Cumulative Risks of Multiple Criminal Justice Outcomes in New York City

Abstract

Previous research has provided estimates of the cumulative risk of felony conviction and imprisonment in the United States. These experiences are, however, also the rarest; most of what happens in the criminal justice system occurs at the level of the misdemeanor rather than the felony. This article addresses our limited understanding of the scope of subfelony justice by providing estimates of the cumulative risk of several lower-level arrest outcomes for one jurisdiction: New York City. Because of excess life table events contributed by nonresidents of New York City, estimates are likely upwardly biased relative to the true values. Nonetheless, they allow us to (1) assess the cumulative risk of misdemeanor conviction and jail sentences and (2) determine to what extent those who enter the world of subfelony justice are distinct from those with felony or imprisonment records.

Introduction

Demographic research on the American criminal justice system has focused on the most extreme forms of contact: felony conviction and imprisonment. Central to this body of research is a set of studies that employ life table techniques to demonstrate variations in the risk of own imprisonment (Pettit and Western 2004), parental or kin imprisonment (Chung and Hepburn 2018; Wildeman 2009), and felony conviction (Shannon et al. 2017). These studies lay bare the reach and racial inequalities of the criminal justice system.

Felony arrest, however, is not the most common form of criminal justice encounter (Kohler-Hausmann 2013, 2014, 2018; LaFountain et al. 2012; Natapoff 2012, 2015), nor is imprisonment the most common outcome. Although the literature provides insight into high-level criminal justice contacts and the most serious outcomes from those contacts, knowledge about subfelony justice is surprisingly limited. Notably lacking is an equivalent set of estimates that allow researchers to (1) assess the cumulative risk of misdemeanor conviction and jail sentences and (2) determine the extent to which those who enter the world of subfelony justice are distinct from those with felony or imprisonment records.

This article addresses these gaps in one jurisdiction: New York City. The choice of site is driven both by data availability and by New York City’s role as a pioneer in policing tactics emphasizing low-level enforcement. We employ demographic methods to estimate the cumulative risk of misdemeanor conviction, felony conviction, and receipt of a jail or prison sentence by sex, racial/ethnic group, and birth cohort. Notably, we are able to estimate the lifetime risk of misdemeanor conviction alone—without any prior or subsequent felony conviction—and demonstrate how that risk has changed as misdemeanor enforcement increased in New York City. This allows us to address the question of how low-level convictions extend beyond the populations touched by felony justice.

Subfelony Justice in the Era of Mass Imprisonment

The world of subfelony justice dwarfs that of felony justice. Natapoff (2012) suggested the metaphor of an iceberg, with only the well-studied outcomes of felony conviction and imprisonment visible above the surface. As with any iceberg, more is hidden below. For example, Stevenson and Mayson (2018) estimated that the arrest rate for misdemeanor offenses, nationally, was roughly 15 times that for violent felonies over the period 1995–2015. They found that misdemeanor case filings outnumbered felony case filings by a ratio of at least three to one.

Subfelony convictions bear penalties. The National Inventory of Collateral Consequences of Conviction documented 354 legal and regulatory sanctions triggered in New York State by different types of misdemeanor conviction (Justice Center 2018). Such sanctions can range from being barred from adopting or fostering a child to being denied tenancy in a public housing authority building to being ineligible for a barber shop owner’s license.¹ Even those arrested but not convicted of a misdemeanor can see their odds of future employment decline (Uggen et al. 2014), and Internet access has vastly expanded employers’ access to subfelony criminal justice records (Lageson 2016; Lageson and Maruna 2018).

In New York City, misdemeanor arrests have increased as felony arrests have fallen (Kohler-Hausmann 2018). New York City saw approximately 148,000 felony arrests in 1990, 113,000 in 2000, and 92,000 in 2010.² Over the same period, however, misdemeanor arrests increased dramatically: in 1990, there were approximately 118,000 misdemeanor arrests in the city; in 2000, that number stood at 224,000; and by 2010, it exceeded 251,000, representing an increase of 113 % from 1990 levels. Consistent with national trends—as well as with New York City–specific patterns in stop, question, and frisk (Fagan et al. 2010; Gelman et al. 2007)—the increase in low-level enforcement has been largely experienced by the city’s black and Hispanic population.

How do these arrest rates—and their uneven racial distribution—translate into long-term criminal justice marks (i.e., convictions and sentences)? Aggregate arrest numbers cannot tell us about the incidence of criminal conviction, much less the overlap between populations touched by different levels of the criminal justice system. The cumulative risk of misdemeanor conviction is a valuable measure of how subfelony enforcement translates into legal outcomes—and thus permanent status markers—and the scope of its effects within subpopulations.

A well-developed body of demographic research has highlighted the distinctions between cross-sectional rates of a given criminal justice event (e.g., imprisonment) and the cumulative risk of the same. The former provides information on point-in-time exposure; the latter allows us to measure the share of the population ever to bear the associated burdens. Cumulative risk measures the likelihood of transition to a given level of contact (treated as an absorbing state) and does not account for repeated exposure; after the first death (or conviction, or prison sentence, and so on), there is no other. This literature has documented variations in lifetime risk of imprisonment by cohort, race, education, and geography (Muller and Wildeman 2016; Pettit and Western 2004; Western and Wildeman 2009). It has also looked beyond imprisonment to the larger population that has a felony conviction or has been sentenced to supervised release in the form of probation without ever going to prison (Jacobs 2015; Manza and Uggen 2006; Pettit 2012; Phelps 2017; Shannon et al. 2017).

Data limitations have been a serious constraint to research on misdemeanors. Researchers are barely able to produce national estimates of the overall number of misdemeanor case filings (Stevenson and Mayson 2018). Studies of imprisonment and felony conviction risk rely on data for which no clear analog exists in the world of subfelony justice, at least nationally. Administrative data, however, offer a unique opportunity to construct estimates of cumulative risk for New York City.

Data and Methods

Our goal in this article is to demonstrate (1) how cumulative risk of two key lower-level criminal justice outcomes—misdemeanor conviction and receipt of a jail sentence—vary by cohort, race/ethnicity, and sex; and (2) how they compare to risk of felony conviction and imprisonment. To do so, we follow the life table methods developed by Bonczar and Beck (1997) and later elaborated by Pettit and Western (2004) and Wildeman (2009). These authors relied on data from multiple waves of the Survey of Inmates in State and Federal Correctional Facilities. No similar surveys provide sufficient data about misdemeanor convictions to extend analyses to subfelony justice. In the absence of such surveys, we take advantage of administrative data.

Data come from the New York State Department of Criminal Justice Services (DCJS), the agency responsible for collecting and maintaining records from local courts and law enforcement agencies and generating criminal histories.³ The data contain the entire population of individuals convicted of a misdemeanor or felony in New York City in seven sample years: 1980, 1985, 1990, 1995, 2000, 2005, and 2010. Following their first criminal conviction, the DCJS assigns individuals a unique New York State ID (NYSID), which is linked to their fingerprints. Based on NYSID, the DCJS provided full arrest and conviction histories for individuals convicted in the seven sample years. We observe all instances of probation, jail, prison, and parole both prior and subsequent to the event that led to sample inclusion. Because the DCJS can maintain a stable NYSID only following first conviction (or one of a small set of other events), these sequences are sometimes left-censored (i.e., we may fail to observe arrests preceding first conviction) but are complete from first conviction onward. They contain some demographic data about each defendant, including age, race/ethnicity, and sex.

The life table methods noted earlier rely on four types of counts, disaggregated by year, age, race/ethnicity, and sex: (1) first criminal justice events, (2) population with a previous criminal justice event, (3) population counts, and (4) mortality counts. DCJS data effectively provide us with the first sort of count, and from those we estimate the second. Because an individual’s NYSID is stable upon first conviction, we can determine for every member of the sample in each sampled year whether that conviction was a first-time or a higher-order event. Based on these data, we produce counts, aggregated by sex, race/ethnicity, and five-year age groups, of first events in multiple categories. (The full set of categories and the underlying logic are provided in the online appendix.) Because of data limitations, we include non-Hispanic white, non-Hispanic black, and Hispanic individuals, but we exclude New Yorkers of all other racial/ethnic groups. We adjust these tallies of events to account for sex-, race/ethnicity–, age-, and period-specific net commuting patterns in New York City. (The process and logic are described in the online appendix.) To estimate the population with a previous event, we aggregate counts of first-time events from previous years for the applicable age category. Decennial census data from IPUMS (Ruggles et al. 2017) provide us with population counts for New York City. We derive population counts for noncensus years by interpolating adjacent age groups across census years making an assumption of stable growth or decline over the period. Death counts are taken from mortality reports periodically issued by the Office of Epidemiology and Statistics in the New York City Department of Health.

Note that our results represent estimates of the cumulative risks that New Yorkers face of experiencing several criminal justice outcomes and that these estimates are likely to be systematically upwardly biased. Some convictions and jail/prison sentences that we record will be to nonresidents of New York City, and some residents of New York City will receive convictions or jail/prison sentences outside New York City that we do not observe in our data.⁴ The former represent “excess” life table events in many cases whereas the latter are “missing” events only for those individuals who never experience that particular event in New York City. Even though we expect that the latter is a nonzero number, particularly given the specificity of some of our outcomes, the former is likely larger. However, in the absence of data that would allow us to adjust our counts for sex-, race/ethnicity–, age-, and period-specific relative risk of these two events, we operate under an assumption that they offset each other. Further research should attempt to assess the validity of this assumption and provide more reasonable bounds.

In recognition that individuals may experience multiple sorts of criminal justice events over their life course and to better understand the overlap between populations entangled in the worlds of felony and subfelony justice, we provide multiple estimates of the cumulative risk of misdemeanor conviction. In increasing order of specificity, these estimates are as follows:

1. Cumulative risk of misdemeanor conviction, including those with felony convictions. How common is it for New Yorkers to be convicted of a misdemeanor? This estimate provides the overall scale of subfelony conviction and is not exclusive of felony conviction.

2. Cumulative risk of misdemeanor conviction, excluding those with felony conviction. How common is it for New Yorkers to be convicted of only a misdemeanor? This estimate allows us to see how individuals experiencing misdemeanor conviction are or are not separate from those experiencing felony conviction.

3. Cumulative risk of misdemeanor conviction, excluding both those with felony convictions and those with misdemeanor convictions from felony arrest. How many New Yorkers have misdemeanor conviction records from only misdemeanor arrest? This estimate allows us to assess the extent to which risk of misdemeanor conviction results from low-level arrests as opposed to charge reduction from felony arrests.

Later categories represent subsets of the prior categories. For example, the second group (those with a misdemeanor conviction but no felony conviction) represents a subset of the first (the total population of individuals with a misdemeanor conviction). In that case, the cumulative risk of holding both a misdemeanor and a felony conviction can be calculated by subtracting estimate 2 from estimate 1.

Results

Table 1 presents the cumulative risks to age 40–44, by race/ethnicity, sex, and cohort, of misdemeanor and felony conviction.

As an example of interpretation, we find that white male New Yorkers born in 1961–1965 had an 8.31 % chance of being convicted of a misdemeanor at least once by age 40–44 (upper-left entry in Table 1). Reading down the first column, we find that 5.64 % of white men in this cohort held a misdemeanor conviction in the absence of a felony conviction, and 2.09 % held only a misdemeanor conviction from a misdemeanor arrest. Put another way, two-thirds (5.96 / 8.86 = .68) of these white men with a misdemeanor conviction held only a misdemeanor conviction, and one-quarter (2.26 / 8.86 = .25) held only a misdemeanor conviction from a misdemeanor arrest. Their black peers had more than four times the risk of a misdemeanor conviction: 36.71 % of black male New Yorkers of this cohort had a misdemeanor conviction by age 40–44. Proportionately fewer black males had a misdemeanor conviction in the absence of a felony conviction (14.42 / 36.71 = .39) or from a misdemeanor conviction only (3.92 / 36.71 = .11).

Figure 1 offers a visual representation of the misdemeanor risk estimates from Table 1. Each panel presents a race- and sex-specific stacked bar plot; each bar shows the cumulative risk of misdemeanor conviction to age 40–44 by birth cohort. The bar is decomposed into risk contributed by those who experienced both misdemeanor and felony conviction (in white on the bottom), those who experienced misdemeanor conviction from felony arrest (in gray in the middle), and those who experienced misdemeanor conviction only from misdemeanor arrest (in black on the top).

Four patterns are noteworthy. First, as the prior literature leads us to expect, cumulative risks of all conviction outcomes were higher for black New Yorkers than whites and higher for men than for women. Racial disparities were much larger for cumulative risks of felony conviction than for misdemeanor conviction. The ratio of black risk to white risk was, within conviction type, relatively stable across cohorts: black men faced approximately four to five times the lifetime risk of a misdemeanor conviction (not exclusive of felony conviction) and seven to nine times the risk of a felony conviction as their white peers. The racial disparities were the lowest for misdemeanor conviction alone resulting from misdemeanor arrest. These risk ratios were similar, if somewhat lower, for women.

Second, Hispanics—who are often omitted from analysis of this sort—experienced cumulative risks of each of these conviction events that fell between those observed for whites and blacks, albeit closer to the latter than the former. For instance, among male New Yorkers born 1966–1970, 6.7 % of white men experienced a misdemeanor conviction by their mid-40s compared with 21.2 % of Hispanic men and 29.8 % of black men. Put another way, Hispanic men in this cohort had more than three times the risk of this outcome (relative to their white peers), whereas black men had more than four times the risk.

Third, among those at risk of holding a misdemeanor conviction, a much larger percentage of white than black or Hispanic New Yorkers held only a misdemeanor conviction (in the absence of a felony conviction). This is reflected in Fig. 1, which shows that the proportion of overall misdemeanor conviction risk contributed by those who receive both misdemeanor and felony convictions (in white on the bottom) was much larger in the black and Hispanic panels.

Finally, we observe a pattern within race and conviction type of declining cumulative risk across cohorts. For example, black men’s cumulative risk of a misdemeanor conviction dropped from 36.7 % in the 1961–1965 cohort to 26.1 % in the 1971–1975 cohort and further downward to 23.9 % in the 1981–1985 cohort. Comparing the 1981–1985 cohort with the 1961–1965 cohort, risks of each type of conviction declined by between 28 % and 53 % for white men, between 18 % and 48 % for black men, and between 28 % and 50 % for Hispanic men. Declines in cumulative risk were higher among women across the board.

These patterns, cumulatively, yield an unexpected conclusion. As the absolute number of black and Hispanic misdemeanor arrests increased over several decades—and as black-to-white and Hispanic-to-white arrest ratios rose—neither the cumulative risk of misdemeanor conviction from misdemeanor arrest nor the respective racial disparities relative to whites increased markedly for members of these minority groups. This suggests that the important site to study racial disparities in the subfelony world may be prevalence and frequency of arrest, especially arrests that do not lead to a criminal conviction.

Table 2 presents an equivalent set of estimates for cumulative risk of receipt of a jail or prison sentence and for risk of receiving only a jail sentence. Among white men born in 1961–1965, 3.67 % were sent to jail or prison by age 40–44; the majority (2.09 / 3.67 = .57) were sent only to jail (i.e., never received a prison sentence). A much smaller percentage of minority group members (relative to their white peers) received only a jail sentence. Across cohorts, roughly 60 % of white men who were sent to jail or prison by their mid-40s were sent only to jail; for black and Hispanic men, the average was approximately 40 %.

In comparing Tables 1 and 2, note the stable pattern whereby cumulative risk of misdemeanor conviction is greater than that of risk of a jail or prison sentence, which is in turn greater than that of felony conviction. Misdemeanor convictions rarely lead to imprisonment, but felony convictions can lead to either prison or jail; neither sort of conviction by definition leads to a carceral sentence. As such, we would expect risk of prison or jail to be somewhat higher than that of felony conviction—because it takes into account jail sentences stemming from misdemeanors—and somewhat lower than that of misdemeanor conviction.

Conclusion

This article presents estimates of the cumulative risk of misdemeanor and felony conviction and receipt of a jail or prison sentence for New York City residents, by race/ethnicity, sex, and birth cohort. These estimates allow us to demonstrate, for America’s largest jurisdiction, the cumulative risk of subfelony legal outcomes. We find that across cohorts, 4 % to 9 % of white male New Yorkers, 24 % to 37 % of black male New Yorkers, and 15 % to 26 % of Hispanic male New Yorkers have ever been convicted of a misdemeanor. The cumulative risk of misdemeanor conviction is much smaller for female New Yorkers.

These estimates allow us to assess the extent to which the populations affected by felony and subfelony justice are distinct. If the penal mechanisms triggered in the subfelony world and the populations that flow through it are meaningfully distinct from those of the felony world, we may need to revise our understanding of how the criminal justice system functions as an inequality-transmitting institution and expand our estimates of the scope of its operations. We find that the cumulative risk of misdemeanor conviction is, as expected and in all cases, larger than that of felony conviction. Depending on race/ethnicity, sex, and birth cohort, between 34 % and 83 % of New Yorkers convicted of a misdemeanor are never convicted of a felony. These ratios—risk of misdemeanor conviction without a felony conviction to overall risk of misdemeanor conviction—are lowest for black and Hispanic male New Yorkers. We see a similar pattern with regard to carceral sentences: among those sentenced to jail or prison, white New Yorkers (relative to their minority peers) have been much more likely to be only sentenced to jail rather than prison.

We find a pattern of declining cumulative risk of each criminal justice outcome over time: those born more recently are at lower risk of misdemeanor conviction, felony conviction, and prison or jail sentences than members of previous cohorts. This finding is consistent with observed national-level and New York State–specific declines in imprisonment rates and conviction rates, including drops in misdemeanor filing rates and drug-related prison admissions (Mauer and Ghandnoosh 2015; Pfaff 2017; Stevenson and Mayson 2018). Nonetheless, work remains to square these finding with the previously cited increases in misdemeanor arrest rates in New York City. A number of questions merit future study. What roles do declining crime rates and lowered thresholds of suspicion for low-level arrest play in these shifts? What link can be established between our findings and the “broken windows” policing model (a collection of tactics that intentionally intensified subfelony enforcement in the city from 1994 onward)? It is important to keep in mind that the estimates of cumulative risk that we present here allow us to assess rates of transition into the absorbing states of “ever convicted” or “ever sentenced.” The lived experience of those with the mark of a misdemeanor conviction or a jail sentence may vary widely and in ways that such estimates cannot capture. Future work should measure processual and iterative encounters with the criminal justice apparatus over the life course, attempting to better describe changes to both misdemeanor arrest frequency and conviction rates.

We caution again that the estimates presented here are likely biased upward because the number of life table events contributed by nonresidents of New York City likely exceeds the number of missed events (cases in which a resident experiences a given event elsewhere but never in the city). A large number of nonresidents flow through New York City each year; our adjustment for net commuting patterns only partially corrects for life table events likely contributed by these individuals. In the online appendix, we discuss how these patterns may affect our results. However, without additional data about the frequency of either excess or missed life table events by age, race/ethnicity, sex, period, or event type, we are unable to systematically adjust our estimates.

This article examines the scope of subfelony justice in New York City. However, the enforcement and even the definitions of misdemeanor justice vary considerably across the country (Stevenson and Mayson 2018). We hope that further work will provide comparable estimates for other jurisdictions. We also strongly encourage the collection and harmonization of more data on subfelony justice.

Acknowledgments

We thank Pil Chung, Christopher Muller, Jennifer Jennings, Michael Kurlaender, Sandra Mayson, Megan Stevenson, and Katie Beth White for comments. Kohler-Hausmann was supported by the Russel Sage Foundation and the Oscar M. Ruebhausen Fund at Yale Law School.

Notes

References

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.