Future, Interrupted: Examining the Impact of a Large Worksite Enforcement Operation on Students’ Educational and Workforce Pathways

Immigration Enforcement as Acts of Community Violence

Scholars have argued that anti-immigrant rhetoric, policies, and actions, including immigration enforcement, should be considered acts of community violence, and they have shown empirically how exposure to such forces can be detrimental to people’s physical health, mental health, development, and well-being (Barajas-Gonzalez et al. 2018, 2021; Kirksey 2024; Kirksey and Sattin-Bajaj 2023). We continue in this line of work by analyzing the consequences of a specific, identifiable act of community violence as evidenced by postsecondary educational indicators and workforce participation data measured at the county level among students most likely to be exposed to an LWEO and its aftermath.

In our conceptualization of legal vulnerability as a collective state shared among members living in the same community, vulnerability may result from multiple forces. First, there may be frequent, active, and extreme immigration enforcement activity that produces ongoing fear, stress, and instability (Watson 2014; Vargas and Pirog 2016; Amuedo-Dorantes et al. 2018; Hong et al. 2025, this issue). Relatedly, undocumented people may live in or near the community, which in turn increases the risk of enforcement activity and heightens the direct harm resulting from detentions and deportations via removals, including family separations, sudden financial instability, housing insecurity, and trauma (Chaudry et al. 2010; Lopez et al. 2022). Next, community members may experience the spillover effects of fear and volatility associated with undocumented status and the threat or enactment of immigration enforcement, including what has been called multigenerational punishment (Enriquez 2015). In this study, we use student data from school districts in Fannin, Grayson, Lamar, and Hunt counties precisely because the dramatic increase in the number of deportation orders filed in immigration courts in these counties in the month after the Load Trail raid reflects a high likelihood of legal vulnerability on these three grounds. In other words, we hypothesize that students attending school in these counties would be more likely than otherwise similar students elsewhere in the state to have been exposed to immigration enforcement activity and to know someone directly or indirectly affected by the August 2018 raid.

Even without enforcement activity, living in a restrictive, anti-immigrant environment could be a final source of legal vulnerability operating at a community level and producing tangible restrictions on people. It could affect the functioning of a community via policies, practices, or climate that limit a sense of freedom, safety, decision-making, and behaviors (Amuedo-Dorantes and Lopez 2015, 2017; Potochnick et al. 2017; Dee and Murphy 2020). In the present study, we focus exclusively on the first form of legal vulnerability—that is, exposure to immigration enforcement events or episodes—and isolate the particular effects on students’ educational choices, behaviors, and outcomes. In what follows, we review the research on how contact with immigration enforcement activity is associated with two main factors that can contribute to students’ post-secondary pathways and decision-making: students’ academic engagement, performance, and college enrollment, and students’ economic precarity and how it influences employment and educational decisions.

Exposure to Immigration Enforcement Activity

Exposure to immigration enforcement activity—be it via an arrest, detention, or deportation of oneself, a loved one, or a member of one’s local community—constitutes a form of legal vulnerability that has been shown to affect students’ academic performance, engagement, and educational decisions in a number of significant ways. Absenteeism is one of the primary indicators that scholars have tracked following enforcement episodes, finding that student absenteeism rates tend to increase in schools located in areas with more frequent and proximate ICE activity, especially immediately following such events (Kirksey 2024; Kirksey and Sattin-Bajaj 2023; Heinrich et al. 2023). Students’ English language arts (ELA) and math test scores also suffer when immigration enforcement actions occur, and school enrollment patterns for Latinx students can be sensitive to immigration enforcement (Kirksey 2023; Kirksey and Sattin-Bajaj 2023; Bellows 2019). Research has shown that student mobility increases concomitant with enforcement activity and declines in Head Start and school enrollment among Latinx students are associated with increased ICE arrests (Kirksey and Sattin-Bajaj 2023; Bucheli et al. 2021; Hong et al. 2025, this issue; Santillano et al. 2020). Less is known, however, about how students’ (and families’) behaviors and choices related to investing in higher education may be impacted by enforcement events, such as fulfilling college entrance requirements, applying for financial aid, and enrolling in four-year versus two-year colleges.

Immigration enforcement can also produce immediate, extreme, and long-term economic devastation for families. Researchers have documented the extent to which lost income due to detention and deportation of key earners produces dramatic changes in families’ financial stability (Capps et al. 2007; Chaudry et al. 2010; Dreby 2012; Lopez et al. 2022). Randolph Capps and colleagues (2016) estimated that families had lost anywhere from 40 to 90 percent of their income within six months of a parent’s immigration-related arrest, detention, or deportation. Another analysis using US Census Bureau data showed that household income falls up to 45 percent when a family member is deported (Preston 2020). Finally, William D. Lopez and colleagues (2022) and others found evidence of heightened food, housing, and financial insecurity in the immediate weeks and months following ICE raids, with lasting, irreversible consequences that persisted over time (Capps et al. 2007; Chaudry et al. 2010).

The mental and physical health effects associated with exposure to immigration enforcement episodes have been well documented, with large-scale events like an LWEO producing complex and sustained sequelae for individuals and broader communities. Research tracking the effects of a raid of a meat-processing plant in Postville, Iowa, in 2008 found that infants born to Latina mothers in the thirty-seven-week period after the raid had a nearly 25 percent higher risk of preterm low birth weights compared to the prior year (Novak et al. 2017). Another study showed higher self-reported stress and lower self-rated health scores for Latinx survey respondents living in a Michigan county where a warehouse raid took place (Lopez et al. 2017). Lastly, Carolyn Heinrich and colleagues (2023) found increased diagnoses of depression, substance use disorder, and higher incidence of self-harm, suicide attempts, and sexual abuse among children in Tennessee after a 2018 LWEO in Morristown.

Taken together, studies of the aftermath of immigration enforcement episodes paint a clear picture of severe and often lingering damage wrought by these incidents that strike at the heart of communities. This growing body of evidence elucidates the precise ways that enforcement penetrates school buildings and classrooms, via impacts on student attendance, health and well-being, and performance on &exams. Yet, large gaps remain in empirical understandings of how exposure to immigration enforcement activities shapes students’ educational trajectories—in particular, their pathways to and through college. Using robust, longitudinal data from Texas students, our analysis focuses on identifying the effects of an LWEO on key postsecondary outcomes, such as direct enrollment and degree completion, while also exploring potential mechanisms underlying these impacts. This work advances knowledge of the ways that young people experience and react to an act of community violence in the form of an LWEO via their postsecondary choices and actions. We draw attention to the enduring consequences that immigration enforcement can produce for students’ educational opportunities and outcomes—consequences that may have long-range implications for personal, professional, and financial success at the individual level and for community well-being overall.

Constructing a Comparable Control Group

After identifying the sample of students from the affected counties, we sought to construct a control group that would provide a comparison to measure the potential impact of the LWEO on student outcomes. This control group needed to represent what the affected counties might have looked like in the absence of the LWEO, accounting for preexisting trends in student demographics and educational outcomes. To achieve this, we employed an optimization algorithm that allowed us to select a group of non-affected counties, referred to as donor counties in our analysis, using the synthetic control method (SCM), whose characteristics closely mirrored those of the affected counties prior to the LWEO. The SCM employs an optimization algorithm to assign weights to donor counties based on their similarity to the affected counties in terms of key pre-intervention characteristics, such as student demographic composition, economic conditions, and outcomes, including prior trends in postsecondary enrollment, college readiness, and high school employment rates. The algorithm assigned greater weights to counties that exhibited pre-LWEO trends most similar to those of the affected counties, combining them into a synthetic group that closely matches the affected counties on observable characteristics.

Figure 1 presents differences in demographic characteristics of students enrolled in school districts located in counties affected by the LWEO prior to the raid compared to those in the constructed synthetic control group. All differences in demographic characteristics between the LWEO and donor counties are at or near zero. From the sample of students in the affected counties, we also constructed distinct subsamples for Latinx and English-learner students to examine whether the LWEO had disproportionate impacts on particular populations. Within the overall affected-county sample, 1 percent of students were Asian, 9 percent were Black, 22 percent were White Latinx, and 61 percent were White non-Latinx. During the 2018–2019 school year, 78 percent of these students were considered economically disadvantaged, 11 percent participated in a gifted and talented program, 1 percent were identified as immigrant students, 15 percent were current English learners, and 21 percent received special education services.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Differences Between LWEO Counties and Synthetic Control Group

Source: Authors’ calculations.

Note: Points to the right of the center line indicate higher values in LWEO counties.

Key Variables

Our study centers on two sets of outcomes. The first set of outcomes deals with the postsecondary trajectories, specifically analyzing direct enrollment in two-year and four-year colleges. We define direct enrollment as the enrollment of students in a postsecondary institution within one year following their high school graduation. This measure provides insights into the immediate educational progression of students and serves as a gauge for the broader educational aspirations and accessibility of higher education opportunities in the wake of the LWEO.

The second set of outcomes addresses several critical high school indicators that precede and likely influence postsecondary education decisions. These include assessing whether students met the college readiness standards for four-year institutions as defined by the THECB. Additionally, we evaluate completion rates of either the Free Application for Federal Student Aid (FAFSA) or the Texas Application for State Financial Aid (TAFSA) applications for financial aid to attend college. We also explore the likelihood of students working during their high school years, a factor that can significantly influence academic performance and the ability to engage in extracurricular activities that enhance college applications.

Appendix table A.1 presents the mean differences in outcomes for students in LWEO counties before and after the LWEO in 2019. The four-year college direct enrollment rate increased from 3.24 percent to 4.56 percent. Conversely, four-year college readiness dropped from 4.42 percent pre-LWEO to 3.22 percent post-LWEO. FAFSA and TAFSA application rates decreased from 57.84 percent to 46.06 percent. The proportion of students working during high school showed a minor decline, though this difference was not statistically significant. Lastly, two-year college direct enrollment experienced a small decrease, but this change was also not statistically significant. While these descriptive results show mixed trends in student outcomes following the LWEO, these findings do not account for broader time trends and confounding factors that might influence these outcomes.

We incorporate a rich set of covariates to account for factors that may influence the outcomes of interest. The covariates included a broad set of demographic information aggregated to the county level, providing a detailed snapshot of the student population within each county. This includes the percentage of students from various racial and ethnic backgrounds, the sex of the students, economic disadvantage status, English learner status, and the percentage of students receiving special education services. These covariates account for a wide range of observable characteristics that could confound estimates in identifying the effects of the LWEO on students’ educational trajectories.

Empirical Strategy

In our preferred analysis of the impact of the LWEO on educational outcomes, we adopt the SCM to construct a comparable counterfactual scenario representing what would have transpired had the raid not occurred. This approach is particularly apt for assessing events like the LWEO that affect a relatively small number of observational units—counties, in this case—where a constructed counterfactual can provide insight into what outcomes would have occurred in the absence of the intervention. The SCM technique leverages observed characteristics aggregated across multiple counties over time to create a synthetic control group that closely mirrors the attributes of the four affected counties (Abadie et al. 2010). Through SCM, we generated a synthetic control group for each model employed, incorporating covariates and lagged measures of outcomes in three of the preceding academic years (2013–2014, 2015–2016, and 2017–2018). The construction of the synthetic control uses an optimization algorithm to determine the appropriate weights for potential control units, which consist of other counties not impacted by the LWEO. By optimizing the selection of these weights, the synthetic control group was designed to closely mirror the pre-intervention characteristics of the LWEO-affected counties. This methodological approach allowed us to create a credible baseline against which we could measure the actual impacts of the LWEO, providing a clear view of the intervention’s effects on student educational trajectories. Figure 1 illustrates the similarity between the synthetic control group and the four affected counties, showing negligible differences across observable characteristics. This consistency was observed across all models employed in this study. Our analysis uses SCM models to examine outcomes for all students, as well as subsamples of Latinx and English-learner students, aligning with previous research highlighting variations in the effects of immigration enforcement across different student populations (Kirksey and Sattin-Bajaj 2021, 2023).

Tests of Robustness

In both appendix table A.1 and figure 2, we observe increases in four-year direct enrollment rates for the LWEO counties post-raid. While the SCM results suggest that the LWEO had a net negative effect—since the increase in LWEO counties’ enrollment rates is not as large as that of the donor counties—there remains a concern that the effect we observe from the raid is confounded by other unobserved factors. To address this, we implement a placebo test (Abadie et al. 2010). The placebo test helps determine whether the observed effect is unique to the LWEO counties by applying the SCM to a set of pseudo-treated counties, none of which were exposed to the LWEO. This robustness check simulates the LWEO in the donor counties to check for spurious effects due to preexisting trends in those counties, which ensures effects attributed to the LWEO are not merely coincidental. Appendix figure A.1 displays the results from the placebo test of LWEO’s impact on four-year direct enrollment rates. The bold line represents the four-year direct enrollment rate for the LWEO counties, and the lighter lines represent individual placebo counties within the donor pool. The placebo test results show that the observed increase in four-year direct enrollment rates for the LWEO counties post-raid is notably smaller than most of the placebo effects. While some fluctuation is observed in the placebo counties, the LWEO counties demonstrate a clear difference in enrollment rates post-raid, suggesting that the impact seen in the LWEO counties is unlikely to be driven by random variation. Additionally, we conducted another placebo test by shifting the intervention start year to one year prior to the actual LWEO, treating it as the baseline. This did not result in statistically significant changes. Lastly, we employed a leave-one-out sensitivity analysis, sequentially removing donor counties with large weights (> 0.1) to ensure that no individual county disproportionately influenced the synthetic control. While this slightly affected the estimated magnitude of the LWEO’s impact, the divergence between the affected counties and the synthetic control group remained consistent across outcomes and analyses.

Another potential concern with our analysis is that the observed negative effects of the LWEO immigration raid on student outcomes could be confounded by disruptions associated with the COVID-19 pandemic, which began shortly after the LWEO event. While we cannot rule out possible confounding effects attributable to the pandemic, we conducted additional analyses using a difference-in-differences framework, comparing our original estimates (based on the full cohort range, 2014–2023) with estimates derived from a narrower analytic window tightly centered around the LWEO event year (±2 years: 2017–2021). Our results remain robust under this narrower window. For example, when examining declines in enrollment in four-year colleges, these models produce similar, statistically significant declines in college enrollment of roughly 1 percentage point. Moreover, our baseline synthetic control analysis indicates that enrollment gaps persist and notably widen to approximately 2 percentage points in subsequent cohorts, strongly suggesting that the negative enrollment impacts we attribute to the LWEO event extend beyond, and are distinct from, pandemic-related disruptions alone.

Limitations and Further Research

While the LWEO appears to correspond to a measurable negative influence on students’ postsecondary outcomes, we acknowledge several limitations of our study. Foremost among these is the potential confounding role of the COVID-19 pandemic, which coincided with our post-raid observational period. The pandemic created unprecedented disruptions to schooling through shifts to remote learning, widespread school closures, and substantial changes in education policies and programs. These disruptions disproportionately affected economically disadvantaged communities, and it is plausible that pandemic-related impacts on school policies, resource allocation, and family circumstances may have compounded or interacted with the effects of the LWEO. Many students and families experienced significant instability during this period, potentially further diminishing their capacity to engage with college preparation, readiness, and enrollment processes. Although our inclusion of data through 2023 extends the observational window beyond the peak of pandemic disruptions, we cannot fully disentangle the observed LWEO effects from broader systemic educational disruptions that arose during the COVID-19 period.

Another limitation concerns our reliance on administrative, district-reported data, which does not allow us to identify specific students or families directly affected by the raid or their precise sources of legal vulnerability. Without individual-level data on immigration status or personal experiences of the raid, our findings necessarily reflect broader community-level impacts rather than direct causal effects at the individual level. Furthermore, while we employed rigorous methods—such as the synthetic control method and short-window difference-in-differences analyses—to mitigate potential confounding factors, these approaches depend heavily on the assumption that our control groups accurately reflect the counterfactual scenario in the absence of the LWEO.

Future studies should also explore explicitly the interactions between immigration enforcement and public health crises to better disentangle these overlapping effects. Moreover, longitudinal analyses that track students over extended periods may capture delayed impacts not observable in shorter-term, immediate post-raid studies. Qualitative investigations could further enrich our understanding by illuminating student experiences and institutional practices that promote resilience in the face of legal vulnerabilities. Lastly, the absence of observed effects in certain domains, such as two-year college enrollment, underscores the importance of continuing research into conditions and contexts under which educational outcomes are either resilient or susceptible to immigration enforcement disruptions. Such insights would meaningfully inform targeted policy interventions and educational supports for legally vulnerable student populations.