Patterns of Family Visitation During Immigration Detention
The population detained by Immigration and Customs Enforcement more than doubled between 2001 and 2013, swelling to over 477,000 individuals. Despite this growth, few studies analyze the experiences of detained immigrants. We draw from one of the first studies of detention in the United States, analyzing survey data from 565 noncitizens detained for six months or longer in California. Criminal incarceration literature finds that family visitation helps maintain social ties but is not evenly distributed. We analyze the predictors of contact and visitation with children during immigration detention. Results indicate that demographic background, the type of detention facility, and children's legal status substantially affect contact and visitation experiences. Findings suggest that immigration detention replicates experiences of criminal incarceration and is perpetuating inequality in immigrant communities.
immigration detention, visitation, immigration, incarceration, undocumented immigration
A large body of social science research analyzes the causes and consequences of the rise of mass incarceration, yet very few studies document the growth and consequences of a parallel system: mass immigration detention.1 Indeed, the last three decades have brought about an unprecedented convergence in immigration and criminal laws, leading to an influx of noncitizens into the federal criminal justice system as well as an explosion in the United States' detention and deportation systems. For instance, around half of the individuals sentenced in federal courts are noncitizens (Light, Massoglia, and King 2014) and 9.1 percent of federal prisoners (approximately seventeen thousand inmates) are incarcerated for immigration-related offenses as of October 2015 (Federal Bureau of Prisons 2015). Noncitizens in general and undocumented immigrants in particular are more likely to be incarcerated pre-trial and receive longer criminal sentences than U.S. citizens (Light, Massoglia, and King 2014).
At the same time, the growth of the U.S. immigration detention system has skyrocketed. In 2013, 477,000 immigrants were detained by [End Page 18] Immigration and Customs Enforcement (ICE), representing a nearly 2.5-fold increase in the detainee population since 2001, at a cost of more than $2 billion—or $161 per detainee per day (Simanski 2014; National Immigration Forum 2014). Despite recent efforts to reduce the populations of jails and prisons, immigration detention remains a growth area. Since 2009, Congress has mandated that ICE continue to fill thirty-four thousand detention beds daily, thereby allocating more funding for detention than was requested by the White House in 2014 (National Immigration Forum 2014). Though the scope and cost of immigration detention in the United States has rapidly expanded in recent decades, largely due to issues of access, social science literature on this phenomenon remains scant.
U.S. immigration law is civil rather than criminal and is therefore legally considered nonpunitive. Because few constitutional limits are set on the length of detention, however, noncitizens who commit triggering offenses under U.S. immigration law are often held mandatorily for the entirety of their removal proceedings, in local jails or in facilities run by private prison corporations. In 2013, ICE detained more than thirty thousand immigrants for three months or longer and ten thousand for six months or longer (Transactional Records Access Clearinghouse 2013). A recent study found that more than 15 percent of Mexicans deported from the interior of the United States had been held in ICE custody for more than one year prior to deportation—and half of those were held for more than three years (Bermudez, n.d.). These figures suggest that immigration detention, though legally nonpunitive, has become much more akin to incarceration than ever before. The literature on the experiences of the incarcerated thus provides an important jumping-off point for the present study.
We seek to understand the impact of long-term detention on detained immigrants and their families.2 One component of this inquiry is how immigrant families experience the long-term detention of a parent, particularly through access to regular communication and visitation. We therefore seek to answer the following research questions: What factors influence whether detained parents have any contact at all (such as letters, phone calls, or visits) with their children? What factors influence whether detained parents have face-to-face visitation with their children? Does the legal status of a detained parent's spouse or child predict visitation?
These questions are particularly important for at least three reasons. First, research on incarceration in the criminal context finds that family visitation is an important predictor of recidivism and can help families stay connected during the separation period (Bales and Mears 2008). However, although existing research has examined variation across inmate visitation in jails and prisons (see, for example, Cochran, Mears, and Bales 2014), no studies have considered the experiences of those held under the jurisdiction of immigration authorities.
Second, the existing assumptions about the importance and value of visitation are based on accessibility of visitation—from both a logistical and a legal standpoint. The extent to which visitation is differentially accessible for immigrant families from certain backgrounds, or when visiting certain facilities, could indicate inequalities within the system of immigration detention. We therefore examine, for the first time, whether experiences of visitation during immigration detention vary depending on the characteristics of detainees, their family members, or the facilities in which they are housed. Third, and relatedly, we pay particular attention to whether spouse and children's immigration statuses predict visitation patterns. If these family members are unable or less able to visit, we might conclude that immigration detention is especially punitive for families containing undocumented family members.
The paper draws empirically from one of the first studies of immigration detention and release in the United States. We analyze data collected in 2013 and 2014 from 565 immigrants who had been detained for six months or longer [End Page 19] in California. The four facilities in which study detainees were held—three jails and one privately operated facility, each subcontracted by ICE to house immigrant detainees—represent the universe of facilities housing long-term detainees in the federal judicial district in California where the study took place. Findings suggest that detainees' ethnicity and sex predict the likelihood of any contact with children (such as phone calls or letters) as well as whether detainees received any face-to-face visits from children. In addition, both the type of facility where detained immigrants are held and the undocumented legal status of detainees' children substantially affect contact and visitation experiences with children. Detainees held in private immigration detention facilities rather than county or city jails experienced lower likelihoods of receiving any in-person visitation with their children as well as fewer total visits. Finally, having undocumented children exhibited mixed effects in our analyses. Although having undocumented children marginally increased the likelihood of receiving any contact from children (such as letters or phone calls), detained parents with undocumented children received comparatively fewer visits, on average, from their children.
This study contributes to several bodies of literature. First, although the magnitude and costs of immigration detention have continued to increase, empirical studies of detention and its impacts on individuals, households, and communities are scarce. Advocacy organizations have released reports documenting punitive conditions within detention facilities as well as difficulties with reentry following release (Chaudry et al. 2010; Amnesty International 2009). However, given the challenges to accessing the detained population, studies of the experiences of immigration detainees are few and far between.
Second, this study provides an important comparative context to research on criminal incarceration and reentry. In contrast to the lack of empirical studies on detention, a broad history of research on incarceration helps situate the current study. Social scientists have explored the impacts of visitation on family relationships (Poehlmann 2005) and the extent to which visitation may be stratified across certain groups of prisoners (Cochran, Mears, and Bales 2014). This body of research has greatly enhanced our understanding about the complex ways in which mass incarceration is perpetuating cycles of poverty and marginalization in communities experiencing high rates of incarceration. By applying the insights from research on incarceration and reentry, this study fills significant gaps in our understanding of the social and economic consequences of prolonged immigrant detention on families. As the criminologist Joshua Cochran and his colleagues point out, an inquiry into visitation is critical because if visitation is differentially accessible, it can indicate inequality in the collateral consequences of imprisonment (2014). Finally, this study allows us to explore whether family members' immigration statuses compound with detention policies to enforce and reinforce family separation.
This study is also relevant to current political and social debates. The first of its kind, it sheds light on some of the experiences and consequences of immigration detention for mixed-immigration-status families at a time when detention is becoming an increasingly common reality for many families and is a source of national discussion. Second, it illustrates some of the implications of the increased criminalization of immigrant communities and the extent to which enforcement measures may be experienced across these communities, which are made up of citizens and noncitizens alike. Relatedly, it is sociologically important for its implications in understanding immigrant integration and immigrant identity formation. Many detained immigrants go on to win their cases and need to reintegrate into their communities afterward. Maintaining family cohesion during the lengthy detention process may be critical to this transition.
The expansion of mass immigration detention
The imprisonment of noncitizens for violating U.S. immigration law is not a new phenomenon, though the recent and prodigious expansion in the size and scope of the detention system is unparalleled. Scholars have documented such an unprecedented convergence of immigration and criminal law in recent decades that [End Page 20] they have begun to refer to the phenomenon as crimmigration (Stumpf 2006). Changes to immigration and criminal laws since the 1980s—especially during the War on Drugs—have vastly inflated the immigration detention system by, for example, drastically lowering the bar for deportable offenses and expanding the categories of individuals who can be held mandatorily pending a judicial decision on their removal proceedings (Stumpf 2006; Sayed 2011; García Hernández 2014; Coutin 2011). The legal scholar César Cuauhtémoc García Hernández argues that Congress has used immigration detention "as a central tool in the nation's burgeoning war on drugs" (2014, 1349).
The vast expansion of deportable offenses and detention authority has led to a boom in immigration detention, with the detained population expanding from just over 200,000 in 2001 to nearly 480,000 in 2013 (Simanski 2014), at a cost of around $161 per detainee per day (National Immigration Forum 2014). This incredible growth has also led to significant delays in the adjudication process in removal proceedings, resulting in longer stays within detention facilities. As mentioned, in 2013, more than thirty thousand individuals were held for three months or longer and more than ten thousand for six months or longer.
U.S. immigration law is civil rather than criminal in nature; therefore, the Supreme Court has held time and time again that immigration detention is legally considered non-punitive (García Hernández 2014). Given this definition, detained immigrants are not subject to the same constitutional protections available in a criminal context—for example, the Sixth Amendment right-to-counsel provisions or other due process guarantees under the Fifth Amendment (Kaufman 2008; García Hernández 2014). Such constitutional concerns, combined with the vast growth of crim-migration, have led legal scholars to argue that immigration detention has become so akin to criminal incarceration that it should be considered punitive:
Individuals in immigration confinement are frequently perceived to be no different than individuals in penal confinement. … They are represented as a threat to public safety, locked behind barbed wire, often in remote facilities, and subjected to the detailed control emblematic of all secure environments. Often they are held alongside their criminal counterparts. … By so intertwining immigration detention and penal incarceration, Congress created an immigration detention legal architecture that, in contrast with the prevailing legal characterization, is formally punitive.
Scholars and advocates have also raised concerns about the expansion and scope of mandatory immigration detention. Noncitizens who are subject to mandatory detention are required to be held by ICE for the entire length of their removal proceedings, which can last months or even years in some cases (Transactional Records Access Clearinghouse 2013; Bermudez, n.d.). The Immigration and Nationality Act specifies that individuals subject to mandatory detention include, for example, those convicted of most felony offenses and multiple misdemeanors, including minor drug offenses and some traffic offenses.3 To be clear, mandatory detainees are not being held on criminal charges, but are instead imprisoned while awaiting adjudication in their removal proceedings. The average length of detention for respondents in our sample at the time of the survey was 271 days (approximately nine months).
In summary, changes to immigration and criminal law over the past several decades have led to a prodigious expansion of immigration detention alongside the growth of mass incarceration. The size of the detained population has ballooned in unprecedented fashion, with few constitutional limits on the length of detention. Although detention is legally considered nonpunitive, detained noncitizens are often perceived as criminals, held in the same jails as criminal offenders, and for similar periods as many criminal inmates in local jails. Nearly half a million individuals are held in immigration detention facilities each year, yet few studies have been able to investigate the [End Page 21] experiences of immigrant detainees. We know relatively little about how detainees and their families experience immigration imprisonment, and with what consequences. Given the similarities between the systems of mass incarceration and mass detention, it makes theoretical as well as empirical sense to draw from literature on incarceration as a helpful starting point in understanding the experiences of immigrants detained for long periods of time. This paper proceeds with an inquiry into one important aspect of the experience of imprisonment: contact and visitation with family.
family contact and visitation during imprisonment
For incarcerated parents, imprisonment represents a removal from the family unit. Even for those who strive to remain close with their families during incarceration, contact does not come easily. Phone calls are expensive and limited, letters are read by guards before delivery, and visitation is both limited to certain hours and often logistically difficult for families. In some facilities, no contact visits are allowed, so all in-person communication must be done through glass shields or—more recently—video conferencing. Despite these challenges, studies confirm the importance of visitation during incarceration. Indeed, visitation is one of the only opportunities for communication with the outside world and for maintaining social ties. Contact with family has been found to reduce stress among inmates, in turn reducing negative behaviors within prison settings (Cochran 2012). Visitation also allows imprisoned individuals to continue to access social resources and social capital that can support positive reentry and reduce recidivism (Bales and Mears 2008). Increased visitation has also been linked to health outcomes among the incarcerated, such as by reducing depressive symptoms among mothers (Poehlmann 2005).
Visitation can be a positive experience for incarcerated individuals but it is not equally distributed. In one of the only studies of the predictors of inmate visitation, Joshua Cochran, Daniel Mears, and William Bales find that African Americans, the elderly, and those more frequently incarcerated receive fewer visits (2014). Such results suggest troubling barriers for certain inmates to the benefits of visitation. However, to our knowledge, no studies examine whether the family members' legal status has any impact on visitation.
The Bureau of Justice Statistics reports that 60 percent of mothers and fathers in jails and prisons never receive a visit from their children (Glaze and Maruschak 2008). Understanding the factors that predict such disparities in visitation is important for several reasons. First, visitation rules, however stringent, are generally consistent across imprisoned individuals within the same facilities, with some exceptions. That visitation is more difficult and less frequent for certain groups may constitute an added form of punishment for imprisoned individuals from those groups (Cochran, Mears, and Bales 2014). Second, differential access to visitation may represent a group-specific collateral consequence for family members of the imprisoned (Travis 2005). Finally, as Cochran and his colleagues argue, "to the extent that such punishment [differential visitation] is patterned along social and demographic lines, it raises questions about the social inequality in punishment in America" (2014, 5).
Even if visitation rules were consistent for all inmates, the accessibility of visitation between different types of facilities varies. A recent study analyzed the types of contact allowed at jails, state prisons, and federal prisons and found discrepancies in inmates' access to different types of contact (letter writing and phone calls) and face-to-face visitation by facility type (Shlafer, Loper, and Schillmoeller 2015). For example, local jails often allow only noncontact visits, but state prisons almost always allow contact visits. The authors conclude that the type of facility has important implications for families and children: "The type of facility can impact the inmates' proximity to their families, the probable frequency of contact, as well as the format and rules for contact and visitation" (Shlafer, Loper, and Schillmoeller 2015, 2).
Given such discrepancies, we might expect to see differences, especially in access to face-to-face visitation, across the facilities in which immigrant detainees are held. However, although many immigrant detainees are held in local and county jails alongside criminal offenders, [End Page 22] approximately 62 percent of detention beds are subcontracted to private, for-profit prison corporations (Carson and Diaz 2015). Although similar to government-operated facilities in their custodial role, research suggests important differences between private facilities and local or county jails. For example, Alissa Ackerman and Rich Furman provide a review of literature on private prisons, arguing that the reduction of operating costs, a key selling point for private prison contracting, often "comes at the expense of quality" (2013, 258). In particular, the authors note that the conditions in private facilities, such as personnel training and inmate health care, are often of lower quality than their government-operated counterparts. Moreover, it is unclear whether private facilities offer visitation privileges similar to those offered in public facilities. To our knowledge, no research has assessed families' access to these privately operated facilities. However, recent media coverage of U.S. immigration detention facilities housing Central American children and families has demonstrated extensive barriers to accessing such facilities. For example, a related New York Times Magazine feature reported some of the logistical challenges attorneys faced in accessing privately operated facilities in Arizona and Texas (see Hylton 2015). Moreover, private facilities are often built far outside metropolitan areas that offer few options for public transportation, which can create additional barriers for potential visitors.
Even when imprisoned individuals can receive visits, many factors could influence the frequency of visits. First, visitation (and contact in general) can be logistically challenging and expensive. In a study of women who visited family members in a large state prison, the clinical psychologist Olga Grinstead and her colleagues documented emotional, social, and financial challenges (2001). Women reported that the economic burden of having an incarcerated family member was worsened by the financial costs of visitation, phone calls, and sending packages. Women reported spending an average of $292 per month on these items, representing more than a quarter of monthly income for those in the lowest economic brackets in the study. To make matters worse, jails and prisons are often located far outside city centers, requiring long drives or bus rides that often take up entire days (Christian 2005). Coupled with challenges of distance and transportation, visitation schedules within facilities may make visitation difficult for working spouses and children.
An additional challenge to family visitation is the negative psychological experience of visitation for family members. In a study of women with partners incarcerated at San Quentin State Prison in California, the sociologist Megan Comfort describes the "secondary prisonization" experienced by the families of the detained (2003). Legally innocent people come to experience the effects of incarceration indirectly due to their sustained contact with the correctional institution, characterized, for example, by dealing with guards and being subject to invasive searches (Comfort 2003, 2007, 2009). Comfort also demonstrates that partners and spouses feel emotionally strained with worry about their incarcerated partner (2009). Children also experience secondary prisonization experiences. In response, some families may decide that it is better for the child not to visit the incarcerated parent (Martin 2001; Poehlmann 2005). Other parents may decide that they don't want children to visit, either because of the nature of the incarcerated parent's criminal record or the deterioration of parents' relationship (Edin, Nelson, and Paranal 2004). In noncitizen families, additional barriers may include language, fear of authorities, and immigration status, an issue we address in the following section.
immigrant families and visitation
Despite the vast similarities between criminal incarceration and immigration detention, it is plausible that immigrant families experience immigration imprisonment differently than families of U.S. citizens do criminal incarceration. In particular, the precarious legal status of family members may impose a barrier to regular visitation.
An emerging body of literature on the impacts of unauthorized status has emerged over the past decade. It has exposed how immigration laws can structure the everyday experiences of undocumented immigrants and their [End Page 23] families, with great consequences for their economic, social, and health outcomes (Menjívar and Abrego 2012; Dreby 2012; Coutin 2011; Dreby 2015). Scholars are increasingly concerned with how laws and policies produce and sustain the inequality of individuals, families, and communities (Ngai 2004; De Genova 2002; Coutin 2000; Willen 2007). The geographer Nicholas De Genova argues that the threat of deportability engenders persistent feelings of the "revocability of the promise of the future" among undocumented immigrants (2002, 427). The anthropologists Guillermina Gina Nuñez and Josiah Heyman find that increased immigration enforcement affects undocumented immigrants to such an extent that some will structure their daily decisions to avoid discovery, for example, by selectively choosing which streets to drive on (2007). The fear of exposure may deter undocumented immigrant family members from visiting detained loved ones. This is similar to the concept of system avoidance (Brayne 2014), whereby individuals who have had criminal justice contact steer clear of "surveilling institutions" that keep formal records. For undocumented individuals, whose statuses already make them vulnerable, an immigration detention facility may be the most directly threatening form of surveilling institutions. We therefore expect that undocumented families may view visitation as effectively entering the belly of the beast and may feel they have no choice but to avoid it.
An additional challenge to undocumented family members' visitation is logistical. The anthropologist Susan Coutin demonstrates how illegality turns the mundane into the illicit—for example, undocumented immigrants are faced with the daily predicament of unlicensed driving and lack of official government documents (2000). Many undocumented immigrants are barred from getting valid drivers' licenses, which may be a particular impediment in the context of family visitation for two reasons. First, at least one facility in this study requires visitors to show a driver's license and proof of insurance to enter the facility grounds by vehicle. Second, most facilities require visitors to provide photo-identification before visiting family members. These logistical challenges could present barriers to visitation for many low-income families, but are especially problematic for undocumented families. In particular, we might expect undocumented children (or those with undocumented parents who are not imprisoned) not to visit, or to visit with less frequency than documented children.
Even in families that can overcome the logistical and legal challenges discussed, visitation might still be a traumatic event, for reasons that may be particular to immigrant families. To be sure, the ever-present fear of enforcement makes its way through entire families regardless of citizenship status. The sociologist Joana Dreby interviewed the young children of immigrants and finds that they equate police with ICE and exhibit observable fear of family separation and deportation even when no one they know is detained (2012, 2015). Dreby also finds that the children's mothers worried so much about detention and deportation of their spouses that they developed severe symptoms of stress and other mental health concerns (2015). It is possible, therefore, that the children of detained immigrants will visit their detained parents less frequently because the experience could trigger not just the negative emotions associated with having an incarcerated parent, but also increased worry about the potential for parents' deportation and prolonged family separation. It seems likely that undocumented immigration status will only compound the disadvantages that children with detained parents face.
data and methods
This study draws from original survey data collected in 2013 and 2014 from 565 detainees in California. The list of participants was drawn from the universe of individuals who had been held for at least six months in the federal judicial district in California where the study took place, and had been scheduled a hearing under Rodriguez v. Robbins, ongoing class action litigation requiring custody redetermination hearings (bond hearings) before an immigration judge for individuals who have been detained for 180 days or longer. We focus on detention lasting six months or longer for three reasons. First, in Rodriguez v. Robbins, the Ninth Circuit Court of Appeals held that a detention of this length is a profound deprivation of liberty and [End Page 24] raises serious constitutional concerns. The court's decision rested on a body of law recognizing that detentions longer than six months require heightened protections, suggesting that long-term detention is particularly punitive. Second, a large portion of jail and prison inmates in the state of California serve less than one year, making this population an important comparison group. Third, in other Western countries, reentry outcomes such as employment are significantly affected by time served lasting six months or longer (see Ra-makers et al. 2014). Given these reasons, it makes empirical sense to consider that by the six-month mark much of immigrant detainees' experiences of imprisonment—and the implications of those experiences—will have come to mirror those of the criminally incarcerated.
The 565 detainees in our sample were held at four detention facilities in California subcontracted by ICE to house immigrant detainees. Three facilities are county or city jails; a private, for-profit correctional corporation operates the fourth. Surveys were conducted in person, in English or Spanish, and all participants were at least eighteen years old. Surveys lasted between 90 to 120 minutes and respondents did not receive any incentive for participating. Of detainees who received information about the survey from the interviewers, 92 percent completed the survey. There were no significant differences in refusal rates by gender or country of origin.
To our knowledge, this study is the first to access such a large sample of long-term detainees during their detention. As explained, the sample was drawn from a list of all respondents who had been held for six months or longer and had been scheduled a bond hearing under Rodriguez v. Robbins. Because the Department of Homeland Security (DHS) does not release detailed demographic or case-specific information about immigrant detainees, it is not possible to determine how well this sample represents the detained population in the United States. However, it is important to note that the sample may have different characteristics than other detainees held throughout the United States. First, the average respondent in the sample had been detained for 271 days at the time of the survey, which may differentiate them from shorter-term detainees.4 For example, long-term detainees (many of whom are held mandatorily) may be more likely to have some kind of criminal record that triggered a removal proceeding under immigration law. In our sample, 96 percent of respondents had at least one misdemeanor or felony conviction.5 The two most common convictions in our sample are traffic and drug-related (both 44 percent). Unfortunately, we are unable to determine whether any particular conviction may have triggered deportation proceedings. Finally, on average, respondents in our sample had lived in the United States for nearly two decades, which may differentiate them from detainees held in other locations—for example, in facilities along the U.S.-Mexico border. Though this sample may not represent the universe of detained individuals in the United States, it is the first of its kind to allow individual-level analysis of the experiences of individuals detained for long periods.
Measures and Method
The survey gathers data on the demographic, family, employment, health, immigration, and criminal background of each detainee. It also explores detainees' experiences in detention, including information about family visitation. We draw on visitation literature from the criminal context, as well as literature on the broader impacts of immigration law enforcement, to construct a series of measures to capture the [End Page 25] contact and visitation experiences of detained immigrants and the factors predicting these experiences.
We created three dependent variables to measure detainees' contact and visitation with their children. We excluded 107 respondents from the sample who reported having no children, bringing our sample size to 462. First, we generated a binary variable indicating whether respondents received any contact from their children while detained (visits, letters, phone calls, or news from others). Detainees reporting any of these forms of contact during detention were coded as 1; no contact was coded as 0.
A second binary variable measured whether respondents experienced any face-to-face visitation with their children during detention. The visitation variable provided a more specific examination than the overall contact variable because in-person visitation may carry different meaning than other forms of contact. Moreover, direct visitation experiences have a greater impact for both detainees and their children than other types of contact (Shlafer, Loper, and Schillmoeller 2015). Detainees who reported at least one visit with their children during detention were coded as 1; those who reported no visitation were coded as 0.
We generated a separate count variable capturing the extent of visitation experiences among the sample by measuring the total number of visits detainees received from their children. Respondents were asked whether they received visitation from each of their children and subsequently asked the number of times that they were visited by each child. Aggregating the number of reported visits from each child produced a single count measure measuring respondents' total exposure to their children during detention.6 To account for positive skewness in the variable distribution, and for consistency with prior research on visitation (Cochran, Mears, and Bales 2014), we recoded the count variable to a maximum of forty visits.7
We constructed a series of demographic, detention facility, family, and legal status variables to predict detainees' contact and visitation experiences. Demographic measures included respondents' gender (male = 1, female = 0), self-identified ethnicity (1 = Hispanic-Latino, 0 = not Hispanic-Latino), and age at the time of the survey. To capture detainees' previous socioeconomic status, we included a measure of detainees' self-reported average income through the six months prior to detention by computing average weekly earnings in hundreds of dollars.
Existing literature on criminal incarceration shows that inmates' criminal records significantly predict visitation and that inmates with more frequent experiences of incarceration are less likely to receive visits (Cochran, Mears, and Bales 2014). In addition, inmates with certain types of violent or serious crimes may have less or no contact with children given that family members may deem visitation unsafe or unhealthy for children (Shlafer, Loper, and Schillmoeller 2015). Finally, inmates who have been incarcerated for long periods may have fewer or weakened ties to their families and may therefore receive fewer visits (Cochran, Mears, and Bales 2014; Christian 2005). In our models, three variables measured respondents' self-reported criminal history. First, a count measure captured the total number of prior criminal convictions for each respondent. Next, two binary measures indicated any felony and violent convictions, respectively. We also controlled for the total months detained for each [End Page 26] respondent at the time of interviewing to account for potential spuriousness in our models, because longer detention periods may correspond with more visits. Time spent in detention varied from just under six months to over four and a half years at the time of surveying. To account for positive skewness in detention length, we computed a new variable by taking the natural log of total months detained.
The locations where individuals are detained can have important implications for contact and visitation because detention facilities feature their own rules and regulations surrounding visitation with family and are located in different geographic areas (Shlafer, Loper, and Schillmoeller 2015). In light of these potential influences, we created two variables to account for immigrant detainees' locations. First, we generated a dummy variable indicating the type of facility housing each respondent. Detainees in a privately operated detention facility were coded as 1; those in a county or city jail were coded as 0. We also computed the spatial distance between detainees' current detention facility and the city where each individual reported entering ICE custody prior to detention.8 This distance measure served two purposes in our analyses. First, greater distances between detention facilities and detainees' families may increase the difficulty of visitation, particularly for lower-income families, and therefore are a salient predictor of overall visitation experiences (Shlafer, Loper, and Schillmoeller 2015; Christian 2005).9 Second, this measure allowed us to examine the unique effects of privately operated detention facilities while controlling for facilities' geographic locations. To account for positive skewness in the distance measure, we calculated the natural log of computed distances.
Last, we created four variables to account for respondents' family structures and family members' legal statuses. We created a binary variable for whether respondents reported being married or having a domestic partner. Similarly, a binary measure indicated whether detainees had any dependent children. We defined dependent children as those age seventeen or younger; related biologically, through adoption, or as a stepchild; and currently residing in the United States. Additionally, two binary variables indicated whether the respondent reported having an undocumented spouse-partner or child, respectively.
We examined detained immigrants' contact and visitation experiences with their children across three dependent variables, using logistic regression (any contact, any visits) and negative binomial regression (number of visits) models.10 Prior to model estimation, we investigated missing data patterns within the dataset. Nearly all variables featured 1 percent or fewer missing observations except for marital status and spouse legal status, each missing [End Page 27] approximately 43 to 45 percent of observations.11 We preserved cases with missing data by generating multiply imputed datasets (Rubin 1987) using chained equations (the multivariate imputation by chained equations, or MICE, method). MICE allows users to specify each variable by type (such as count, dichotomous) for all variables and relaxes the assumption of normal distribution. We estimated fifty multiply imputed datasets to account for the high percentage of missing data on the spouse variables (Graham, Olchowski, and Gilreath 2007).12 After multiple imputation, we ran diagnostic tests to determine the proper modeling strategy for our count-based visitation measure and verified the test results across all fifty imputed datasets. Initial tests demonstrated significant overdispersion in the visitation count distribution, which indicated the appropriateness of negative binomial models rather than a Poisson regression approach.
Table 1 provides a descriptive overview of the study respondents. Respondents were typically male (92 percent), and were approximately thirty-eight years old on average. The majority (85 percent) reported Latino-Hispanic ethno-racial identification. The most common countries of origin were Mexico (50 percent), El Salvador (21 percent), Guatemala (12 percent), and Honduras (4 percent). Altogether, respondents indicated forty-three originating nations. Just over half (52 percent) reported being married, and 29 percent indicated having an undocumented spouse. Moreover, 72 percent of respondents indicated having at least one dependent child (biological, step, or adopted child, living in the United States, and under eighteen years old) and 8 percent reported having at least one child with undocumented legal status. In the six months prior to detention, these individuals earned on average about $750 per week. Respondents had an average of about three prior criminal convictions, with approximately 26 percent reporting at least one felony conviction and 40 percent reporting at least one violent conviction. At the time of interviewing, respondents had been detained around nine months and 44 percent were housed in a privately operated detention facility. The average distance between the city of arrest and current detention facility was sixty-two miles. During detention, 83 percent of respondents reported receiving at least some form of contact with their children; more than half (53 percent) reported receiving at least one in-person visit. Moreover, detainees reported an average of approximately eight visits while held in detention, but nearly half (47 percent) indicated no visitation.
We analyzed each of the three dependent variables for contact and visitation experiences by estimating models in a two-step approach.13 First, we regressed the dependent variable on all independent variables except for the legal [End Page 28] statuses of spouses and children. Next, we added the legal status measures into the model to identify their unique effects on contact and visitation after controlling for other model covariates.
Table 2 presents odds ratios from logistic regression estimates of whether, during detention, detainees received any form of contact from their children. An odds ratio higher than 1 indicates an increase in the odds associated with a one-unit increase in a given independent variable. A ratio between 0 and 1 indicates a corresponding decrease in the odds associated with a one-unit increase in a given independent variable. Without including legal status measures, the model indicated that Latino detainees had an approximately 114 percent increase in the odds of receiving contact over their non-Latino counterparts (p < .05), and having any dependent children increased the odds by nearly 700 percent (p < .001), controlling [End Page 29] for other variables in the model. Although marginally significant, male detainees also had greater odds of contact with their children than female detainees.
After adding legal status measures for spouses and children, the observed effects for Latino detainees became marginally significant and the effects of having any dependent children became stronger, increasing the odds of receiving any contact by approximately 824 percent (p < .001). Moreover, the legal status of detainees' children demonstrated a marginally significant relationship in the model, where detained parents with any undocumented children had higher odds of receiving any contact from their children.
Next, we estimated logistic regression models predicting the likelihood of respondents receiving any direct visitation with their children (table 3). These analyses yielded results similar to those of the general contact models but also some divergent findings. Before accounting for the legal statuses of spouses and children, the model indicated that male detainees had an approximate 162 percent increase in their odds of visitation relative to female detainees (p < .05) and Latino detainees had approximately 84 percent higher odds than non-Latinos (p < .05), controlling for other variables. Similar to previous findings, detained parents with dependent children showed an approximate 443 percent increase in their odds of visitation (p < .001). Unlike in the previous models, however, the facility where respondents were detained showed a significant relationship with visitation. Compared with detainees in city-and county-operated facilities, individuals held in a private detention facility experienced a nearly 60 percent decrease in the odds of any child visitation (p < .001). The model also indicated a marginally significant and negative effect for respondents convicted of violent crimes in the past relative to detainees without a violent criminal history.
Adding legal status measures to the binary visitation analyses produced almost no change in covariates' relationships with the predicted odds of visitation. The odds of visitation with their children increased for both male and Latino detainees. Moreover, having dependent [End Page 30] children significantly increased the odds of visitation but being detained in a private facility decreased them. Neither spouses' nor children's undocumented legal statuses significantly predicted the likelihood of visitation.
The final set of analyses, presented in table 4, examined the relationships between model covariates and the total number of visits that detained individuals had with their children. Before accounting for spouses' and children's legal statuses, model results indicated that respondents' age was significantly associated with visitation, where each year of age corresponded with an approximately 3 percent increase in the expected number of visits (p < .05), controlling for other variables. In addition, having a dependent child increased the estimated number of visits by approximately 208 percent (p < .001). The type of facility housing respondents was significantly related to the number of visits with children: respondents in a private facility saw an approximately 59 percent decrease in the expected number of visits while in detention (p < .01), after controlling for other variables.
Adding legal status measures to the analysis changed several of the model's estimated effects. Detention in a private facility remained statistically significant but its coefficient grew in magnitude, suggesting a 64 percent decrease in the expected number of visits (p < .001). The legal status of detainees' children showed a significant and negative relationship with the number of received visits. Respondents with undocumented children showed a 64 percent decrease in the predicted number of visits (p <.05), controlling for other variables. Thus, although legal status did not appear to influence the likelihood of having any visitation with children, having undocumented children did suggest a reduction in the overall number of visits for individuals held in immigration detention facilities. Respondents' age remained a salient predictor, with each year of age predicting an approximate 4 percent increase in the expected number of visits (p < .01), and having any dependent children increased the expected number of visits by approximately 241 percent (p < .001).
[End Page 31]
Although research on the causes and consequences of mass incarceration is extensive, few studies empirically examine the impacts of a parallel system: mass immigration detention. Legal scholars argue that immigration detention has become increasingly punitive, bringing up several constitutional concerns despite detention's nature as administrative law (García Hernández 2014; Kaufman 2008). This study aimed to examine, for the first time, patterns of family visitation among immigrant detainees imprisoned approximately six months or longer. Differences across family visitation experiences could suggest unequal collateral consequences from the detention system, which could perpetuate inequality and stratification in immigrant communities.
Several findings emerge from our analysis. Overall, being held in private detention facilities reduces the likelihood of face-to-face visitation with children but not general contact, whereas the demographic characteristics of the detainee (sex, ethnicity, and dependent children) increase the likelihood of contact and visitation. Legal status is also a salient predictor: undocumented children are more likely to be in contact with their detained parents but to make comparatively fewer visits to see them.
Controlling for other relevant characteristics, males and Latinos are more likely to have contact with children and to receive a visit from them. Scholars argue that deportation (and detention, as its frequent precursor) is a gendered and racialized process that disproportionately affects Latino males and will therefore have gendered and racialized effects in immigration communities (Golash-Boza and Hondagneu-Sotelo 2013). Our results support this hypothesis. Females may have differential access to the family and community support systems necessary to maintain child visitation during detention. Females may also be more stigmatized for their imprisonment because they are less likely to be policed in the first place (Rios 2011). Further research could explore the mechanisms explaining why Latinos and men are more likely to remain in contact with children during their detention.
A key finding from this study is that being detained in a privately operated facility (compared to a county or city jail) reduces the likelihood of receiving any face-to-face visits from children, as well as the total number of visitors. Although research on criminal incarceration documents discrepancies across local, state, and federal jails and prisons, to our knowledge, ours is the first on facilities run by private corporations (see Shlafer, Loper, and Schillmoeller 2015). Several factors may contribute to our finding that visitation is curtailed in privately operated facilities. Although the private facility in our study is much farther from major cities in the study region than other facilities, distance is not a significant predictor after controlling for the facility.
At first glance, this finding may seem to contradict prior research suggesting the impact of distance on the likelihood of visitation among criminally incarcerated individuals (see, for example, Tahamont 2013). When we reestimate our models and exclude the private facility variable, distance significantly predicts visitation. Yet this effect disappears when the private facility variable is included. Thus, the private immigration detention facility continued to exhibit a separate and distinct influence on the likelihood of visitation and the number of visits. An alternative explanation, therefore, could be that logistical access is more important than physical distance. For example, Google Maps estimates that it will take more than five hours on public transportation, each way, to make the approximately hundred-mile commute from a city where many families reside to the private facility (including four buses, one train, and a two-mile walk). Families who cannot access reliable transportation may be less likely to visit (Christian 2005). Second, the ways that private facilities operate relative to county or city jails may differ. Private facilities may have different rules and regulations for visitation, staff may be trained differently, and facilities may have a different ambiance and style of operations. For example, the private facility in this study allows visitation on only one weekend day, whereas the other facilities have access to visitation throughout the weekend. Additional studies that can address these qualitative differences across facilities would be timely and relevant. Indeed, privatized detention facilities are now the norm rather than the [End Page 32] exception: 62 percent of detention beds are subcontracted to private facilities, up from 49 percent in 2009, during which period the quota for detention beds increased by 46 percent (Carson and Diaz 2015).
The legal status of detained immigrants' children exhibits differential effects in our analyses, where undocumented children are more likely to be in contact with their detained parents but make comparatively fewer total face-to-face visits. These results suggest that although undocumented families make efforts to stay in touch during detention, they may face systematic barriers from visitation that might otherwise yield positive effects for both parents and children (Poehlmann 2005; Shlafer, Loper, and Schillmoeller 2015). As a result, undocumented children may be more strongly penalized by a parent's detention than their counterparts with legal status, suggesting unequal collateral consequences across families.
Our analysis contributes to several theoretical, empirical, and political debates. First, it provides one of the first empirical analyses of the experiences and consequences of immigration detention. Applying what we know about criminal incarceration to the immigration detention context enables us to examine the unique aspects of immigration detention and its impacts on detainees and their families. Our findings reinforce the notion that the immigration detention experience mirrors criminal incarceration in many ways, yet remains distinct in others.
Second, we know that family relationships are affected by imprisonment. Visitation provides opportunities for maintaining social ties that can support positive reentry and reduce recidivism (Bales and Mears 2008). Visitation can also reduce negative behavior, stress, and other detrimental health outcomes among the incarcerated (Cochran 2012; Poehlmann 2005). Therefore, "disparities in visitation constitute a form of potentially unequal punishment, a collateral consequence, concentrated more among some groups … than others" (Cochran, Mears, and Bales 2014, 2; see also Bales and Mears 2008; Western 2006). That detainees with different demographic backgrounds and across types of facilities experience detention differently suggests that they may also experience the collateral consequences of the detention experience differently, which could lead to further stratification and marginalization.
Third, differential access to visitation may cause an additional hardship to immigrant detainees that incarcerated individuals do not experience. As the criminologist Mary Bosworth has written of immigration detention in the United Kingdom, detention is a process defined by uncertainty (2014). For example, one way that detention is unlike incarceration is that detained individuals are not serving a sentence, simply awaiting adjudication on their removal proceedings. They have no way to gauge how long they may be held, must fight their legal cases while imprisoned, and do not enjoy the privilege of cost-free access to public defenders (Kaufman 2008). This becomes critical because adjustment of status cases can revolve around the participation of family members and others; for example, by providing affidavits of support or confirmation that they will house released detainees (Morando Lakhani 2013). To the extent that families are vital to the legal process, visitation may be a critical way to communicate about legal cases. Unequal access to visitation may indicate a troubling form of legal inequality that could have severe repercussions.
Fourth, this study provides additional evidence of how immigration enforcement actions are experienced across immigrant communities and of varying legal statuses. We observed that detainees with undocumented children receive fewer face-to-face visits. Although a body of research has documented the detrimental effects of "illegality" and of the threats of enforcement mechanisms, to our knowledge, this study is the first to hone in on the experiences of detained immigrants and their families while detention is ongoing (Menjívar and Abrego 2012; Dreby 2015; Núñez and Heyman 2007; De Genova 2002; Brabeck and Xu 2010). It is possible that decreased access to face-to-face visitation could lead to increased despair and reduce family cohesion in immigrant families (Dreby 2015).
Finally, this study is important for its implications in understanding immigrant integration and immigrant identity formation. Future [End Page 33] research could explore the ways that the children of detained immigrants, across legal statuses, experience and understand their parent's detention. As these children interact with the legal system, how do they come to understand their identities and their place in American society? In an era characterized by ever-expanding criteria for detention and deportation, these questions may be more important than ever before.
Caitlin Patler is assistant professor of sociology at the University of California, Davis.
Nicholas Branic is a PhD student in the Department of Criminology, Law and Society at the University of California, Irvine.
1. Throughout this paper, we refer to immigration detention and immigration imprisonment interchangeably.
2. We conceptualize long-term detention as lasting approximately six months or longer for several reasons discussed in the methods section.
3. Mandatory detention is specified in the Immigration and Nationality Act, §236 (8 U.S.C.A. §1226).
4. According to one estimate, the average immigrant detainee in the United States is released or deported in thirty-one days, often because he or she chooses not to contest a deportation order (Transactional Records Access Clearinghouse 2013). However, because DHS does not release individual-level information on detainees, it is impossible to compare how our sample aligns with this "average" detainee or even with an average long-term detainee who has been detained for six months or longer.
5. Compare this with data from September 22, 2012 (one day for which data are available on aggregate criminal histories via a Freedom of Information Act request) finding that 61 percent of detainees in ICE custody on that day had a criminal record (Kerwin, Alulema, and Tu 2015).
6. This aggregated count variable approximates the total number of visits that respondents experienced during detention. The detention facilities in the study stipulate limitations on the number of family members allowed to visit at one time (for example, two adults and one child, one adult and two children) and permit a maximum of one visit per day. Thus, even detainees with multiple children would only be able to see one or two children per scheduled visit.
7. Imposing a maximum threshold for visitation counts resulted in thirty-six cases recoded at forty visits.
8. Using ArcGIS version 10.2, we obtained the XY coordinates for city centroids and detention facilities' address-level coordinates. We then computed distance in miles between the coordinates using the economist Austin Nichols's vincenty program in Stata 14. City coordinates approximate detainees' predetention residence and are a proxy for where families are likely to reside. We acknowledge the imprecision of using city coordinates to draw these assumptions, which proves a limitation of the data and warrants consideration when interpreting results. Given the theoretical relevance of distance as a predictor of visitation, however, we elected to include the variable in our analyses.
9. Although families with personal vehicles may take advantage of freeways and comparatively lower travel times, families reliant on public transportation face greater hardships when planning to visit with a detained parent.
10. Using Stata 14, we tested for multicollinearity by estimating ordinary least squares regression models for each dependent variable while including all covariates and obtaining variance inflation factor scores. Results indicated that multicollinearity was not a concern in the models. We then tested for outliers using the Hadi statistic and identified twelve outlier cases that we subsequently omitted from the sample.
11. The original version of the survey did not include questions about respondents' marital status or the legal status of the partner, resulting in high frequencies of missing data. Given the high percentage of missing data on the spouse variables, we reestimated the multiple imputation and regression analyses after excluding the spouse measures; model results did not change significantly from models, including the spouse variables.
12. Because of the later addition of the number of visits questions in a revised version of the survey instrument, missing values on the number of visits with children were perfectly predicted by measures of any contact and any visitation. To prevent model convergence issues during multiple imputation, we estimated three sets of fifty imputations—one for each dependent variable. Comparing descriptive statistics between the three sets of imputations revealed essentially identical means and standard errors. The social psychologist John Graham and his colleagues show that greater proportions of missing data require greater numbers of imputations in order to reduce statistical power loss (2007). The authors recommend forty imputations for models with approximately 50 percent missing data; we estimated fifty imputations as a more conservative approach.
13. Scholars debate whether to include or omit imputed values for dependent variables. Because our imputation models include the presence of auxiliary variables that are not included in the regression models, we opted to retain the imputed Y values after multiple imputation. Comparing the model estimates from both approaches yielded essentially identical results.