Sunday, March 13, 2011

Crime and Immigration: Facts and Fabrications

Utah’s public discourse on illegal immigration has greatly changed in the last few years. In 2009, the Utah legislature passed HB81, a law that allows local police departments to train their officers to enforce federal immigration law, and ups the consequences ($$$) for employing undocumented immigrants. But this year, the legislative session has churned out a couple of surprises.
Rather than passing Representative Stephen Sandstrom’s Arizona-like law (HB70) in it’s intended form, legislators modified the law, removing from it the provision that would have allowed police to ask immigration questions based on a “reasonable suspicion” that a person they stopped had illegal immigrant status. Perhaps, they didn’t want the anti-racial-profiling response that continues to affect Arizona. The law now requires officers to check the immigration status of anyone they arrest who is on a felony or class B or C misdemeanor charge, something the federal immigration police (ICE) do in the county jails anyway.
The legislature also passed Representative Bill Wright’s bill, (HB116), one that allows for a guest worker program for undocumented immigrants already living in Utah.
These laws continue to be controversial even among immigration activists. The guest worker program will cost immigrants a couple thousand dollars to join, and there’s still some question as to whether or not the federal government will consider a state immigration plan constitutional. However, these legislative moves are landmarks that indicate Utah’s famously ultra-conservative stance on immigration may have shifted to a more moderate position. The big question is, why? What sparked the change? The media has suggested these reasons:
  • The Utah Compact, a declaration of humane principles that should guide immigration legislation, which happens to be signed by several powerful groups and endorsed by one of Utah’s most influential organizations, The Church of Jesus Christ of Latter-Day Saints.
  • Business and Farm lobbyist arguments for the need of migrant labor.
  • Major efforts by the conservative media to portray undocumented immigrants in a more sympathetic light. (KSL, Deseret News)
  • New (2011) comprehensive research on undocumented immigrants and crime by Sutherland Institute and Brigham Young University, which reveals that of all the inmates in Utah jails and the state prison, only 5.4% are undocumented immigrants (3.8% if you don’t count the illegal-status inmates who have already served their criminal sentence and are simply detained as they await trial for federal immigration violation).
The last item was a brilliant response to a concern held by Utahns for the past few years that while very few reputable studies have been published to give us a clear picture of immigration and crime, politicians and activists have proposed conflicting answers based on estimations formulated without following academic standards essential for accurate statistics.
"There are three kinds of lies: lies, damned lies, and statistics." (Mark Twain)
The intent of this blog post is to discuss the nature of the immigration-and-crime concern, to describe the events leading up to the new research, and to reveal the flaws in previous research.

It was and still is commonly supposed that the majority of illegal immigrants are involved in not just a lack of proper documentation, but in a variety of other crimes. Here are a few examples of Utahns' concerns about immigration and crime taken from the online comments following a 2008 Deseret News article on illegal immigration and compassion:
The majority of violent crimes, foreclosures, and other negative situations where I live directly involve foreigners.
Illegal aliens have killed more Americans since 911 than have died in the war in Iraq.
We don't need law breakers in the USA who bring drugs, murder and all other kinds of chaos to our country.
I don't think anyone cares if people come here legally. It's the ones who lie, steal and break American laws to get here who are NOT WELCOME!
The criminal mind believes that everything legal Americans have worked hard for belongs to them FREE OF CHARGE. All you can do is arrest them and quit using religion to keep them here.
Although these quotations certainly take an extreme position, the assumed connection between undocumented immigrants and "the criminal mind" outlines a paradigm that would be difficult for people to question, especially if reliable sociological crime statistics are scarce.
In June of 2009, an organization finally published some significant statistics in answer to this question: What percent of Utah crime is committed by undocumented immigrants? Sutherland Institute produced a simple study called, Just the Facts showing that illegal immigrants represent a surprisingly small percentage of jail populations. Quickly afterward, state representative, Chris Herrod (R), made public criticism of Sutherland’s study, reasoning that Sutherland’s numbers couldn’t possibly be true, and providing a few shocking numbers of his own. It wasn’t until July 2010 that Professor Charlie Morgan and his sociological research team at BYU completed a thorough study of statistics on undocumented immigrants in Utah jails and prisons. In February, 2011, Professor Morgan collaborated with Sutherland Institute to put out the information as a formal paper, How Much Crime Do Undocumented Immigrants Commit in Utah?
The following is my detailed analysis and assessment of the statistics produced by Sutherland Institute’s first study, and by Representative Herrod. I explain what they said, where they went wrong, and why a more comprehensive study was needful.

Just the Facts

In Sutherland Institute’s Just the Facts, the most significant finding was that only 3.9 percent of county jail inmates in Utah are undocumented immigrants. This figure is similar to the percent of illegal immigrants estimated to be residing in Utah (4%). That is to say, illegal immigrants' representation in county jails is equally proportional to their representation in the Utah population.This news was shocking to those who imagine that illegal immigrants are committing much more crime than other demographic groups.
Sutherland's research methods were legitimate, but limited. They were justified in using county jail statistics, a more accurate method of crime measurement than other proposed methods, such as the use of arrest rates. (The use of arrest statistics would be unreasonable, because legal status in most cases remains undetermined until the arrested arrive in jail.) However, Sutherland's study included only data collected on one occasion, January, 2009. Here is the information they give regarding their method of data collection: Sutherland contacted every county in Utah between January 15 and January 30, 2009 to request information about the legal status and ethnic origins of inmates in county jails. Receiving and verifying the responses required several months. Twelve jails responded with ethnicity data and 17 responded with data on the number of inmates with criminal charges who have also been identified as undocumented immigrants and are being held for federal immigration officials.
In a July 15th, 2009 press release, Sutherland responded to critics of their article and included the following explanation:
We contacted the county jails and they responded with the figures we reported. Several counties, such as Salt Lake, provided documentation by fax or email. Others, such as Utah county, reported them over the phone.
What remained unclear, in Sutherland's data collection, is whether the statistics they received from each county were the jail residence numbers over the whole month of January, or the numbers of inmates in residence only on the day they happened to report to Sutherland. The day the numbers represent and even the time of day may be significant, because illegal status inmates are sometimes shifted from one county jail to another, and may be under-counted or double-counted. A greater question might be this: how much do the jail population numbers fluctuate over the course of a month or a year? A good quantitative report on a jail population must take into consideration the logistics and procedures practiced in the jail system.

County Jail Logistics and Procedures

When a U.S. citizen or legal resident is arrested, he or she is brought to the county jail, may be bailed out before trial, attends trial, and is given a sentence. If his or her sentence is to serve time (rather than to simply pay a fine or complete community service) he or she will be either returned to the county jail, or sent to the state penitentiary, depending on the severity of the crime. When an undocumented immigrant commits is arrested, he or she is required to follow the same procedures. Additionally, the county jail records label such an inmate with a special status that indicates he or she is undocumented and is being monitored by ICE, the federal immigration police. The inmate's status would be "ICE-Hold with Misdemeanor" or "ICE-Hold with Felony." After an undocumented immigrant completes a county jail or state prison sentence he or she is not released to go home. Instead, he or she is put on "ICE-Hold Only" status and is relocated to one of the two county jails designated for "ICE-Hold Only" inmates (Weber county or Utah county jail). This is like a second arrest; the inmate waits in the Weber or Utah county jail until he or she is granted a trial, at which time he or she will likely be sentenced to deportation.
Interestingly, if an undocumented immigrant is arrested for a felony or misdemeanor, and found innocent by the court, he or she will still be put on an ICE-hold and may be deported.
Sutherland's report made a significant flaw: the study put all undocumented inmates into one group without distinguishing between those have not yet served time for their felonies and misdemeanors from those who only remain in the facility due to their illegal status. If the public is interested in knowing the numbers of those serving time in the jails for property and violent crimes, then a mixed sum of ICE-Hold with Felony/Misdemeanor + ICE-Hold Only inmates is unhelpful. How could jail statistics help us understand what percent of crime is due to the undocumented population if undocumented inmates are made to stay in the jail much longer than other demographic groups? The Simple Answer: unveil which ICE-Hold categories the inmates are in, and it will be clear how many inmates have already served their sentences. The recent (2011) study compiled by Sutherland and BYU’s Professor Charlie Morgan, does make just such a distinction and shows that the vast majority of undocumented inmates are ICE-Hold Only inmates.

Legislator Guesswork

A few days following the publication of Sutherland's study, Utah Representative, Chris Herrod (R-Orem), a zealous advocate for SB81, sent a vehement email message to his fellow state legislators. It was published for the public on The Senate Site--the Utah Senate Majority blog. The intent of Herrod's letter was to warn others that Just the Facts had distorted the facts. Unfortunately, Herrod's letter contains a few distortions of its own.

Herrod's Letter in Summary

Herrod's Points:

Where He Went Wrong:

The percentages published by Sutherland Institute must be inaccurate--they can't really know how many illegal immigrants are in the jails until they train police as ICE officers as HB81 suggests.

The police at county jails can and do perform accurate documentation checks on their inmates to determine immigration status.

Sutherland's data looks skewed because the percent of illegal immigrant jail population is not proportional to the percent of total illegal immigrant population for some counties. Some county jails have a noticeably lower (or higher) number of illegal immigrant inmates relative to the corresponding county population of illegal immigrants.

What Herrod didn't know is there are only two Utah jail facilities authorized to hold ICE-Hold Only inmates. This is the reason Weber and Utah County facilities hold many more undocumented inmates than what would seem proportionally normal.

There are better ways to determine what percentage of jail inmates are illegal, such as taking the percent of Hispanics in the jails and multiplying it by 38-48%, which is the estimated percent of Utah Hispanics who are illegal immigrants.

"Hispanics" and "illegal immigrants" are not the same group. (Nationally, only 76% of undocumented immigrants are Hispanic. And only about a third of Utah Hispanics are undocumented immigrants.) It is completely unreasonable to make guesses about the number of undocumented jail mates when we can simply count them.

It doesn't make sense that illegal immigrants would be committing crimes at the same rate as legal Hispanic immigrants, so the percentage of illegal inmates in the jails is probably much higher than 38-48% of the number of Hispanic jail mates.

Herrod forgets that "Hispanic," is a broad group that includes U.S. born Hispanic citizens, not just immigrants. According to National statistics, second and third generation immigrants (including Hispanics) are more likely to be involved in crime than their immigrant parents or grandparents. This is likely the case in Utah.

Police department arrest rates for Hispanics might be an even better way to determine the number of illegal immigrants committing crime in Utah.

Herrod is again conflating “Hispanic” with “illegal immigrant” in order to substantiate his argument, not to mention the fact that arrest rates are more likely to include a racial profiling bias against immigrants.

Herrod was right about Sutherland Institute’s original study not being comprehensive enough, but he was wrong about the numbers.

Herrod's Letter in Detail

Herrod cites an example to support his argument that Sutherland's inmate numbers must be wrong. "For Salt Lake County Jail, Sutherland lists only 29 inmates on federal hold. Yet, monthly reports by the Salt Lake County Jail from January to May of 2009 show no month with less than 85 inmates detained for an immigration hold." Thus, Herrod legitimately points out a possible flaw in Sutherland's data-gathering method. It is unclear whether the numbers jails reported to Sutherland represented the number of undocumented inmates present on one day of the month, or the number of inmates on the roster throughout the entire month. Likewise, a one-month look at jail numbers is much less informative than a month-by-month year long survey of the jail population.
Because Herrod believes that simply asking the county jails to count is not an accurate way to determine the number of illegal immigrants in the jails, he performs a speculative series of quasi-calculations to hypothesize what, to him, seems a more reasonable statistic range. He insists that "until better data is available, ethnicity must be used [rather than jail attendance counts] since race and ethnicity are often the only information available." Although Herrod does not cite specific publications, he explains that some of his population statistics come from the Pew Hispanic Center, a reputable, nonpartisan research organization. He states that according to the Pew Hispanic Center, 76% of illegal aliens are Hispanic, and 38-48% of Hispanics in Utah are illegal aliens. He claims that the percent of Utah crimes being committed by illegal aliens can simply be determined by taking 38-48% of Hispanic inmates as a starting point. This is a significant problem. First of all, we really can count the number of undocumented immigrants in the jails--thus, heuristically guessing at the number is unnecessary. Secondly, "Hispanics," and "undocumented immigrants" are completely different types of demographic groups. Sociologists and statisticians know that while these two groups certainly overlap, the social, recreational, or criminal tendencies of one group do not determine the tendencies of the other. The fact that 2 out of 5 Hispanics in Utah are illegal does not mean that 2 out of 5 Hispanics at Bountiful High School are illegal. Nor does it mean that 2 out of 5 Hispanics at the bowling alley are illegal, or that 2 out of 5 Hispanics at the post office are illegal. It would be inaccurate to assume that 2 out of 5 Hispanics in the county jail are illegal.
Moreover, national statistics show that crime trends for "foreign-born” groups, and "Hispanics" are very different. 76% of illegal immigrants are Hispanic (a national percentage); Since not all illegal immigrants are Hispanic, the two demographics are only partially correlated. To assume complete correlation in order to supply information from one demographic in place of unknown information on the other is partially speculative and unreliable. But Herrod doesn't stop there. In his quest to rebut Sutherland statistics by manipulating some of his own, he confesses that more guessing needs to be done, and describes three options for tinkering with his estimate:
The only options are that illegal Hispanics are causing crime at a greater rate than the general population, legal Hispanics are committing more crime, or both categories are equally committing crime at a higher rate. It must be one of these three. I believe it is the first.
This statement is a little confusing, but I believe what he means to say here is that he thinks there are probably more illegal immigrants in the jails than just 38-48% of the total number of Hispanic inmates. Herrod then explains his concern with Sutherland's numbers regarding the Utah Hispanic population, which state that in the county jails Hispanics make up 17.7 percent of the inmate population, whereas the Hispanic representation in the total Utah population is 11.6 percent (in 2007). Herrod goes on to explain his numbers further and then takes the opportunity to interject his own bias.
Ironically, Sutherland's own data shows that Hispanics commit crime at roughly 50% higher than the general population. If Sutherland's research is correct and illegal aliens do not commit crimes at a higher rate than the general population, then Sutherland is asserting that Legal Hispanics have a higher crime rate than illegal Hispanics. does Sutherland really believe this? I certainly do not. this is precisely why so many legal Hispanics are against illegal immigration. Legal immigrants have gone through a screening process. Foreign students have as well. It only makes sense that legal immigrants have lower crime rate otherwise the U.S. is wasting a lot of money at embassies throughout the world.
When Herrod cites Sutherland as saying that illegal aliens do not commit crimes at a higher rate than the general population, he's referring to Sutherland's finding that illegal immigrants' representation in the county jails (3.9%) was proportional to their representation in the Utah population (an estimated 4%). His concept of the crime rate of "the general population" seems to imply that the way he sees it, most people belong to a demographic group whose percent representation in county jails is equally proportional to it's percent representation in the Utah population. However, his use of the term "the general population" to mean groups other than Hispanics and immigrants, adds a biased Us versus Them theme to his writing. He's in disbelief over the notion that Sutherland's statistics do not support his assumption that illegal aliens are much more likely to commit crimes than "the general population."
Herrod's main concern as he compares legal Hispanics and illegal Hispanics is badly stated, and shows he misunderstands the data. First of all, when he claims it doesn't make sense that, "legal Hispanics," should be overrepresented in county jails, he clearly means legal Hispanic immigrants, but Sutherland's data showing that Hispanics are overrepresented in county jails refers to Hispanics in general, not specifically to legal Hispanic immigrants. It includes both legal and illegal Hispanic immigrants as well as U.S. born Hispanics. Therefore, his logic that legal immigrants go through a screening process, and should, therefore, be less likely to commit crime than illegal immigrants, sounds reasonable, but is mostly irrelevant. Meanwhile, national studies on immigration and crime show evidence that second and third generation immigrants (including U.S. born Hispanics) are more likely to commit crime than first generation immigrants. This is likely the reason Utah Hispanics are over-proportionally in the jails, while illegal immigrants are proportionately represented.
Although Rep. Herrod does not complete a start-to-finish calculation of the percent of illegal immigrants committing crimes, his statements imply that he believes it must be significantly more than 6.7-8.5% of county jail inmates (38-48% of the Hispanic jail inmate population). In addition to the fact that his conjecture is purely guesswork, Herrod's real error is using "Hispanics" as a near proxy for illegal immigrants.
Herrod further projects that "a much better indicator of crime would be arrest rates which show a dramatically higher rate of crime for Hispanics." Not only does he continue to use statistics on Hispanics to estimate data for illegal immigrants, but he delves into a region of numbers most social scientists would consider "off-limits," due to the high margin of error created by the ever-prevailing phenomenon of racial profiling. That is to say, because police may tend to find people with darker skin more suspicious, it is very likely that rates of police interaction would be an inaccurate depiction of crime rates. Herrod quotes Salt Lake City Police Department's Chief Burbank's expression of concern about racial profiling in the second half of his message. He also refutes anticipated opposition to the use of this kind of data by reasoning that, "Political correctness is destroying this nation."
However, Herrod describes some statistics he received from the Salt Lake City Police Department. Their numbers for 2008 were: "24.53% Hispanic, and 75.47% Non-Hispanic" adult arrests. His criticism of the percentages is interesting:
What was most troubling, however, is what the new data showed. It showed that SLCPD had included "Unknown" in the "Non-Hispanic" column the month before thus greatly skewing the data. "Unknown" are not "Non-Hispanic." While 24.53% Hispanic is higher than the general is not so dramatic. But divide "Hispanic" by "Hispanic" plus "Non-Hispanic" as any reasonable person would and the results become startling.
A few paragraphs later, Herrod gives some information that explains why his way of dividing "Hispanic" by "Hispanic," plus "Non-Hispanic" would be startling:
I added the column on % of Hispanics using Hispanics and Non-Hispanics as the denominator--not including "Unknown Ethnicity."
One would assume that by this he means rather than doing the simple percentage calculation Salt Lake City Police Department did:

# of Hispanic Arrests

# of Total Arrests

% of Hispanic Arrests

He instead performed this operation:

# of Hispanic Arrests

# of Total Arrests

% of Hispanic Arrests

which would indeed make the results startling. Unfortunately, his mathematical procedure is incorrect. The only way to calculate a percentage is to divide the category in question by the total; you cannot simply take a chunk out of the total in the denominator because you don't want it there. Herrod made a similar mistake when calculating a percentage he presented to fellow legislators (not in the email message I discuss here) including Representative Carl Wimmer, who reported it as "hard statistics" to KSL News. According to Herrod, 81% of Salt Lake City's homicide arrests in 2008 were Hispanic. In a May 3, 2010 report KSL explains:
Herrod told KSL he received the numbers from Utah's Bureau of Criminal Identification (BCI), so we analyzed those numbers.
According to the BCI statistics for 2008 in Salt Lake City (shown above), there were 18 arrests for various types of homicide. Of those, 9 arrests were of suspects described as Hispanic, two were non-Hispanic, and seven were of an unknown ethnicity.
Herrod says he discarded arrests in the "unknown" category because there was no way to tell what ethnicity they were, and that's how he came up with the percentage of 81.
"If somebody is listed as unknown, should they be counted? Well, you can't determine their ethnicity, so I believe they should be thrown ... that stat should be thrown out," Herrod said.
By simply ignoring over a third of the total, Herrod manages to manipulate the statistic, stretching it out of proportion. Meanwhile, I must impress again that using crime statistics regarding Hispanics is not an accurate way to determine what percentage of crime involves undocumented immigrants.

To Conclude

Statistical errors, like the ones Herrod made, are not uncommon among politicians, and other persons of power and influence. People make mistakes. As a society, it is our responsibility to question the numbers we hear in the media, to go to the formal research sources, and to check the math.

1 comment:

  1. When I purchased copies of several crime reports in which I was the victim in Utah in 1995, I was dismayed to find out that the police of my county were forbidden from checking the box labeled "Hispanic" on the form. I brought the "error" of identifying the criminal properly to the attention of the officer present who happened to be black, and she told me of the policy change.
    Among many things I can say about this is: Re-categorizing the criminals' ethnicity does not fix the problem.