individuals who pose a danger to public safety before those individuals are released into our communities.”

In Texas, there’s almost no chance that illegal immigrants could be arrested nine times and still be in the state and on the loose. In all likelihood, they would either be deported or locked up long before then. This would explain why the murder rate for illegal immigrants who remain in Texas is lower than that of U.S. natives.

This also may be why Cato finds relatively low overall conviction rates for illegal immigrants in Texas—while nationwide—the Congressional Research Service reports that the incarceration rate for noncitizens has “corresponded closely to that of noncitizens in the U.S. population….” Again, since roughly half of noncitizens are in the U.S. legally, and legal immigrants tend to be very law-abiding, the incarceration rate of illegal immigrants is probably about twice as high as for all noncitizens.

Also, unauthorized immigrants are more apt to literally get away with murder than the general population. This is because murders committed by minorities are less likely to be solved. In the sanctuary city of Chicago, for example, the portion of murders that resulted in a suspect being identified and acted upon by the criminal justice system was 19% in 2016, as compared to a nationwide average of 59%.

On top of this, information from the Social Security Administration and other sources shows that most illegal immigrants engage in identity fraud. This is a federal felony punishable by up to five years in prison.

All of this proves that illegal immigrants have much greater levels of serious criminality than their arrest and conviction rates indicate. Those metrics only tell part of the story, particularly in a state like Texas that consistently works with the Department of Homeland Security to deport criminal immigrants.

Other Farces

Beyond the fallacies in this AP “fact check,” various journalists and scholars have misled the public about this issue by using bait-and-switch tactics and statistical techniques that are inappropriate to the data. Per another book about data analysis, “Statistical analysis is very easy to misuse and misinterpret. Any method of analysis used, whenever applied to data, will provide a result, and all statistical results look authoritative.”


When it comes to illegal immigration and crime, media outlets often lavish attention on fatally flawed studies while ignoring straightforward, comprehensive, and rigorously documented facts about this issue. Such facts show that Trump is correct on this point. Illegal immigrants are much more likely to commit serious crimes than the general public.

However, the opposite is true when it comes to legal immigrants, especially those who become U.S. citizens. They are 79% less likely than the general public to be incarcerated in adult correctional facilities. In this case, deportations are not a factor, because immigrants who become citizens generally cannot be deported unless they are stripped of their citizenship, which is a rare occurrence.

There’s good reason why immigrant citizens are so law-abiding. To become U.S. citizens, they must pass a full FBI background check, demonstrate they have good moral character, show they will not be a financial burden on taxpayers, and take a public oath of allegiance to the U.S. Constitution. Illegal immigrants avoid such vetting, and this allows masses of criminals to enter the U.S.

15 thoughts on “Illegal Immigrants Are Far More Likely to Commit Serious Crimes Than the U.S. Public

  1. The left blends the propensity for crime of very law-abiding LEGAL immigrants with the far less law abiding ILLEGAL immigrants to make the case that “immigrants” commit less crime. In any case, illegal aliens are killing Americans. Here are 115 dead Americans killed by an illegal alien. Unfortunately, there are many, many more.

  2. This article is way too complicated and obtuse to be much use in countering the current narrative. Getting sidetracked into a discussion of statistical misuse simply makes people’s eyes glaze over. This article really did not do much for bringing clarity to the issue. It appears it is not really based on any independent research of actual crime rates vis a vis the illegal immigrant community. Further, the issue is kind of irrelevant. The fact is that criminal aliens are here because of open borders and lack of enforcement. The causation between high numbers of illegal immigrants and criminal alien crime is the same. Failure of the government to enforce the laws of the land. As such, we have hundreds of thousands of crimes and tens of thousands of homicides that were avoidable but for this failure.

    One tidbit that is helpful if elaborated upon is the low level closure rate on homicides in low income minority community which nationwide is about 35%. And, the fact that many criminals are simply deported. Using the low level of closure and using the 25,000 homicide incarcerations reported in the 2011 GAO’s Criminal Alien Statistics Report it might be possible to project a range of say, 25,000 to 75,000 total potentially avoidable homicides due to open borders. Even this report by the GAO appears to be purposely opaque. But, the data is out there most likely in SCAAP (federal reimbursements for incarcerating criminal aliens) data.

    • Sorry, but comprehensive, straightforward, and rigorously documented facts rarely fit on bumper stickers. Understanding details is the often price we must pay for being informed instead of indoctrinated.

      Furthermore, partisans on opposing sides of this issue continuously cite conflicting studies, which does little to advance the truth. This is why it is important for people to understand the flaws in such studies.

  3. Sadly, most people care not the spend the time necessary to learn the truth. They will ignorantly parrot anything they want to be true, even if it is provably false. This is the lazy minded populace we have to try to inform.

  4. Just one question — in looking at incarceration rates of illegal aliens, did you factor out or include crimes strictly related to their immigration status, i.e., individuals incarcerated pending adjudication or deportation whose “crime” is being an undocumented immigrant?

    • Good question. The incarceration rates are strictly for convicts in correctional institutions, so they don’t include immigrants who were incarcerated pending adjudication or deportation.

      There may be some immigrants in correctional institutions strictly for immigration offenses, but such cases are rare, because the government typically does not imprison people just for illegal entry, and unlawful presence in the U.S. is generally a civil violation that cannot result in a prison sentence or criminal conviction. Instead, the government merely returns or removes these people to their homelands.

      If any individuals are in correctional institutions just for breaking immigration laws, they would be in federal prisons, because these are federal laws. Only about 15% of all non-citizen inmates are in federal prisons, and per the U.S. Department of Justice, these convicts have:

      relatively serious prior criminal records. In 2010, 66% of immigration offenders charged in U.S. district court had a prior felony arrest and 57% had a prior felony conviction. During 2010, among immigration offenders with a prior felony conviction, 21% had a prior felony drug conviction and 17% had a prior felony violent conviction.

  5. Excellent and thoughtful examination of the subject. I have been drinking the Cato Kool-Aid on this subject. You opened my eyes.

  6. Jim,

    We need for your group to take another step: write this in the shortest, most rememberable phrasing. This would help get the correct word out.

    • Per the American Heritage Dictionary of the English Language, an “immigrant” is “a person who leaves one country to settle permanently in another.” An “illegal immigrant” is someone who does that illegally. This is a descriptive, straightforward, and accurate term.

      Per U.S. Code Title 8, Chapter 12, Subchapter I, Section 1101: “The term ‘alien’ means any person not a citizen or national of the United States.” This is different than “immigrant,” because it includes foreigners who are temporarily visiting the U.S., and it excludes immigrants who have become citizens.

  7. this is terribly incorrect, purposefully misleading and a pretty good example of bad statistics. of several egregious errors, here are some of the most notable:

    1. you claim that “U.S. Census data from 2011 to 2015 shows that noncitizens are 7% more likely than the U.S. population to be incarcerated in adult correctional facilities” and that “this alone debunks the common media narrative [that immigrants are less prone to crime than native born citizens].” This is obviously untrue; you are comparing two populations that are incarcerated for different reasons. Whereas incarceration is a strong indicator of having committed a crime for native born americans, incarceration for the “foreign born” group that you use to compare to native born americans can be either As someone with an interest (if not an expertise) in conservative public policy you should be familiar with 287g agreements, which allow the federal government to grant funds to municipal authorites to detain undocumented immigrants for no reason other than their immigration status (which is a civil infracture rather than a crime). In fact there are 2,000 immigrants being held in three county jails in New Jersey alone as per their 287g agreements; this number alone brings your supposed 7% difference down to 5%.There are 16 states with 287g agreements and many, many more immigrants detained for their immigration status than the 2,000 in NJ. That’s just one pretty obvious mistake in the causal model you propose above that you could’ve avoided by referring to the economics textbook you linked above on causation in statistical analysis. If you’d read the section on causation, you’d see that adequately proving causation requires empirical correlation, direction of influence and NON-SPURIOUSNESS. Your causation is spurious; the relationship you purport to prove is explainable by an extraneous third factor.

    2. you claim throughout the piece that 10x as many immigrants were deported in 2015 as there were people incarcerated in US jails and prisons. Furthermore, you make the claim that 1.5 million immigrants were deported in 2015. Basic division suggests that you believe that there are 150,000 people incarcerated in US jails and prisons. This is so obviously factually untrue that I have some difficulty believing you made this error by mistake, especially because your argument above about the US census requires that you read US census data that shows that there are upwards of 2 million people in adult correctional facilities (doesn’t include some county jails and halfway houses). While there are roughly 150,000 people in federal prisons (because federal prisons only incarcerate people convicted in federal courts) there are about 3 million people incarcerated in US jails and prisons, which include county jails, halfway houses, state prisons and private prisons. Therefore, your claim that ten times as many immigrants were deported in 2015 as there were people incarcerated in US jails and prisons is at least twenty times wrong.

    3. your third point is nonsensical – besides the obvious error (see point 2, 10 times as many people being deported is flatly wrong) you don’t prove anything that you claim to prove. You claim that “after deporting 10 times more noncitizens convicted of crimes than are in U.S. prisons and jails, they are still 7% more likely to be incarcerated than the general public. This indicates a level of criminality that is multiplicatively higher than the U.S. population.” This is badly phrased and appeals to a statistic that we know isn’t true, so I’ll try to tease out a truth claim for you that we can actually discuss.

    In short, you argue
    a. the entire incarcerated population has a high recidivism rate and therefore often becomes incarcerated again.
    b. by not allowing noncitizens to remain in the country, we do not allow them to reoffend and therefore become incarcerated again
    b. therefore, deporting incarcerated noncitizens should lower the the amount of incarcerated noncitizens in the future.

    This isn’t bad logic, it just doesn’t do what you claim it does. It’s possible that deporting incarcerated noncitizens who committed a crime could reduce the incarcerated noncitizen population. this is a question you would have to test with a causal model to find out the size of the effect on the population in question. instead, you argue without any proof of any kind that the effect of deporting incarcerated noncitizens would be multiplicative. Using the exact same (lack of) logic, I could argue that the effect of my initial criticism (in which i argued that your causal model about crime was spurious because noncitizen people are incarcerated for different reasons than native born people) is negatively multiplicative or for that matter exponential. Without the adequate rigor you wind up skipping the argument and just blowing smoke at the statistic.

    4. Your data misuse section utilizes fails to adequately criticize the second paper you cite. I’ll ignore the first one because it’s basically just intended to be a fact sheet and you break about as many statistical rules as that study does in your purported analysis above.

    Re: the second paper – it proposes multiple causal models in “SELECTION, NETWORKS, AND IMMIGRANT REVITALIZATION” that are put together and tested in “DATA, METHOD, AND LOGIC OF ANALYSIS.” Your discussion above suggests that you cannot interpret a regression or the construction of one; the regressions constructed in the section marked “DATA, METHOD AND LOGIC OF ANALYSIS” test several possible causes for the association between immigration and low crime rates, including states and municipal authorities hiring more police, cultural diffusion, increased social control and the effect of ethnic enclaves on crime. In short, it poses several theories of causation and tests for each. Your criticism was unwarranted and bad.

    I’m not going to go much further on this one. Your criticism of the Cato study is poor and really isn’t deserving of critical thought. Don’t sponsor posts like this if you’re going to claim that you do real research on public policy issues.

    • 1. You either failed to understand the article or are deliberately misrepresenting it:

      • The article does not state that “1.5 million immigrants were deported in 2015.” It states that “1.5 million immigrants who were convicted of committing crimes in the U.S.” were deported “in the decade ending in 2015.” Illegal presence in the U.S. is not a crime but a civil offense, so these are not typical illegal immigrants. These are convicted criminals.

      • The article does not state “that 10x as many immigrants were deported in 2015 as there were people incarcerated in US jails and prisons.” It states that 10 times as many convicted criminal immigrants were deported in the decade ending in 2015 than the number of noncitizens in U.S. adult correctional facilities during 2015.

      That’s five falsehoods you crammed into just two sentences. Any careful reader with basic comprehension skills would never make such errors.

      2. You are wrong about U.S. immigration policy:

      • 287(g) agreements vary by jurisdiction, but they mainly allow local law enforcement to briefly detain illegal immigrants only after they are arrested for committing non-immigration crimes. For example, Cape Cod’s Barnstable County Sheriff states that 287(g) gives its officers “authorization to identify, process, and when appropriate to further detain immigration offenders already in their custody. For the Sheriff’s Office, this means those who have already been arrested, arraigned and placed in our custody by a state judge on a separate and unique local criminal offense.”

      • Your claim that 2,000 immigrants are being held in three county jails in New Jersey under 287(g) agreements for “no reason other than their immigration status” is hogwash. Perhaps that’s why you didn’t provide a shred of evidence to support it.

      3. You don’t seem to understand how regressions work:

      • The second paper I critiqued uses “regression models” of nonexperimental data, and these cannot prove causation. Per the book Regression With Social Data: Modeling Continuous and Limited Response Variables:

      Regression modeling of nonexperimental data for the purpose of making causal inferences is ubiquitous in the social sciences. … This practice is fraught with controversy….

      Friedman … is especially critical of drawing causal inferences from observational data, since all that can be “discovered,” regardless of the statistical candlepower used, is association. Causation has to be assumed into the structure from the beginning. Or, as Friedman … says: “If you want to pull a causal rabbit out of the hat, you have to put the rabbit into the hat.” …

      First, regardless of the sophistication of our methods, statistical techniques only allow us to examine associations among variables.

      Likewise, the book Multiple Regression: A Primer states:

      Multiple regression shares an additional problem with all methods of statistical control…. Unfortunately, there’s no way that we can measure all the variables that might conceivably affect the dependent variable.

      • Worsening the problem above, the second paper tries to infer the criminality of illegal immigrants by measuring the total crime rates of entire states. As the article explains, this is absurd given that illegal immigrants comprise only 4% of the U.S. population in this study. In other words, the signal-to-noise ratio is far too low to determine causation.

      • One does not need a “causal model” to estimate the relative crime rates of non-citizens to the general public. This can be done with straightforward math and logic, and that’s just what this article does.

  8. The footnote in table 41 doesn’t say that these 1.5 million aliens were removed for committing crimes in the U.S. It says they were removed for a previous criminal conviction. The footnote doesn’t indicate where the criminal conviction occurred.

    • It is true that the report containing Table 41 only says these are “persons removed who have a prior criminal conviction.” The fact that these are criminal convictions in the U.S. is proven by an ICE report that provides an appendix of “Key Terms and Definitions” used by ICE. It states, “Convicted Criminal: An individual convicted in the United States for one or more criminal offenses. This does not include civil traffic offenses.”

      This report is cited in the footnotes of Just Facts’ comprehensive research on immigration and crime.

Leave a Reply to Daniel Coghlan Cancel reply

Your email address will not be published. Required fields are marked *