Citizenship_Status_and_Arrest_.pdf

Citizenship Status and Arrest Patterns for Violent
and Narcotic-Related Offenses in Federal Judicial
Districts along the U.S./Mexico Border

Deborah Sibila1 & Wendi Pollock2 & Scott Menard3

Received: 8 September 2016 /Accepted: 30 October 2016 /
Published online: 10 November 2016
# Southern Criminal Justice Association 2016

Abstract Media reports routinely reference the drug-related violence in Mexico,
linking crime in communities along the Southwest U.S. Border to illegal immigrants.
The primary purpose of the current research is to examine whether the media assertions
can be supported. Logistic regression models were run to determine the impact of
citizenship on the likelihood of disproportionate arrest for federal drug and violent
crimes, along the U.S./Mexico border. In arrests for homicide, assault, robbery, and
weapons offenses, U.S. citizens were disproportionately more likely than non-citizens
to be arrested. The only federal crime where non-citizens were disproportionately more
likely to be arrested than were U.S. citizens was for marijuana offenses. Results of the
current study challenge the myth of the criminal immigrant.

Keywords Citizenship . Arrest . Criminal immigrant . Gender

The myth of the criminal immigrant is perhaps one of the single most controversial
factors contributing to America’s present day anti-immigrant fervor. In their book, The

Am J Crim Just (2017) 42:469–488
DOI 10.1007/s12103-016-9375-1

* Wendi Pollock
[email protected]

Deborah Sibila
[email protected]

Scott Menard
[email protected]

1 Department of Government, Stephen F. Austin State University, Box 13045 SFA Station,
Nacogdoches, TX 75962, USA

2 Department of Social Sciences, Texas A&M University, 6300 Ocean Drive, Corpus Christi,
TX 78412, USA

3 Institute of Behavioral Science, University of Colorado, Boulder, USA

http://crossmark.crossref.org/dialog/?doi=10.1007/s12103-016-9375-1&domain=pdf

Immigration Time Bomb, authors Richard D. Lamm and Gary Imhoff contend that the
issue of immigration and crime is a critically divisive topic easily subject to misinter-
pretation (1985, p. 21). The belief that immigrants are more crime-prone than native-
born is not a twentieth century development. Debates on this controversy date back
more than 100 years (Hagan & Palloni, 1998; Martinez & Lee, 2000). Hagan and
Pallon believed that the nexus between immigration and crime is so misleading that it
constitutes a mythology (1999, p. 630). In a special report for the Immigration Policy
Center, professors Ruben Rumbaut and Walter Ewing wrote B[The] misperception that
the foreign-born, especially illegal, immigrants are responsible for higher crime rates is
deeply rooted in American public opinion and sustained by media anecdote and
popular myth^ (2007, p. 3). Lee (2013) similarly argues that immigrants have a long
history of serving as scapegoats for a vast array of America’s societal problems
including crime.

Public opinion surveys suggest that a significant number of Americans believe that
immigrants, particularly illegal immigrants, are associated with higher crime rates
(Kohut et al., 2006; Muste, 2013; Sohoni & Sohoni, 2014). Media sources routinely
associate immigration, especially Hispanic immigrants, with crime (Bender, 2003;
Martinez, 2002). Politicians also play a key role in perpetuating the belief that
immigrants and crime are interrelated. Arizona Governor Jan. Brewer, Senator John
McCain and former presidential hopeful Patrick Buchanan are just a few political
figures that have gone on record linking immigration directly with high crime rates
(Butcher & Piehl, 1998; USA Today, 2011). On May 15, 2006, during a presidential
address to the nation on immigration reform, former President George W. Bush
asserted that BIllegal immigration puts pressure on public schools and hospitals, it
strains state and local budgets and brings crime to our communities.^ According to
Rumbaut and Ewing (2007), regardless of the much-publicized media stereotyping
and harsh political rhetoric, empirical evidence simply does not support the popular
misperception that immigration is the cause of higher crime rates in America.

History of Immigration and Crime Theory and Research

Explanations of the link between immigration and crime have been offered from the
perspectives of culture conflict, acculturation, social disorganization and the
immigration revitalization perspective. From the culture conflict perspective, Sellin
(1938) suggested that the conflict between the norms of behavior for divergent
cultures, as represented by native born Americans versus immigrants, was one source
of crime. Sutherland (1924, 1934) posited that it was not immigration itself, but rather
acculturation, that led to the association between immigration and crime, and noted
that second-generation immigrants had higher crime rates than first-generation immi-
grants. From the social disorganization perspective, researchers from the Chicago
School linked immigration to a number of social issues, including not only crime but
also poverty, unemployment, poor housing, and substandard schools (Park et al.,
1925; Shaw, 1929; Shaw & McKay, 1931, 1942; Thomas & Znaniecki, 1920,
1958). Shaw and McKay, in particular, suggested that it was not the characteristics
of the immigrants themselves, but the characteristics of the urban neighborhoods in
which they resided, that led to the apparent link between immigration and crime (see

470 Am J Crim Just (2017) 42:469–488

also Taylor, 1931). In contrast to the culture conflict, acculturation, and social
disorganization perspectives, the immigration revitalization perspective (Lee, 2013;
Martinez, 2006) suggests that immigrants tend to be less criminal than native-born
Americans, and that the informal social controls that are an integral part of the culture
in predominantly immigrant neighborhoods result in higher levels of social organiza-
tion and lower rates of crime. The present paper is informed by, but not a direct test
of, these theories, which disagree about whether immigrants should have dispropor-
tionately higher (culture conflict, social disorganization; acculturation for the second
generation), or lower (immigrant revitalization; acculturation for the first generation)
involvement in illegal behavior.

Early empirical investigations into the link between immigration and crime were
limited and were focused at the individual level (Abbott, 1915; Hourwich, 1912; Lind,
1930; Taft, 1933, 1936; Van Vechten, 1941). These studies found little evidence of a
causal relationship between immigration and crime. Three major government commis-
sions (the Industrial Commission of 1901, the [Dillingham] Immigration Commission
of 1911, and the [Wickersham] National Commission on Observance and
Enforcement of 1931) similarly explored the issue of whether immigration increases
crime. Each of the commissions found that immigrants were less likely to commit
crime than were their native-born counterparts.

More recent studies investigating the association between immigration and crime
tend to agree with earlier research, specifically that immigrants are not disproportion-
ately involved in crime and are oftentimes significantly less involved than native-born
Americans (Hagan & Palloni, 1998; Martinez & Lee, 2000; Mears, 2002; Rumbaut &
Ewing, 2007). Contemporary researchers examining the immigrant-violent crime
nexus have concentrated their efforts on macro-level studies, conducting both neigh-
borhood and city-level studies. Findings from macro-level research surrounding the
immigration-violent crime question have been more inconsistent than the individual-
level studies. There are a handful recent studies that find some positive relationships
when examining the impact of immigration and certain crime variables under specific
conditions (Lee, et al., 2000; Martinez, 2000, 2003; Sampson & Raudenbush, 1999).
A significant portion of contemporary research examining the relationship between
immigration and crime has focused on ethnic gangs and violent crime, especially
homicide. More recently, researchers have explored the idea that immigration has
actually contributed to the decline in U.S. crime rates since the 1990s (Sampson,
2006, 2008). In addition to recent academic studies, at least one government com-
mission was tasked with examining the issue of immigration and crime during the last
20 years. The U.S. Commission on Immigration Reform (1994) examined the impact
of Mexican immigration on crime rates in metropolitan areas along the Southwest
border as compared to those in non-border cities. The commission found that crimes
rates were typically lower in cities with large Mexican populations along the border
than for non-border cities.

The Southwest Border

Because of the current study’s focus on the impact of immigrant status on violent
crime and narcotic-related offenses in federal jurisdictions along the United States-

Am J Crim Just (2017) 42:469–488 471

Mexico border during a time of unprecedented drug violence in Mexico, it is
important to provide some background information regarding the location and
timeframe for the research. The U.S.-Mexico border commonly referred to, as the
Southwest border is an international boundary separating the United States and
Mexico. The border was created in 1848 after the end of the Mexican-American
War under the Treaty of Guadalupe-Hidalgo. The Southwest border is one of the
longest borders in the world (1954 miles) and runs along the U.S. states of
California, Arizona, New Mexico and Texas on the northern side and the Mexican
states of Baja California, Sonora, Chihuahua, Coahuila, Nuevo Leon and Tamaulipas
on the southern side. It is also considered to be the busiest international boundary in
the world (Andreas, 2000). The La Paz Agreement signed by the United States and
Mexico in 1983 specifies that the Bborder region^ between the U.S.-Mexico en-
compass the band of land that stretches 100 km (approximately 62.5 miles) on either
side of the boundary line. There were approximately 70.9 million residents living in
the four border states in 2010, with approximately 7.5 million people residing in 37
counties that comprise the border region (Rex, 2014). According to the U.S. Census
Bureau, the border region contains the largest concentration of Hispanic population
in the country — from 25 % to more than 50 % of the population (Ennis et al.,
2011). The Pew Research Center estimates that in 2010 approximately 4.7 million
illegal immigrants lived in the four states along the Southwest border (Passel &
Cohn, 2011).

The movement of illegal drugs and movement of unauthorized immigrants from
Mexico into the United States via Southwest border have long been areas of
contention in U.S.-Mexico relations (Andrews, 2012; Domínguez & De Castro,
2009; Payan, 2006; Seelke, 2010). For decades, marijuana and heroin (primarily
Mexican brown and black tar heroin) were smuggled across the border without
significant interference by law enforcement in either country (Andreas, 2000).
During the 1980s, Mexico became an important transshipment center for drugs
entering the United States following the crackdown on the importation of
Colombian cocaine and marijuana through South Florida. The crackdown led to
the redirection of federal drug interdiction efforts from South Florida to the
Southwest border. The passage of the Immigration Reform and Control Act of
1986, or IRCA, which granted amnesty to millions of undocumented immigrants
living in the United States, further contributed to public scrutiny of Southwest
border issues and the problem of undocumented immigration (Immigration Reform
and Control Act of 1986, n.d.). Efforts by law enforcement agencies to curb further
illegal immigration across the Southwest border following the passage of the IRCA
legislation proved essentially futile (Payan, 2006). During the 1990s, the U.S.
federal government increased the number and scope of interdiction operations
targeting both drugs and illegal immigrants attempting to cross the United States-
Mexico border (e.g., Operations Gate Keeper, Hold the Line, Safeguard, and Rio
Grande). These operations made little inroad into the drug and immigration prob-
lems plaguing the border region (Andrews, 2012; Domínguez & De Castro, 2009;
Payan, 2006). Meanwhile, Mexico’s efforts to curb the expansion of drug cartel
influence during the 1990s were largely ineffective. The cartels expanded their
influence throughout Mexico, and drug-related corruption was rampant throughout
the country’s government, police organizations and military (Andreas, 2000;

472 Am J Crim Just (2017) 42:469–488

Carpenter, 2009). The U.S. federal government dramatically escalated border polic-
ing following the 9/11 terrorist attacks in New York and Washington. Perhaps the
most affected area in the U.S. after the attacks was the Southwest border (Payan,
2006, p. 13; see also Domínguez & De Castro, 2009). National security concerns
following 9/11 led to the tightening of border security checks, hardening of land and
sea points of entry, crackdowns on unauthorized immigration and increased
militarization of the border. Despite all these safeguards, public perception that
government and law enforcement efforts to secure the Southwest border are
substantially ineffectual persists. Cornelius (2005) suggests that the government’s
efforts to secure the border are not only inadequate, but have been more effective
keeping illegal immigrants inside the U.S. than acting as an deterrent to others
attempting to enter the country (p. 12). Americans remain divided over how to stem
the flow of illegal immigrants across the Southwest border.

The current image of the Southwest border as a region plagued by violence and
crime is intrinsically linked to the rising tide of drug-related violence in Mexico
during the last ten years (Beittel, 2009, 2011; Carpenter, 2012; Payan, 2006).
Violence in Mexico significantly worsened following Mexico’s President Felipe
Calderon’s declaration of war against his country’s drug cartels shortly after his
inauguration in December 2006. The level of drug-related crimes (i.e., homicides,
kidnappings, home invasions, drive-by shootings) increased throughout Mexico,
particularly in the northern territories along the United States border (Beittel, 2009,
2011; Domínguez & De Castro, 2009). According to Human Rights Watch (2013),
approximately 60,000 Mexicans lost their lives as a result of drug-related violence
during President Calderon’s tenure as president (2006–2012).

Since 2007, the media in United States, especially those along the Southwest
border have been preoccupied with the drug violence and bloodshed in Mexico
(Andrews, 2012; del Bosque, 2009). Media sources routinely sensationalize re-
ports of U.S. citizens dying and being kidnapped in suspected drug-related
incidents in cities along the Southwest border. The media coverage has persuaded
Americans that the turmoil in Mexico is no longer confined to that country
(Correa-Cabrera, 2012). Many are convinced that Mexico’s drug-related violence
has Bspilled over^ into communities along the Southwest border. ‘Spillover’ has
become a new media buzzword (del Bosque, 2009). While there is no firm
definition for the term spillover violence it is generally understood to be violence
that occurs as a result of drug trafficking. Civilians, law enforcement officers and
other criminals or criminal organizations can all be the targets of spillover
violence (Lee & Olson, 2013, p. 100). There is some debate as to whether
spillover violence is an actual phenomenon (Correa-Cabrera, 2012; Lee &
Olson, 2013). Without a concise definition, the ability of law enforcement to
identify and measure spillover violence is problematic. In a report to Congress
regarding the issue of Southwest border violence, Finklea and colleagues note that
Bno comprehensive, publicly available data exists that can definitively answer the
question of whether there has been a significant spillover of drug trafficking-
related violence into the United States^ (2010, p. 19). Further complicating the
issue of spillover violence has been the media linkage of the problem to the
Bflood^ of unauthorized immigrants moving across the Southwest border (del
Bosque, 2009). Proponents of stricter immigration controls argue that increased

Am J Crim Just (2017) 42:469–488 473

border security will significantly impede the flow of undocumented immigrants
thereby reducing rates of violent and property crimes in border communities
(Nevins, 2002; Payan, 2006).

The Federal Criminal Justice System and the Southwest Border

Criminal offenses occurring along the Southwest border can be prosecuted in
either state or federal court. The primary trial court in the federal judiciary is
the U.S. district court. There are currently 94 federal judicial districts in the United
States, each with its own district court. Each district has different judges, court-
house cultures and specialized local needs (Abrams et al., 2010). There are five
federal districts along the Southwest border: The District of Arizona, the District
of New Mexico, the Southern District of California, the Southern District of
Texas, and the Western District of Texas (Fig. 1).

Each district has its own U.S. Attorney. The U.S. Attorneys’ Office (USAO)
represents the federal government in all cases where the United States is a party. It
has the discretion to decide whether federal charges will be filed in a case or if the
case should be referred to the state system. Some factors that may affect the
USAO’s choice between federal or state prosecution include: (1) primary investi-
gative jurisdiction; (2) custody of the suspect; (3) possibility of a duplicative
prosecution; (4) caseload and resources; (5) legal advantage; and (6) inter-
agency relationships and relations among agents (Abrams et al., 2010). Federal
crimes prosecuted by the USAO are listed under various titles of the United States
Code (USC) including Title 18 (the primary criminal and penal code of the U.S.
federal government) and Title 26 (the Internal Revenue Code).

Fig. 1 Counties in southwestern judicial districts

474 Am J Crim Just (2017) 42:469–488

Current Study

The following research questions will be addressed:
Research question 1. What are the characteristics (age, gender, ethnicity, marital

status and citizenship) of individuals arrested for the following federal offenses along
the U.S.-Mexico border?

& 1a.Homicide
& 1b. Assault
& 1c. Robbery
& 1d. Marijuana
& 1e. Hard Drugs
& 1f. Weapons

Research question 2. Are non-citizen arrestees disproportionately more likely to be
arrested for any of the following offenses, when compared to U.S. citizen arrestees?

& 1a.Homicide
& 1b. Assault
& 1c. Robbery
& 1d. Marijuana
& 1e. Hard Drugs
& 1f. Weapons

Data

Data for the current study comes from the Federal Justice Program (FJSP):
Arrests and Bookings for Federal Offenses (2007–2010) research series. The datasets
contain comprehensive information about suspects and defendants processed in the
federal criminal justice system during fiscal years 2007 to 2010. Offenders arrested for
federal offenses are transferred to the custody of the United States Marshals Service
(USMS) for processing, transportation, and detention. The current dataset was con-
structed from the USMS Prisoner Tracking System (PTS) database. Records include
arrests made by federal law enforcement agencies (including the USMS), state and
local agencies, and self-surrenders. The USMS uses the PTS application to maintain
tracking information for all federal prisoners in USMS custody. The PTS was imple-
mented by the USMS in March 1993 to maintain tracking information for federal
prisoners and to monitor federal prisoners in state and local detention facilities under
contract to the USMS. The PTS replaced the Prisoner Population Management System.
The PTS contains information that is specific to each individual prisoner, including the
prisoner’s personal data (e.g., gender, ethnicity, age, marital status and citizenship
status), property, medical information, criminal information, and location. Prisoners’
records are created using information derived from key source documents (e.g., USMS
and other agency arrest sheets), and this information is entered into the PTS.

The rationale for the location and time setting of this research warrants some
discussion. While USMS arrest statistics are collected in all federal judicial districts

Am J Crim Just (2017) 42:469–488 475

nationwide, the current research will only utilize data from the five federal districts on
the U.S.-Mexico border (Arizona, New Mexico, California Southern, the Texas
Western and Southern Texas). The five Southwest border districts accounted for
56 % of all federal suspects arrested and booked in the U.S. and 90 % of all
immigration arrests in 2010 (Motivans, 2012). This study departs from most existing
research on immigration and crime, in that the focus is on a variety of federal offenses
committed in five federal judicial districts whose jurisdiction extends to multiple states
along the Southwest border. Previous research has almost been exclusively limited to
examining a limited number of offenses (i.e. homicide) at either the city or neigh-
borhood level (Martinez, 2006). The time period covered by this research (2007–
2010) is significant because it covers a period of unprecedented drug violence in
Mexico that was thought to contribute to crime levels in states along the Southwest
border. Media reports routinely referenced the drug-related violence in Mexico,
linking crime in communities along the Southwest border to drug cartel members
and illegal immigrants.

Variables

The dependent variables in this study are six different measurements of arrests. In each
case the prevalence of arrest was used, where 1 = arrested and 0 = not arrested in the
respective year (2007–2010). The six forms of arrest include marijuana, hard drugs
(heroin, cocaine, hallucinogens, other opiates, amphetamines, barbiturates, and syn-
thetic drugs), homicide, robbery, assault, and weapon offenses. Prevalence was used
instead of frequency of arrest due to the fact that it is extremely rare that an individual
would be arrested twice in the same fiscal year for a federal offense.

Independent variables in the current study include age, sex, race, citizenship status,
and marital status.

The arrestee’s sex is measured with a dichotomous variable that is coded B1^ if the
offender is male and B0^ if the offender is female. Ethnicity of the arrestee is broken
down into three categories: White/White Hispanic, Black/Black Hispanic, and Other.
Ethnicity is broken into three dummy coded variables where 1 = being a member of
that ethnic category and 0 = non-membership in that category. In each model White/
White Hispanic was used as the reference category because it is the largest category.
This predictor variable was coded using the same ethnicity categories provided in FJSP
datasets, the source of information used in this analysis. As is common in many
datasets, Hispanic defendants in the FJSP research series are categorized as either
white or black. The lack of ethnicity and citizenship information in criminal justice
datasets is a significant handicap to researchers examining the interaction between
citizenship status and race/ethnicity.

Citizenship status is broken into three dummy coded variables: (1) U.S. citizen, (2)
Non-citizen, and (3) Unknown status. For each citizenship status variable, 1 = being a
member of that citizenship category and 0 = non-membership in that category. For the
purpose of this study, U.S. citizens include individuals born in the U.S.; individuals
who were born outside the U.S., but who have at least one parent who is a U.S. citizen;
and individuals who were born alien but who have lawfully become citizens of the
United States (i.e., Bnaturalized citizens^). Non-citizens are individuals who are legal
(resident) aliens, unauthorized immigrants or individuals without U.S. citizenship

476 Am J Crim Just (2017) 42:469–488

whose immigration status is unknown. An unauthorized immigrant is a person who
resides in the United States but who is not a U.S. citizen, has not been admitted for
permanent residence, and is not in a set of specific authorized temporary statuses
permitting longer-term residence and work (see Passel et al., 2004).

In each model U.S. citizenship was used as the reference category because it is
the largest category. Descriptive analysis revealed that approximately 3178 cases
(9.1 %) of total cases were missing citizenship status. Rather than excluding those
cases it was decided to recode them as unknown in order to identify possible
patterns in the missing values and to determine if those patterns were consistent
with or different from other categories. One author of this study has 22 years of
personal experience as a federal criminal investigator and supervisory special
agent. Based on that experience, the following are possible explanations for
missing demographic information in the PTS database (source of information for
FJSP datasets used in this analysis). Reasons include the inaccurate completion of
investigative reports due to human error and a lack of quality review of those
reports by supervisory personnel. Additionally, the U.S. Department of Justice
Office of the Inspector General (2004) determined that the USMS (proprietor of
the PTS database) lacked proper internal controls to identify and correct erroneous
or missing information contained within the computer database.

Marital status is also broken into three dummy variables, (1) married, (2)
Unmarried (including divorced, single, and widowed individuals), and (3) un-
known marital status. Like other dummy coded variables in this study, 1 = being a
member of that marital category and 0 = non-membership in that category. In each
model Unmarried was used as the reference category because it is the largest
category. Descriptive analysis revealed that approximately 15,166 cases (43.5 %)
of total cases were missing marital status. Rather than excluding those cases it was
decided to recode them as unknown in order to identify possible patterns in the
missing values and to determine if those patterns were consistent with or different
from other categories. Possible explanations for the missing values are discussed
above. Age, the final independent variable, was measured as age in years, at the
time that the individual was arrested. Prior to any analysis being run, these
variables were checked for collinearity, and none was found. VIF scores ranged
from 1.023 (Tolerance = .977) to 1.415 (Tolerance = .707).

Analytic Strategy

Logistic regression was chosen to analyze the preceding research questions, because it
is appropriate for use when the dependent variable or variables (in this case the
prevalence of arrest for each type of offense) are dichotomous, and it is appropriate
with all types of independent variables (Menard, 2010). Model statistical significance
will be determined using the p-value associated with the model chi-squared statistic,
and the model substantive significance will be determined using the likelihood ratio R2,
also known as McFadden R2 or R2L. This measure of model substantive significance
was chosen because it is conceptually the closest to R2 in OLS regression, in that it
reflects the proportional reduction in the quantity actually being minimized (Menard,
2010). Predictor statistical significance will be determined using p-values computed

Am J Crim Just (2017) 42:469–488 477

from the Wald statistic, while predictor substantive significance will be determined
using fully standardized regression coefficients. These coefficients have the same
interpretation in logistic regression as in ordinary least squares regression: A one
standard deviation change in the predictor is associated with a b* standard deviation
change in the outcome, where b* is the fully standardized regression coefficient. For a
full description of how these coefficients are calculated, see Menard, 2010. While it is
the fully standardized regression coefficients that will be interpreted by the authors,
odds ratios have also been inserted into the tables, for readers who prefer that measure
of predictor substantive significance. Table 1 illustrates the descriptive statistics for the
variables in the current study.

Results

Table 2 illustrates the results of the logistic regression arrest models, for each offense
type. On this table we can see that the federal homicide arrest model is statistically
(p = .000) and substantively (R2L=.276) significant. Accordingly, the model explains
27.6 % of the variation between federal homicide arrests and federal arrests for other
offenses. According to Wald statistics, individuals who are disproportionately arrested
for a federal homicide offense are more likely to be male (p = .002), White/White
Hispanic versus Black or Black/Hispanic (p = .016), a member of an Bother^ racial

Table 1 Descriptive statistics

N Mean Standard deviation

Dependent variables

Homicide Arrests 34829 .0088 .09362

Assault Arrests 34829 .0313 .17412

Robbery Arrests 34829 .0146 .11989

Marijuana Arrests 34829 .5671 .49548

Hard Drug Arrests 34829 .2804 .44919

Weapons Arrests 34829 .0978 .29704

Independent variables

Age 34099 31.36 10.412

Sex 34828 .8410 .36566

White/White Hispanic 34828 .8984 .30207

Black/ …