Abstract
Studies estimating the cumulative risk of imprisonment have been central to documenting the rise of incarceration and its consequences for inequality in the United States. Data limitations have stymied efforts to document the risk of criminal legal contact falling short of imprisonment, however, and few studies estimate risks among groups other than Black and White men. Although less severe than imprisonment, subfelony contact also carries damaging and lifelong consequences. Using more than 25 years of administrative criminal record data from Texas, this study estimates the prevalence of both subfelony and felony arrest, prosecution, conviction, and incarceration across five cohorts of Black, White, and Hispanic men and women born between 1971 and 1995. Findings show that subfelony contact is 3‒18 times as prevalent, and far more widely distributed across the population, as felony conviction and incarceration: fewer than 3% of Black women go to prison by age 24, but at least 22% are arrested at least once. Although risks of felony events have fallen substantially for some groups, subfelony criminalization has persisted at high and increasingly unequal rates. These findings have implications for our understanding of the reach of the criminal legal system and its role in shaping inequality in the twenty-first century.
Introduction
The rise of mass incarceration beginning in the 1970s dramatically altered the social landscape in the United States. An extensive number of cohort studies estimating the population prevalence of imprisonment have revealed that incarceration became a life course event for an entire generation of poor Black men and deepened racial inequality across social and economic domains (Muller and Wildeman 2016; Pettit and Western 2004; Robey et al. 2023; Roehrkasse and Wildeman 2022; Shannon et al. 2017). Moreover, mass incarceration greatly expanded the criminal legal system's capacity to surveil, punish, and control large swaths of the population beyond the prison walls, a phenomenon that can be described as mass criminalization (Beckett and Murakawa 2012; Hinton and Cook 2021; Kohler-Hausmann 2018).
Criminalization, in this usage, refers to the front-end processes and low-level contact, such as arrest, prosecution, adjudication, misdemeanors, and alternative sanctions, that fall short of felony conviction and imprisonment but nonetheless operate to demarcate select individuals as “criminal” and thus subject to punitive control (Hinton and Cook 2021; Jenness 2004). These front-end processes and “subfelony” (Kohler-Hausmann 2022) events, particularly at the misdemeanor level, make up the vast majority of criminal legal activity but are poorly captured in national corrections data (Finlay and Mueller-Smith 2021; Stevenson and Mayson 2018). The life table studies that revealed the high and highly concentrated prevalence of incarceration among Black men relied on extrapolations from federal and state data sources providing point-in-time enumerations and representative surveys of the population under correctional supervision. No comparable data sources document criminalization falling short of felony conviction and incarceration, a fact that has significantly hindered our understanding of the full scope of the criminal legal system and the prevalence of subfelony contact across the population.
Although less severe than imprisonment, the experiences and marks left by subfelony criminalization can last a lifetime and carry damaging consequences. For example, arrest is a traumatic event shown to have measurable negative impacts on the health and well-being of the person arrested and their family (Geller et al. 2014; Legewie and Fagan 2019; Sewell and Jefferson 2016; Turney 2022). The process of being prosecuted, even when it does not end in conviction, can involve extended detention in jail, where conditions are frequently unsanitary, degrading, and dangerous (Walker 2022; Wildeman et al. 2018). Even relatively short periods of jail confinement decrease the likelihood of a favorable court outcome, damage employment (Dobbie et al. 2018), and decrease voter turnout (White 2019). Importantly, each criminal legal event generates a mark—or record—that is legible on background checks and can negatively affect access to resources and opportunities ranging from housing to employment to dating apps (Lageson 2022; Pager 2007). When exposure to criminal legal contact is both high and highly unequal, these damaging consequences can exacerbate racial and class inequalities at the population level (Lee et al. 2014; Western and Muller 2013; Wildeman 2016).
In this study, I draw on more than 25 years of individual-level state criminal record data from Texas that are inclusive of all recorded misdemeanor and felony arrest, prosecution, adjudication, and sentencing events to estimate the cumulative prevalence of criminalization to age 24 across five cohorts of Black, White, and Hispanic men and women born between 1971 and 1995. Estimation of the changing prevalence of each stage of criminalization at both the misdemeanor and felony levels advances our understanding of the scale, scope, and dynamics of the criminal legal system over a nearly three-decade period of immense social, political, and economic change. These birth cohorts entered adolescence and early adulthood—the peak ages for arrest and lawbreaking—between the early 1990s, when crime rates were high and the prison boom was in full swing, and the 2010s, an era of historically low crime rates and emergent bipartisan penal reform.
This study contributes to a growing literature documenting cohort differences in the prevalence of criminalization. Prior research in this area has focused on singular events, estimating changes in the risk of jail incarceration (Western et al. 2021) and misdemeanor conviction (Hepburn et al. 2019) in New York City and arrest among young people in Chicago (Neil and Sampson 2021) and nationwide (Brame et al. 2014; Weaver et al. 2019). Given the limited measures and few reliable estimates of criminal legal contact among women, open questions remain about the prevalence of contact across the U.S. population and whether, how, and for whom the risk of criminalization has changed.
Background
Cohort Studies of Incarceration
A well-established body of research comparing the prevalence of imprisonment across different birth cohorts has documented the historic rise of incarceration through the twentieth century (Muller and Wildeman 2016; Pettit and Western 2004; Robey et al. 2023; Roehrkasse and Wildeman 2022; Shen et al. 2020). Although the incarceration rate—a measure of the proportion of people incarcerated at any given time—shows that imprisonment increased by nearly 500% from 1973 to 2007 (Travis et al. 2014), only prevalence statistics can identify the proportion and demographic contours of the population who has ever been incarcerated. Prevalence estimates revealed that, among cohorts coming of age amid rising incarceration, two in three Black men without a high school diploma and close to one in three Black men overall would go to prison by midlife (Bonczar 2003; Roehrkasse and Wildeman 2022; Western and Pettit 2010; Western and Wildeman 2009).
Studies that compare prevalence rates across cohorts uniquely reveal how macrosocial and policy changes that transform the role, capacity, and power of institutions can reshape the life course trajectories of entire cohorts (Ryder 1965). Pettit and Western's (2004) formative study of changes in the cumulative risk of imprisonment among men coming of age before and during the prison boom showed how the period conditions present in the 1970s and 1980s—namely, White anxiety over racial progress, rising crime rates, and moral panic around drug use that fueled hyperpunitive public policy (Travis et al. 2014)—coalesced to reshape the life course trajectories of young people over the next several decades.
By 2010, however, period conditions thought to drive mass incarceration had changed significantly. Crime rates had fallen by close to 50% since peaking in the early 1990s, and the moral panic around crime and drug use had waned (Western et al. 2023). Budget shortfalls, worsened by the Great Recession, prompted states to reconsider the logic of lengthy and mandatory prison sentences for low-level drug and property crimes (Beckett et al. 2016; Dagan and Teles 2016; Green 2015; Phelps and Pager 2016). In short, these shifts portended a historic decline in imprisonment. Indeed, as the macrosocial and policy environment changed, so too did the risk of imprisonment. Black men born in 1991 are almost half as likely as Black men born in 1971 to go to prison by midlife (Robey et al. 2023). Whether these declines in the risk of imprisonment have been matched by comparable declines at other levels of criminalization remains an open question.
The national corrections data used in these cohort studies of incarceration do not include comparable measures of subfelony and front-end processes, making it difficult to know just how many people have experienced criminalization and whether and for whom the prevalence of contact has changed amid shifting period conditions. Data limitations have also led to the omission of women from most cohort studies of criminal legal contact; because women have much lower rates of arrest and imprisonment than men, national surveys often fail to capture a sufficient number of women with contact to estimate valid prevalence statistics.
Cohort Studies of Subfelony Criminalization
In the absence of national data sources on subfelony and front-end criminalization, researchers have turned to longitudinal surveys and administrative data from single jurisdictions to estimate the risk of arrest (Brame et al. 2014; Neil and Sampson 2021; Weaver et al. 2019), misdemeanor conviction (Hepburn et al. 2019), and jail incarceration (Western et al. 2021), providing the first estimates of the prevalence of contact falling short of felony conviction and incarceration. These studies are informative of diverging trends in felony and subfelony contact at the turn of the century.
Consistent with trends in incarceration, these studies of subfelony contact find that young people born in the early 1980s and who came of age in the late 1990s and early 2000s experienced criminalization at levels far greater than either national crime rates or self-disclosed involvement in crime and risky behavior would suggest (Brame et al. 2014; Weaver et al. 2019). For the 1980‒1984 birth cohort, approximately 49% of Black men and 38% of White men had been arrested at least once by age 23, prevalence rates much greater than were found for the 1957‒1964 birth cohort (Brame et al. 2014; Weaver et al. 2019). Using data from Chicago, Neil and Sampson (2021) found that arrest exposure among 1980s birth cohorts was very high and highly concentrated by race and class, with as many as 70% of economically disadvantaged Black men experiencing an arrest by their mid-20s.
These studies are also informative of trends in criminalization across cohorts. Comparisons of self-reported crime and arrest between the two study cohorts of the National Longitudinal Survey of Youth (NLSY) show that individuals in the 1980‒1984 cohort, from NLSY97, were more likely to have been arrested but less likely to have engaged in crime than individuals in the 1957‒1964 cohort surveyed in the NLSY79 (Weaver et al. 2019). This finding runs counter to evidence from New York City, which showed that the cumulative risk of criminalization through misdemeanor conviction was highest among the 1961‒1965 cohort and has since declined with each subsequent birth cohort up to 1981‒1985 (Hepburn et al. 2019). However, arrest and misdemeanor conviction are different events that are responsive to different forces. In New York City, for example, the number of misdemeanor arrests doubled between 1985 and 2012, yet the misdemeanor conviction rate dropped by half because prosecutors increasingly disposed of cases through dismissal (Kohler-Hausmann 2018). In other words, changes in prosecution practices, such as the introduction or expansion of conditional dismissals and other alternatives to conviction, will affect the prevalence of conviction without directly impacting the prevalence of arrest.
There are few estimates of the risk of any type of criminalization among cohorts born after 1985, but the available evidence suggests that they are at a much lower risk of all types of criminal legal contact than their predecessors. Neil and Sampson (2021) found that the risk of arrest for young people in Chicago dropped substantially between 2000 and 2016, such that people born in the mid-1990s were half as likely as those born in the 1980s to be arrested during adolescence. Western and colleagues (2021) similarly found that the cumulative risk of jail incarceration by age 29 in New York City fell between 2008 and 2017, dropping by close to 50% for Black and Latino men. These trends suggest that subfelony criminalization has declined across birth cohorts in much the same way as felony incarceration.
Trends in exposure provided in these studies diverge along gender lines. Theoretical accounts arguing that criminalization remains central to the control of racially marginalized populations suggest that both women and men of color are surveilled and criminalized at much higher rates than Whites, even when crime rates are low (Beckett 2018; Hinton and Cook 2021). Evidence from New York City is consistent with this theory, showing that race‒gender disparity in subfelony contact is so high that Black women are at greater risk of misdemeanor conviction and jail incarceration than both White women and men (Hepburn et al. 2019; Western et al. 2021). This finding is surprising because research has consistently shown that men are arrested and incarcerated at much higher rates than women; estimates from the NLSY97 show an arrest risk of 38% for White men versus 10% for Black women (Brame et al. 2014).
It is likely that prior studies of criminalization underestimate contact among women, however. Western and colleagues (2021) provided robust estimates of women's jail incarceration in New York City, but because not all arrests result in jail detention, their estimates represent only a lower bound of overall criminalization. Hepburn and colleagues (2019) used data that capture misdemeanor convictions in New York City, but because large numbers of misdemeanor cases are dismissed (Kohler-Hausmann 2018), the study cannot estimate the number of women (or men) ever arrested or charged with a misdemeanor. Estimates of women's arrest from the NLSY97 are unreliable because the data rely on self-reporting, which has been shown to vary across demographic groups (Kirk 2006), and contain so few women reporting any criminal legal contact that race-specific rates cannot be estimated (Brame et al. 2014).
Nonetheless, this body of research highlights important patterns in the prevalence of subfelony criminalization that warrant further investigation and call for the expansion of research to jurisdictions beyond New York City and Chicago. These studies capture only single stages of the criminal legal process, meaning they cannot document historical change in the processing of cases from arrest to prosecution to conviction, nor do they capture differences in felony- and subfelony-level events. It is imperative to capture the full criminal process because preadjudication procedures likely contribute to racial disparity in conviction and incarceration (Bing and Gutierrez 2025; Omori and Petersen 2020; Western et al. 2023) and are significant life events that can deprive people of liberty, safety, privacy, and dignity (Geller et al. 2014; Legewie and Fagan 2019; Rios et al. 2020; Walker 2022).
The aims of this study are twofold. The first is to document the reach of criminalization at both subfelony and felony levels. Prior studies showing that misdemeanors far outnumber felonies (Stevenson and Mayson 2018) suggest that subfelony criminalization is a common life event in the transition to adulthood for a large swath of the population, but data limitations have prevented scholars from documenting its prevalence. The second aim is to identify historic change in the scale and sociodemographic concentration of criminalization across birth cohorts. Trends showing declining risks of felony imprisonment at the national level (Robey et al. 2023), particularly among Black men, and declining arrest and jail incarceration risks in New York City and Chicago (Neil et al. 2021; Western et al. 2021) suggest that exposure to other forms of criminalization may have declined for recent birth cohorts as well. Yet, as some scholars have suggested, subfelony criminalization may be responsive to forces that differ from those driving felony criminalization (Beckett et al. 2016; Kohler-Hausmann 2018), and changes in penal policy or practice may benefit some groups more than others (Lofstrom et al. 2020; Mauri et al. 2025). Although identifying the driving forces behind mass criminalization remains outside the scope of this study, documenting its scale, dynamics, and demographic distribution across birth cohorts coming of age during and after the prison boom is an important first step.
Data, Measures, and Methods
Texas
Texas is an ideal state for this study. First, as a politically conservative, southern state with substantial urban and rural populations, Texas provides an opportunity to study trends in criminalization in a different sociopolitical environment than New York and Chicago—where prior research on subfelony contact has focused. Texas was also at the forefront of the recession-era wave of penal reforms and was one of the first states to pursue a Justice Reinvestment Initiative—a package of reforms implemented in more than 30 states between 2007 and 2016 (Dagan and Teles 2016; Green 2015; Harvell et al. 2016; Sabol and Baumann 2020). Thus, findings from Texas may be generalizable to other politically and socially similar locales.
Texas is also home to the largest population of Black Americans in the United States (U.S. Census Bureau 2022). Even if findings from this study are isolated to Texas, the scale and racial disparity of punitive surveillance and punishment are consequential for the life chances and well-being of nearly 4 million Black Americans—an important impact in and of itself. Additionally, it is also home to a growing number of both U.S.-born and immigrant Hispanic and Latino/a individuals. As of 2020, more than 40% of the state population identified as Hispanic1 (U.S. Census Bureau 2022). Growing anti-immigrant attitudes and policies, championed by conservatives in Texas and elsewhere, have heightened the typification of Hispanics as criminal and raised the stakes of criminalization for immigrant and mixed-status Hispanic families (Alcalá and Montoya 2018; Beckett and Evans 2015; Light et al. 2014). Documenting trends in criminalization among Hispanics in Texas will shed light on the impacts of social and policy changes in the early twenty-first century.
Data
Criminal record data come from the Texas Department of Public Safety's Computerized Criminal History Database (CCH); data were obtained directly from the department through a data use agreement in adherence to Texas Government Code 411.083. The CCH is the state repository for criminal record information and contains the universe of recorded arrest, prosecution, and sentencing data for all misdemeanor2 and felony offenses for persons 17 or older. The CCH contains records of more than 22 million arrests of approximately 7.7 million unique individuals, and recent audits show that it is one of the most comprehensive and complete state criminal record repositories in the country (Goggins and DeBacco 2020).
Texas CCH data contain an identity file that includes an array of identifiers, including full name, known alias, date of birth, Social Security number, fingerprint codes, residential address, and driver's license number, as well as demographic descriptors including race and ethnicity (recorded separately), country of birth, and country of citizenship. Entry in the CCH identity file occurs at the point of first arrest, regardless of whether that arrest results in a conviction. Individuals are then assigned a unique system ID number that is biometrically linked to all future arrest, prosecution, court, and sentencing events recorded in the data. A detailed description of the contents, structure, and reporting practices for the CCH, including the data dictionary and codebooks, is publicly available on the Texas Department of Public Safety website.3
Although CCH coverage dates begin in 1975, records from early years suffer from underreporting and missing data. Therefore, I limit analyses to cohorts coming of age after 1990, when data quality improved. The data were obtained in the spring of 2018 and contain complete coverage of events occurring in 2017 but only partial coverage of 2018. Estimates for cohort members reaching age 24 in 2018 and 2019 are thus interpolated using age-specific rates from cohort members observed to age 24. A description of the interpolation technique is located in the online appendix.
Population data used for the denominators of the prevalence estimates come from the Centers for Disease Control Bridged-Race Population Estimates, accessed via CDC WONDER (Centers for Disease Control and Prevention [CDC] 2020). These data contain single-year age-, sex-, and race/ethnicity-specific population estimates for Texas for 1995 to 2019, the years in which birth cohort members reached age 24, and are already adjusted for migration and mortality. Age-, sex-, and race/ethnicity-specific mortality rates used to adjust for cohort mortality for the numerators, or the members of the cohort observed experiencing the event, come from the CDC compressed mortality file, also accessed through CDC WONDER (CDC 2017).
Measures
For each cohort, I estimate the prevalence of misdemeanor (subfelony) and felony arrest, prosecution, conviction, and incarceration to age 24. Because these measures of criminal legal contact refer to both processes and outcomes, I refer to them collectively as criminal legal events. These measures do not describe crime or “offending.” Criminal records document the actions of police, prosecutors, and courts—not defendant behavior. Although crime and arrest are not wholly unrelated, readers should be cautious not to misinterpret estimates from this study—or any study reliant solely on criminal record data—as evidence of cohort or sociodemographic differences in lawbreaking behavior.
Measures of the four legal events come from distinct record types in the CCH and logical statements consistent with the sequence of criminal processing. Arrests are measured through records appearing in the arrest file, charges from records found in the prosecution file, and conviction and incarceration sentences from records in the adjudication file. Because of the sequential nature of case processing, missing earlier data are imputed using later records (e.g., adjudication implies a prior arrest and prosecution). Incarceration sentences following a misdemeanor conviction, which are served in county jails, are coded as subfelony incarceration. Incarceration sentences following a felony conviction, which are typically served in state jail or prison, are coded as felony incarceration.
In Texas, as elsewhere, acquittals (not guilty verdicts) are extremely rare. Rather, nearly all people who are prosecuted but not convicted have their case dismissed for something like insufficient evidence or after satisfying some set of conditions, such as paying restitution or completing probation. In such cases of conditional dismissal, which are formalized as deferred adjudication in Texas, the failure to complete the agreed upon conditions can result in a final conviction. In this study, deferred adjudication is counted as conviction only if the final conviction entry occurred during the cohort observation period.
Race and ethnicity of individuals come from the CCH identity files. For simplicity, I refer to “race and ethnicity” as “race,” but this usage is not intended to conflate the two characteristics. For analyses, race and ethnicity categories in the CCH data are coded to match the 1990s census categories, which include non-Hispanic White, Black (of any ethnicity), and Hispanic White.4 Coding for Hispanic ethnicity required additional steps because, as evidence shows, the measurement and recording of Hispanic ethnicity in administrative criminal justice records has varied over time and across data sources, leading to biased estimates of racial and ethnic differences in contact (Beck and Blumstein 2018; Tonry and Melewski 2008).
Texas CCH data record race and ethnicity separately, overcoming many of the challenges of missing race data for Hispanic individuals noted by Beck and Blumstein (2018) to bias estimates from Bureau of Justice Statistics data. Even without missing data on race and ethnicity, there remains a risk of Hispanic individuals being identified as non-Hispanic because the data are not always based on self-report. To address this potential source of bias, I code Hispanic ethnicity using country of birth and citizenship data in the CCH identity file and parent/self-reported ethnicity in linked K-12 public school and higher education data from the State Longitudinal Data System–Texas (SLDS–Texas). SLDS–Texas data were accessed through the Education Research Center at the University of Texas at Austin. The CCH and SLDS–Texas data were linked using a combination of name, date of birth, and Social Security number. See the online appendix for information on matching procedure and match rates.
Individuals are coded as Hispanic if they are (1) marked as Hispanic in the CCH record, (2) marked as born in or citizens of Mexico or another Latin American country in the CCH identity file, or (3) marked as Hispanic in SLDS–Texas education records. Hispanic ethnicity could be verified across at least two administrative data sources (CCH citizenship record, CCH race/ethnicity record, SLDS–Texas education records) for approximately 79% of individuals. Nearly all of those who were newly identified as Hispanic through this process were recorded as White in the race variable. A few individuals were identified as both Black and Hispanic (∼4%), and these individuals were coded as Black to align with 1990 census population categories.
Across all cohorts, approximately 17% of individuals coded as non-Hispanic White in the CCH identity file were identified as Hispanic via citizenship, country of birth, or education data. Citizenship and country of birth data were available for 100% of the sample. Linked SLDS–Texas data were available for between 74% and 84% of those born in 1976–1995, but for only 46% of those born in 1971–1975. This difference is because SLDS–Texas data coverage begins in the early 1990s, when the majority of the 1971–1975 birth cohort had already aged out of K-12 education.
Methods
Unique individuals in the CCH are grouped by birth year into sex- and race-specific five-year birth cohorts, beginning with 1971–1975 and ending with 1991–1995. Cumulative rates are inclusive of criminal legal events occurring before or during the year in which cohort members reach age 24. Because the CCH contains the universe of recorded arrests, charges, convictions, and sanctions for Class B misdemeanors or greater in Texas, I directly estimate the prevalence of criminal legal events in the population by dividing the total number of people who experienced the event at least once by age 24 by the same sex, race, and age population living in Texas. For the 1971–1975 birth cohort, for example, this is the number of unique individuals of each sex‒race grouping who experienced the criminal legal event at least once before or during the year each member reached age 24 (1995–1999) divided by the same sex and race population of 24-year-olds living in Texas in 1995–1999. Analyses also compare conditional conviction rates, or the proportion of people charged with an offense at least once by age 24 who were marked with conviction by age 24. Conditional conviction rates highlight divergent trends across different stages of the system and how case processing can shape sociodemographic disparity.
I estimate risks to age 24, rather than age 30 or 40, for both methodological and conceptual reasons. Methodologically, doing so allows the direct estimation of risks for cohorts born after 1986—the birth cohorts that prior research suggested are the most likely to experience declining risks of criminalization (Neil and Sampson 2021). Extending estimates beyond age 24 would require extrapolation beyond what is observable in the data. Because these extrapolations use patterns observed among older cohorts to predict unobserved trends among younger cohorts, extending estimates beyond age 24 may undermine the study's capacity to detect change. Furthermore, extending estimates to older ages offers diminishing returns to measures of cumulative risk because the vast majority of first contacts with the legal system occur before age 24 (Gottfredson and Hirschi 1990; Sampson and Laub 1992).
Conceptually, estimating to age 24 captures the cohorts during their transition to adulthood—an important period in the life course during which young people are solidifying their identities, gaining independence, and developing experiences, relationships, and credentials that will shape the rest of their lives (DeMarco et al. 2021; Ryder 1965). Documenting cohort and sociodemographic differences in criminalization during this pivotal moment in the life course can help to uncover potential drivers of inequality at later ages and foreground predictions about the future of mass incarceration and criminalization. Indeed, evidence shows that cohort incarceration rates later in life are greatly influenced by exposure to criminalization in adolescence and early adulthood (Shen et al. 2020).
Migration and mortality adjustments for the numerator—the population who experienced a criminal legal event—were made in the following manner. For migration, people whose arrest record listed an out-of-state residential address were dropped. Estimates are not further adjusted because out-migration from Texas over the period was low and greatly exceeded by in-migration (U.S. Census Bureau 2000, 2003, 2012; White et al. 2016), suggesting that leavers were more than replaced by entrants. Furthermore, evidence suggests that geographic mobility is lower among the system-involved population than the general population (Kirk 2009). Supplemental analyses, available in the online appendix, also show that results are robust to adjustments for several different levels of out-migration.
For mortality, sex-, age-, and race-specific mortality estimates for Texas from the CDC WONDER compressed mortality file were used to adjust for expected cohort mortality between ages 17 and 24. Upper confidence interval estimates were used to produce conservative estimates and account for higher-than-average mortality among the system-involved population (CDC 2017; Shannon et al. 2017). Age-specific mortality data for Hispanic individuals are not available for the 1990s. For these years, Hispanic mortality estimates were interpolated from 2000 estimates for Hispanics using the rate of change observed for Whites between 1990 and 2000, adjusted for White‒Hispanic mortality differences from 2000 to 2019.
Table 1 displays summary statistics describing the population observed in the CCH who experienced any criminal legal event by age 24, arranged by birth cohort. Summary statistics describing the overall population (the full birth cohort alive and in Texas at age 24) are included in italicized numbers to show how the population with criminal legal contact varies compositionally from the overall birth cohort. In total, more than 2.2 million Texans born in 1971–1995 acquired a criminal record by early adulthood. Compositional differences across cohorts point to two notable trends. First, the proportion of women has increased over time, composing 29% of the 1991–1995 cohort compared with 22% of the 1971–1975 cohort. Second, Hispanics also make up a growing proportion of those experiencing a criminal legal event by early adulthood, accounting for 47% of the 1991–1995 cohort compared with 28% of the earliest cohort. This change is consistent with population trends in Texas, where the Hispanic population grew by more than 60% between 2000 and 2015, but it also coheres with evidence suggesting that the criminalization of Hispanics has increased over time and in response to anti-immigration sentiment and policy (Light et al. 2023).
Results
Results are shown as plots for each criminal legal event, where the points reflect the estimate for each race‒cohort combination and the lines illustrate the direction and magnitude of difference from one cohort to the next. Results for men and women are plotted separately. A complete table of prevalence rates for all events is available in the online appendix.
Cumulative Risk of Arrest for Men and Women
Figure 1 illustrates that arrest has been a common event in the lives of young people for more than 25 years, particularly among Black men and women and cohorts born before the 1990s. Approximately 55‒61% of Black men, 20‒25% of Black women, 30‒40% of White and Hispanic men, and 8‒15% of White and Hispanic women born in 1971–1990 were arrested at least once by age 24. The risk of arrest among the 1991–1995 cohort, the cohort shown by Neil and Sampson (2021) and Robey et al. (2023) to have the lowest risk of arrest and imprisonment in the last 25 years, was much lower for some race‒sex groups but appeared to have declined little, if at all, for others. Compared with the risk of arrest among cohorts born in the 1970s, the arrest risk for the 1991–1995 cohorts of Black and White men were 29% and 39% lower, respectively. For Black and White women, however, the risk of arrest remained relatively stable across cohorts—with more than one in five Black women and one in 10 White women arrested by age 24. For Hispanic men and women, the risk of arrest broadly increased across birth cohorts and although it was highest for the 1980s cohorts, rates for the 1991–1995 cohort exceeded or were comparable to rates for the 1970s cohorts.
Figure 1 also shows risk of arrest by the level of the alleged offense, revealing that subfelony arrests were both more prevalent and more widely distributed across the population than felony arrests. Subfelony arrest was approximately twice as prevalent as felony arrest for White and Hispanic men and three times as prevalent among White and Hispanic women. By age 24, 14‒17% of White men born in 1971–1985 were arrested on a felony offense, but more than 32% were arrested on a subfelony offense. For Black men and women, the cumulative risk of subfelony arrest among all cohorts was even higher, averaging 45% for men and 20% for women; these groups faced extremely high risks of felony arrest as well, with an average of 31% for Black men and 8% for Black women.
Results for subfelony arrest alone—when people were arrested on a subfelony offense but have no record of a felony arrest by age 24—showed very little racial disparity and very little change across cohorts among men. Except for the youngest cohort of White men, whose risk fell to 16%, between 20% and 25% of all other cohorts of men were arrested only for a subfelony offense by age 24. For women, subfelony arrest alone made up a much larger share of all arrests and thus did not depart from observed racial disparity or change over time trends in other arrest measures.
The most notable change in arrest across birth cohorts is the decline in the risk of felony arrest among Black and White men. For both groups, the risk of felony arrest by age 24 fell by close to 50% between the 1990s and 2010s. For Black men, the decline in felony arrest risk explains much of the decline in the risk of any type of arrest, as the risk of subfelony arrest remained fairly stable up until the very youngest cohort.
Cumulative Risk of Charge, Conviction, and Incarceration for Men
Although many arrests never progress to prosecution, Figure 2 shows that men experienced high cumulative risks of being charged, convicted, and sentenced to jail for a subfelony offense. Nearly one in three Black men and about one in five Hispanic men born in the 1980s were convicted of a subfelony offense by age 24, and similar proportions were sentenced to serve jail time. Risks fell for the 1991–1995 cohort but remained roughly equivalent to rates of the 1971–1975 cohort for all groups except for White men—the only group whose risk fell to more than two points below 1971–1975 cohort levels.
The risk of being charged and convicted for a subfelony offense both increased from 1971–1975 cohort levels, but the conditional conviction rate—or the percentage of people ever charged who are convicted by age 24—decreased across these cohorts. Looking just at subfelony events, 32% of Black men from the 1971–1975 cohort were charged and 23% were convicted—reflecting a conditional conviction rate of 72% (.23 / .32). For the 1986–1990 cohort, 43% of Black men were charged and 30% were convicted—for a conditional conviction rate of 69%. So, while the prevalence of both charges and convictions increased across cohorts, the likelihood that someone charged with a subfelony offense would also be convicted of one by age 24 actually declined.
Results also show great racial disparity in conditional conviction rates, such that the percentage of those charged with a subfelony who were convicted by 24 was about nine percentage points lower for White men than for Black men—a difference that persisted across cohorts. The Black‒White gap in the conditional conviction rate for felonies was larger, ranging from 10 to 15 points, but did decline across cohorts. Although the precise cause of this disparity cannot be determined from prevalence statistics alone, these findings align with evidence from prior studies showing that Black men are systematically disadvantaged in case processing decisions relative to White men and thus receive case dismissals and alternatives to conviction less often (Bing and Gutierrez 2025; Omori and Petersen 2020; Western et al. 2023).
Cumulative Risk of Charge, Conviction, and Incarceration for Women
Figure 3 shows the cumulative risk of subfelony and felony events for women. Over the period of declining crime, young women faced a rapidly rising risk of subfelony prosecution. The risk of subfelony charge doubled for all groups of women between the 1971–1975 cohort and the 1986–1990 cohort. Nearly 20% of Black women born in 1986‒1990 were charged with at least one subfelony offense by age 24, and close to 10% received a subfelony conviction. Changes in subfelony conviction and incarceration were not nearly as large but also increased among cohorts who reached adulthood amid the crime drop, before falling slightly among the 1991–1995 cohort.
At both the subfelony and felony levels, prosecution was much more common than conviction and incarceration for women across all cohorts. The conditional conviction rate for women was much lower than that for men. Of women who were ever charged with a subfelony or felony offense, no more than 58% were marked with conviction. For the youngest cohort, no more than 43% of those who were charged received a conviction. The relatively low conditional conviction rate among women is consistent with sentencing research showing that women tend to receive more lenient treatment in court than men (Doerner and Demuth 2014). It is important to note that people who are charged but not convicted may be required to participate in probation supervision, submit to drug testing, or pay restitution and court fees as conditions of case dismissal. Interventions such as pretrial diversion and deferred adjudication are common in Texas, as they are elsewhere (Wright and Levine 2021).
Racial Disparity
Table 2 shows racial disparity ratios for men and women by event type. Both Black‒White and Hispanic‒White racial disparity at the subfelony level is about half that of the racial disparity observed at the felony level. Racial disparity also increases across successive stages in case processing and with the severity of the event, with the greatest difference in felony incarceration and the greatest parity in misdemeanor arrest. For men, Black‒White racial disparity in subfelony events does not meaningfully decline across cohorts, and at the felony level there are only modest declines (from 4.0 to 3.5) in incarceration. Black‒White disparity in felony events does fall considerably across cohorts of women, resulting in comparable levels of disparity between felony and subfelony events among the youngest birth cohort.
Hispanic‒White disparity differs considerably. Hispanic men and women in the 1971‒1975 cohort were considerably less likely than White men and women to have experienced any type of criminal legal event. However, as more Hispanics in each successive cohort experienced criminalization events, the ratios approach parity and then, at least for men, reverse. Among the 1991–1995 cohort, Hispanic men were more likely to have experienced a criminal legal event than White men.
Discussion
Cohort studies estimating the cumulative risk of imprisonment have been central to documenting the rise of mass incarceration and its consequences for inequality. Data limitations have stymied efforts to document the risk of criminal legal contact falling short of felony conviction and imprisonment. Although less severe than imprisonment, subfelony contact also leaves the mark of a criminal record and can have damaging consequences (Geller et al. 2014; Kohler-Hausmann 2018; Lageson 2020). Drawing on more than 25 years of detailed, individual-level criminal record data from the state of Texas, this study addresses these gaps in the literature by estimating the cumulative risk of both misdemeanor and felony arrest, prosecution, conviction, and incarceration to age 24 across five birth cohorts of Black, White, and Hispanic men and women born between 1971 and 1995. It makes several novel empirical contributions that advance our understanding of the scale and dynamics of mass criminalization in the twenty-first century.
First, findings show that subfelony criminalization was several magnitudes more prevalent than felony conviction and incarceration, particularly among women. By age 24, 55‒60% of Black men, nearly 40% of White and Hispanic men, and approximately 25% of Black women born between 1971 and 1990 had been arrested at least once. Among these cohorts, arrest was 3‒7 times as prevalent as felony conviction for men and 8‒18 times as prevalent for women. Even among the 1991‒1995 cohort, who have been found to face the lowest risks of criminalization, risky behavior, and incarceration in several generations (Baumer et al. 2021; Neil et al. 2021; Robey et al. 2023), at least one in three Black men, one in four Hispanic men, and one in five White men were prosecuted for a subfelony offense by age 24. These risks are 2‒3 times as high as the risk of felony prosecution and are on par with risks experienced by the 1971‒1975 cohort.
Results showing persistently high and unequal exposure to criminalization among women highlight the utility of administrative data in capturing groups that are likely to be underestimated in national survey data. Previous estimates of women's risk of arrest relying on data from the NLSY97 found no evidence of racial disparity and estimated an arrest risk of about one in 10 (Brame et al. 2014). My results using population-level data show that twice as many young Black women were arrested by early adulthood as previously thought. Evidence of the growing criminalization of women and the considerable exposure faced by Black women in particular highlight the need for more research on women's direct experiences with the criminal legal system, especially at the subfelony level.
Second, cohort differences in the prevalence of criminalization varied across demographic groups as well as levels and stages of the system. White men experienced the largest overall declines in criminalization across cohorts. Although Black men from the 1991‒1995 cohort had a 50% lower risk of being sentenced to prison by age 24 than Black men born 20 years earlier, which is consistent with trends shown by Robey and colleagues (2023), I found that their risk of being sentenced to jail for a subfelony offense was not any lower. Fully 20% of all Black men in Texas born in the 1990s received a jail sentence for a misdemeanor offense by age 24.
For Black and White women, there was very little change in criminalization over the birth cohorts studied, whereas among Hispanic men and women—groups often omitted from this type of research because of measurement challenges—the risk of all types of contact generally increased across cohorts. This finding aligns with evidence that Hispanics have become increasingly targeted for criminal legal control during the 2000s (Beckett and Evans 2015; Brown et al. 2018; Muchow 2024). It also aligns with demographic trends showing that second- and third-generation immigrants, who represent a larger share of the 1980s and 1990s cohorts, tend to appear more similar than first-generation immigrants to their nonimmigrant counterparts on a number of measures (Alcalá and Montoya 2018; Bersani 2014).
In this study, Hispanic ethnicity was validated using a novel method of combining multiple sources of administrative data that include self-reported and parent-reported ethnicity from noncriminal justice records. Across all birth cohorts, 17‒30% of individuals who matched to SLDS‒Texas records and were coded as non-Hispanic White in the Texas CCH were self-identified or parent-identified as Hispanic in linked education records, with the greatest proportion among the 1991‒1995 cohort. This approach shows that, even in criminal record data that contain measures of race and Hispanic ethnicity, a large proportion of Hispanics may be miscoded as non-Hispanic White. Properly accounting for the proportion of Whites who identify as Hispanic, rather than the proportion of Hispanics who are white, is an important step in generating valid estimates of the population prevalence of criminal legal contact and racial disparity.
For the cohorts born in the 1970s, high rates of arrest, particularly at the felony level, align with high rates of serious crime during the 1980s and early 1990s. However, most of the people born in the 1980s reached late adolescence and early adulthood between the mid-1990s and early 2010s, when crime rates had dropped substantially. Yet, the risk of subfelony criminalization remained more or less unchanged from rates experienced by the 1976‒1980 cohort. Even among the 1991–1995 cohort, 43% of Black men and 36% of Hispanic men were arrested by early adulthood. These findings highlight the persistence of low-level criminalization amid evidence of declining rates of felony arrest and serious lawbreaking and indicate that cohort differences in criminal offending are likely insufficient to explain cohort differences—and continuities—in criminalization (Neil and Sampson 2021; Western et al. 2023).
Finally, my findings document high yet varying levels of racial disparity across different levels and stages of the criminal justice system. Results showing that racial disparity is much lower at earlier stages in processing and at the subfelony level, when potential penalties are generally less severe, suggest that Black men are increasingly disadvantaged across successive stages of case processing and when the stakes (i.e., potential punishments) are highest. This finding aligns with evidence from studies of case processing (Kutateladze et al. 2014; Omori and Petersen 2020) and prosecutorial discretion (Tuttle 2019), which together suggest that decision-making and procedural inequalities within the criminal legal system, rather than race differences in crime, explain a substantial amount of the racial inequality that accumulates after the point of arrest.
I also find evidence that racial disparity has increased across cohorts at the subfelony level and declined substantially only in felony contact for women. Racial disparity is so pronounced among the 1991‒1995 cohort that Black women share a nearly equivalent risk of arrest as same-aged White men. This is due, in part, to stark declines in the subfelony criminalization of White men that were not matched by comparable declines among Black women.
The persistence of racially unequal criminalization should not be interpreted as evidence of racial differences in crime. Studies of self-reported rates of lawbreaking actually show relatively little racial difference in lawbreaking behavior, and studies routinely show that illicit drug use, which is among the most common criminal offenses, is more prevalent among White and Hispanic people than among Black people (Keyes et al. 2017; Sohoni et al. 2021). Rather, these numbers tell us about inequalities in who is formally detained by law enforcement and marked with a criminal record.
Limitations
This study has several limitations. The CCH data do not provide an exhaustive record of all incidents of criminalization in Texas; for example, the data do not include juvenile arrests occurring between the ages of 10 and 16, reliable measures of pretrial detention, or records of Class C misdemeanors. However, the omission of juvenile contact from the data may not greatly impact estimates. Because juvenile justice involvement is a strong predictor of later justice contact (McGlynn-Wright et al. 2022), many individuals arrested as children are likely picked up in the CCH data for a subsequent arrest by age 24. The close alignment of estimates of men's arrest prevalence for the 1981‒1985 cohort with estimates from Brame and colleagues (2014) for the 1980‒1984 cohort surveyed in the NLSY97 lends further validity to the findings.
The lack of data on pretrial detention and Class C misdemeanors, by contrast, does mean that many instances of criminalization are not captured by this study. My estimates of jail incarceration reflect only postconviction misdemeanor sentences, yet research has found that upward of 70% of people held in jail have not yet been convicted of a crime (Aiken 2017). Thus, I likely underestimate the true prevalence of jail incarceration. Future research on jail detention is greatly needed. Furthermore, Class C misdemeanors, which include all routine traffic violations and petty offenses in Texas, are also a source of criminalization absent from the data. Although these “fine-only” offenses are not punished with jail time and do not create state records, they do carry fines and fees that are shown to have racially unequal impacts (Bing et al. 2022) and can lead to warrants, arrests, and jail incarceration for those unable to pay (Bing et al. 2024; Pager et al. 2022).
Additionally, while the measurement of Hispanic ethnicity through linked records represents a significant improvement over conventional methods, it is still imperfect. Differences in coverage dates across data sources meant that ethnicity coding was available from only one source—identity and citizenship records in the CCH—for a disproportionate number of 1971‒1975 cohort members. Results showed that risk of contact was lower for Hispanics than Whites in the 1971‒1975 cohort and was higher for Hispanics than Whites in later cohorts, which may be an artifact of some Hispanic individuals being misidentified as White in the earliest cohort. In sensitivity analyses, shown in the online appendix, I adjusted race-specific estimates for Whites and Hispanics by assuming that the proportion of matched White adults from each cohort identified as Hispanic in the matched data is representative of the proportion of unmatched White adults who are Hispanic from the same cohort. These supplemental analyses show that trends are robust to alternative coding specifications and only mildly affect estimates for the 1970s cohorts of White and Hispanic men.
Conclusion
This study demonstrates that criminal legal contact is a common life event for an even larger and more diverse share of the population than previous studies have indicated. The macrosocial and public policy changes thought responsible for the declining felony incarceration rate have not had the same effect on subfelony criminalization, at least not for all segments of the U.S. population. Thus, although the risk of imprisonment is declining for many, the high and racially unequal exposure to arrest, prosecution, and conviction by early adulthood found here indicates that criminalization falling short of imprisonment may play an important role in shaping inequality in the life course trajectories of post-prison-boom generations. To be sure, Black and White men born after 1990 face much lower risks of any type of criminalization than those born even five years earlier, yet a staggering proportion remain at risk of arrest and misdemeanor prosecution by early adulthood. The consequences of subfelony criminalization are yet to be fully uncovered, but the high and unequal prevalence documented here makes clear that a research agenda focused on the consequences of arrest, jail, misdemeanors, and prosecution ending without conviction, particularly among women, is urgently needed.
Acknowledgments
I am grateful to Becky Pettit, Bruce Western, Sarah Brayne, Chantal Hailey, and the members of the Crime, Law, and Deviance working group at The University of Texas at Austin and the Justice Lab at Columbia University, as well as the anonymous reviewers and editors at Demography, for their generous feedback on previous drafts of this manuscript. I also acknowledge the Education Research Center at The University of Texas at Austin for the resources and support to conduct this research. The conclusions of this research do not necessarily reflect the opinion or official position of the Texas Education Research Center, the Texas Education Agency, the Texas Higher Education Coordinating Board, the Texas Workforce Commission, or the State of Texas.
This research has received support from the Horowitz Foundation for Social Policy and the Training Program in Population Studies (grant T32HD007081) awarded to the Population Research Center at The University of Texas at Austin by the Eunice Kennedy Shriver National Institute of Child Health and Human Development. The content is solely the responsibility of the author.
Notes
In this article, I use the term Hispanic because polls show that, at least in Texas, Hispanic is the most commonly preferred term among those who self-identify as such (Lopez 2013; Villalba and Jones 2022).
The data exclude Class C misdemeanors, which are the lowest offense class in Texas. Class C misdemeanors include routine traffic violations, municipal infractions, and code violations that typically do not result in an arrest and for which punishment is limited to a fine. In many U.S. states, offenses of this type are deemed noncriminal.
See https://www.dps.texas.gov/section/crime-records/computerized-criminal-history-cch-specifications.
Black Hispanic individuals make up such a small proportion of the Hispanic population in Texas that results cannot be influenced by this coding choice. Prevalence rates are not estimated for Asians or Native Americans because of their relatively small population sizes in Texas during the period.