Does Quality-of-Life Policing Widen the Net? A Partial Analysis

Abstract

In the 1990s, the New York City Police Department instituted a policy of arresting individuals for less serious offenses that impinge on the city’s quality of life (QOL). Critics contend that QOL policing widened the net for arrest, especially among minorities. Alternatively, QOL policing could have created additional opportunities for arresting individuals from the same populations that tend to incur arrests for more serious offenses. This article reports on a comparison of New York City QOL and serious arrestees interviewed in 1999 that provides partial support for this alternative hypothesis; the two groups were similar regarding prior arrests, participation in QOL offenses, and demographic composition. Of note, blacks and Hispanics comprised close to 90% of both arrest populations. These findings suggest that QOL policing does not necessarily widen the net for arrest.

New York City (NYC) experienced a renaissance in the 1990s. The streets became cleaner, the homeless were less visible, the economy was booming, real estate values surged, tourism increased, and serious crime was down, especially violent crime. Many credit aggressive policing by the NYC Police Department (NYPD) for the decline in crime and disorder (Kelling & Sousa, 2001; Silverman, 1999), especially former NYC Mayor Rudolph Guiliani (Guiliani with Kurson, 2002) and former Police Commissioner William Bratton (Bratton with Knobler, 1998). In a comprehensive review, Eck and Maguire (2000) concluded that there was strong evidence that numerous policing initiatives had an effect but that it was not yet possible to know how much of the decline to attribute to each or to other historical factors including the end of the crack epidemic, a strong economy, demographic changes, and a decline in handgun use, particularly among youths.

Part of the NYPD’s aggressive policing strategy was to arrest individuals for less serious but highly visible offenses (such as farebeating¹ or smoking marijuana in public) that detract from the quality of life (QOL) in the city.² In the past, police may have ignored these types of minor misbehaviors. Alternatively, police might have asked individuals to desist and possibly issued a desk appearance ticket requiring the offender to subsequently appear in court where they might have been fined. Of note, Mayor Michael Bloomberg (who entered office in 2002) and Police Commissioner Raymond Kelly have maintained QOL policing as an essential component of the city’s policing strategy (New York Times, 2001a,b).

Many have contended that QOL policing “widens the net” for arrest and disproportionately targets blacks and Hispanics (Amnesty International, 1996; Harcourt, 2001; McArdle & Erzen, 2001; Spitzer, 1999). This article presents a partial test of this concern with data collected under the ADAM (Arrestee Drug Abuse Monitoring program) NYC Policing Study (hereafter the Policing Study), a unique and detailed dataset (see Johnson, Taylor, Golub, & Eterno, 2001). The analysis compares the demographic characteristics (including race/ethnicity), prior offending, participation in QOL offenses, and mainstream status attainment of QOL and serious arrestees³ interviewed in NYC in 1999. If substantial net widening had occurred, then the differences in characteristics of these populations should reflect the populations most affected by the increased law enforcement. Alternatively, a similarity between QOL and serious arrestees would reflect a lack of net widening.

These data do not allow for an ideal evaluation of net widening for two key reasons: 1) comparable data were not collected back in the early 1990s before the broader implementation of QOL policing that could serve as a baseline for analysis; and 2) the police do not explicitly identify who was a QOL arrestee. Despite these limitations, the careful cross-sectional analysis presented is highly suggestive of the extent of net widening from QOL policing in NYC. Moreover, the opportunity to collect the kind of data necessary to definitively evaluate NYC’s experience of QOL policing in the 1990s has already passed. Accordingly, we submit this partial analysis of whether QOL policing in NYC resulted in net widening to advance our knowledge of an important aspect of a widely discussed policing innovation.

The remainder of this introduction reviews the motivation for QOL policing, the evidence supporting this position, and concerns raised about the program. Following the presentation and discussion of findings, the conclusion considers the potential demonstrated by this analysis for the use of the ADAM program data in evaluating policing initiatives.⁴

Prior Research on QOL Policing

The broken windows perspective is frequently used to justify QOL policy (Kelling & Coles, 1996; Skogan, 1990). In a widely cited article published in Atlantic Monthly, Wilson and Kelling (1982) argued that physical decay (such as a broken window) and incivility can initiate a spiral of disorder and decline. Public misbehavior offends the community’s sensibilities, instills a fear of crime, creates a sense of disorder, leads law-abiding residents and visitors to withdraw from public spaces, sends a signal that deviant behavior is tolerated, and ultimately creates an environment conducive to crime.

However, the limited scientific evidence pertaining to the fixing broken windows perspective is mixed so far. In a widely cited effort to link crime and disorder, Skogan (1990) analyzed responses from 13,000 interviews conducted in 40 neighborhoods from 6 cities across the United States. He found that robbery victimization was higher in neighborhoods characterized by disorder, even after controlling for poverty, residential stability, and racial composition using regression. In a reanalysis of the data, however, Harcourt (2001) observed that the robbery-disorder relationship disappeared after excluding Newark (which had the highest incidence of both robbery and disorder) and that the crime-disorder relationship did not hold for other crime types, including burglary, assault, rape, and purse snatching/pickpocketing. Sampson and Raudenbush (1999) shed further doubt on the crime-disorder relationship. Their analysis of over 196 Chicago neighborhoods found that the association between crime and disorder disappeared after controlling for concentrated poverty and collective efficacy (a neighborhood-level measure of cohesion, informal social control, and optimism).

Three recent field studies provide direct tests of QOL policing. Braga et al. (1999) found significantly larger reductions in both disorder and crime in 12 Jersey City high crime areas that received increased attention (with aggressive order maintenance policing a central component of the treatment) compared with 12 matched locations that did not receive the experimental treatment. However, two other studies—one in Chandler, Arizona, (Katz, Webb & Schaefer, 2001) and one in a larger Midwestern city (Novak, Hartman, Holsinger, & Turner, 1999)—found no significant impact.

A secondary justification for QOL policing is that it can sometimes help resolve serious crimes (Silverman, 1999). QOL policing focuses on minor offenses but not necessarily on minor offenders, especially to the extent that the same people commit both QOL and more serious offenses. Advocates like to point to the case of John Royster, Jr. who in 1996 was apprehended for farebeating. A fingerprint match placed him at the scene of a recent murder. He was subsequently linked to four other unsolved cases and eventually convicted of homicide.

Net Widening and Racial Bias Allegations

As index crimes dropped from 1993 to 1995, civilian complaints against the NYPD for abuse of authority doubled (Eterno, 2001). Several high profile cases involving minorities (Anthony Biaz, Abner Louima, Amadou Diallo, and Patrick Dorismond) raised much attention(see McArdle & Erzen, 2001, for a review). Amnesty International (1996) declared an urgent need for the NYPD to address persistent and extensive problems of police brutality and excessive use of force, especially against African-Americans and Latinos. In a study of police stops occurring in 1998 and part of 1999 (in which civilians were temporarily detained, questioned, and sometimes searched), the New York State Attorney General’s Office found substantial racial disproportionality (Spitzer, 1999). Whereas blacks comprised 26% of the city’s population, they accounted for 51% of all stops. Hispanics comprised 24% of the population but accounted for 33% of all stops. In strong contrast, whites comprised 43% of the population yet accounted for only 13% of all stops. There was evidence of racial disproportionality even after controlling for variation in arrest rates by race using regression models. In response to the findings, the NYPD countered that comparing individuals stopped to residents by race was inappropriate (Flynn, 1999; NYPD, 1999). They contended that the representation of blacks and Hispanics among those stopped (85%) was consistent with the rate at which victims of violent crimes described their perpetrators as black or Hispanic (89%).

QOL policing may not have been the central cause of complaints against the NYPD. The NYPD operated several aggressive policing programs in the 1990s. Indeed, in some cases programs were purposely designed to overlap, which further complicates attempts to identify the impact of any particular program. The Getting Guns Off the Streets initiative aggressively pursued illegal firearms using a variety of investigative techniques including stops (NYPD, 1994). From 1994 to 1997, the NYPD confiscated 56,081 guns (Office of Juvenile Justice and Delinquency Prevention, 1999). Spitzer (1999, p. 53) noted that patrol officers frequently reported QOL infractions as the probable cause justifying a stop. In the 1990s, the NYPD continued its use of Tactical Narcotics Teams (TNT) to close down drug markets through sweeps in which a large force of officers search everyone in a neighborhood to comprehensively remove all drug dealers and users (Belenko, 1993; Curtis 1998). NYC also authorized and encouraged numerous Business Improvement Districts (BIDs, not administered by NYPD) to serve as neighborhood-based chambers of commerce and quasi-governmental coalitions (Barr, 2001). These agencies often hired their own private security officers to maintain order.

Unarguably, QOL policing increased the range of arrestable offenses. It is less obvious, however, whether QOL policing resulted in a wider variety of individuals sustaining arrest, people who would have otherwise been unlikely to be sanctioned. To the extent that the same individuals tend to commit both serious crimes and less serious offenses, QOL policing might simply increase the range of violations charged or the frequency with which some individuals are arrested.

Methods

This paper compares 195 QOL and 265 serious arrestees according to demographic characteristics, official New York State (NYS) criminal histories, self-reports of involvement with various QOL behaviors, and recent drug use as detected by urinalysis. We chose to operationally define serious arrestees as those charged with a drug or index felony according to NYS law. Identifying probable QOL arrestees was more problematic (Table 1 presents the result). QOL policing is better characterized as a procedure than a set of statutes. Kelling and Coles (1996) describe how the NYPD often targeted a specific behavior (e.g., sleeping on subways) and applied whatever statute they could. In this manner, some individuals charged with a given offense might be QOL arrestees and others might not.

Table 1.

Operational Definitions of QOL and Serious Arrestees (ADAM-NYC Policing Study)

Arrest Charge	Counta
Quality-of-life arrestees.........................	195
Farebeating, misdemeanor	47
Trespassing, misdemeanor	80
Marijuana possession, misdemeanor	68
Charges not i-ncluded in this study.........	432
Drug possession (7th deg), misdemeanor	98
Prostitution, misdemeanor	49
Assault (3rd degree), misdemeanor	43
Petit larceny, misdemeanor	44
All other less frequent charges	198
Serious arrestees.................	265
Drug possession, felony	42
Drug sale, felony	78
Robbery, felony	39
Burglary, felony	14
Grand larceny, felony	36
Assault, felony	53
Rape/sexual assault, felony	3

^aUnweighted.

For this analysis, we chose to operationalize QOL arrestees as those respondents whose most serious arrest charge is for farebeating, trespassing, or misdemeanor possession of marijuana (hereafter MJ misdemeanor). By limiting the list, we attempted to identify some less serious arrests that were most likely the result of QOL policing, and not other policing activities. The three offenses selected have been discussed in the literature on QOL policing, are common among ADAM-NYC arrestees, carry minimal sanctions, result from highly visible behaviors, and create a sense of disorder.

This list of QOL offenses is clearly incomplete. It is far less comprehensive than the list of QOL offenses included in the Policing Study survey (see Table 4) that identifies such archetypical QOL offenses as aggressive panhandling and drinking in public. We wanted to include only those less serious offenses that were being broadly enforced by the NYPD in 1999. Aggressive panhandling, drinking in public, and other offenses not individually listed in Table 1 had fewer than 40 cases each; they are counted in the 198 less frequent charges not included in this study, along with less serious charges that are not QOL offenses (e.g., possession of stolen property). As of 1999, either few people were committing the QOL offenses other than farebeating, MJ misdemeanor, and trespassing or the NYPD was not actively targeting them for arrest at the time of the study. Of course, the mix of QOL charges may have differed substantially in previous years. The use of a single year of data is clearly a limitation to this study.

Table 4.

Variation in Past-Year QOL Offending Between QOL and Serious Arrestees

	Percent That Engaged in Behavior by Arrest Charge
	Individual QOL Offenses			ALL
	Farebeating	Trespassing	MJ misdemeanor	QOL	Serious
Unweighted sample sizea	(43)	(55)	(34)	(132)	(162)
Engaged in farebeating	82% b	45%	44%	56% b	41%
Engaged in trespassing	21	55 b	28	36 b	20
Smoked marijuana in public	26	39	71 b	45	41
Bought/carried marijuana in public	28	44	63 b	45	35
Sold marijuana in public	3	25	32 b	20	14
Drank alcohol in public	43	32	26	33	36
Smoked in nonsmoking areas	38	14	16	22	22
Urinated in public	35	42	16	32	32
Wrote graffiti	12	1	6	6	4
Littered	33	13	19	21	17
Failed to pick up after your dog	12 b	2	1	5	3
Failed to recycle garbage	7	0	6	4	4
Engaged in disorderly conduct	24	24	25	24	26
Made loud noises in public	12	12	19	14	19
Loitered w/o cause	15	37	16	24	25
Been in a gang	9	0	3	4	4
Hung out in street	43	39	47	42	50
Aggressive panhandling	12	3	3	6	5
Did squeegee work	0	0	3	1	1
Vended w/o license	10	0	6	5	9
Sold counterfeit video/tapes	6	2	6	5	6
Gambled in public	18	13	9	13	12
Engaged in prostitution	4	2	6	4	1
Bought/sold alcohol to minors	0	2	6	3	3
Bought/sold cigarettes to minors	3	5	9	6	6
Drove while intoxicated	9	6	9	8	8
Drove w/o a license/registration	21	20	19	20	22
Ignored red lights and stop signs	9	15	12	12	8
Sped	9	8	12	10	11
Drag raced	6	5	0	4	4
Talked on cell phone while driving	6	5	3	5	3
Violated traffic laws on bicycle	24	5	16	14	11
Jaywalked	33	24	16	25	22
Failed to cooperate w/police	7	18	6	11	16

Note. Estimates weighted to control for overrepresentation of females.

^aCalculations include only respondents with a prior record who disclosed it.

^bDiffers from serious at the a = .01 level.

MJ misdemeanor was included in the QOL designation but misdemeanor possession of a controlled substance (98 cases for drug possession, 7^th degree) was not. Personal use of marijuana in private in small quantities does not carry criminal charges in NYS; it is merely a violation. For many MJ misdemeanants, it would have been their use of marijuana in public that led to arrest. In contrast, a misdemeanor charge for possession of a controlled substance (such as crack or heroin) could have resulted from other narcotics law enforcement activities including TNT sweeps. The misdemeanor severity indicates that the quantity of drugs involved was too small to qualify as a felony. Three other less serious offenses (prostitution, assault, and petit larceny) were also excluded from the QOL arrest category because the charges may have resulted from routine policing activity.

The Policing Study

The Policing Study employed the ADAM program as a convenient and cost-effective platform for data collection (for details, see Johnson et al., 2001). Participation in the ADAM survey is voluntary. At most sites in 1999, more than 80% of arrestees approached agreed to participate (National Institute of Justice, 2000). The ADAM data are kept confidential and used for scientific research purposes only. In 1999, data collection occurred in all five boroughs of NYC.⁵ The Policing Study collected an initial pilot sample in the second quarter of 1999 and additional data in the second half of the year. Prior to the interview, potential participants gave their informed consent for the project to obtain their criminal histories from NYC and NYS agencies. Respondents were promised $15 after release for completing the questionnaire.

Starting in 1999, the ADAM program instituted procedures designed to obtain a representative sample of all arrestees (NIJ, 2000). To check the representativeness, the ADAM program compares the data to a complete census of all arrestees during the time that data collection occurs. Starting with the ADAM 2000 data, the program added sample weights to further assure the generalizability of estimates (Hunt & Rhodes, 2001). To facilitate comparisons across gender, the ADAM program purposely oversamples females, who usually account for about 15% of NYC arrestees. For this analysis, simple weights were employed so that females would constitute 15% of the weighted sample.

Criminal histories for all of the QOL and serious arrestees were obtained from the NYS Division of Criminal Justice Services (DCJS). Prior arrests outside of NYS, in the federal system, or before age 16 were not obtained. In an extensive review of the self-report literature, Thornberry and Krohn (2000) describe how official arrest records are heavily influenced by police priorities. Accordingly, this study interpreted respondents’ official records in their most literal sense, as a measure of previous contact with the criminal justice system and not as a measure of criminality.

A preliminary concern of the Policing Study was whether arrestee self-reports were accurate. Golub, Johnson, Taylor, & Liberty (2002) examined those Policing Study variables that could be confirmed with objective data (prior criminal arrest and recent drug use) to determine how response accuracy varied across questions and individuals. They found that arrestees were highly likely to disclose less stigmatized information such as whether they had ever been arrested before and whether they had used marijuana recently. This suggests that arrestees might be highly likely to disclose whether they engaged in QOL offenses. Arrestees were exceedingly unlikely to disclose a prior index offense, especially a violent index offense. More problematic, disclosure of a prior index offense was substantially higher among arrestees currently charged with an index offense. Hence, an analysis of self-reported index offenses across current offenses would clearly be biased by differential nondisclosure. Consequently, self-reports of serious offending were excluded from this analysis as potentially highly inaccurate.

Golub et al. (2002) also found that arrestees who disclosed having a prior arrest were substantially more likely to disclose other aspects of their criminal behavior. Moreover, arrestees who did not disclose that they had a prior record tended not to disclose other criminal activities. In a multivariate analysis, this preliminary disclosure proved to be the strongest and most consistent predictor of disclosure on other questions—stronger than demographic characteristics, disclosure of recent drug use, and interviewer’s assessment of the respondent’s veracity. Accordingly, it was decided to limit the analysis of self-reported involvement in QOL offending behavior to those arrestees who had a prior record and disclosed it. This included the majority of the serious (61%, 162 of 265) and QOL (68%, 132 of 195) arrestees. The analyses involving variables other than QOL self-reports were not limited to this subsample.

The urine samples collected by the ADAM program provide a particularly valid indicator of recent drug use (NIJ, 2000). The EMIT (Enzyme Multiplied Immunoassay Testing) urinalysis screen used by ADAM can usually detect cocaine (or crack) and opiates (such as heroin) within 48 to 72 hours of use. Marijuana consumption can be detected up to 7 days after last use for infrequent users and up to 30 days or longer for chronic users. The accuracy of the EMIT test depends upon the frequency of use, quantity used, potency, and the time between consumption and the ADAM interview. The EMIT screen used does not distinguish between modes of consumption (e.g., sniffing, smoking, or injecting); hence, this paper refers to detected cocaine/crack use. The EMIT screen also does not distinguish between various opiates. The most common opiate used on the streets of NYC is heroin, and so this paper refers to detected heroin use.

Results

QOL and serious arrestees were quite similar on most characteristics analyzed. On some characteristics, arrestees for the three individual QOL offenses (farebeating, trespassing, MJ misdemeanor) differed substantially. In particular, the average age varied substantially across arrest types. Moreover, the differences across arrest charges in various other attributes (e.g., being single) were often attributable to differences in age. Logistic regression (presented in the Appendix) was used to examine the extent to which any differences across arrest categories remained, after controlling for age. In view of the large number of comparisons presented in this section, the a = .01 level of statistical significance was used.

Demographics

QOL and serious arrestees were similar with regard to gender and race/ ethnicity composition (see Table 2). Most of the QOL and serious arrestees were male (both 88%). Almost two thirds of the QOL and serious arrestees were black (both 64%), and about a quarter of each was Hispanic (26% and 25%, respectively).⁶ White arrestees were uncommon among both QOL and serious arrestees (10% and 11%, respectively). On average, farebeaters and trespassers (both mean age of 35) tended to be significantly older than serious arrestees (mean age of 30) and MJ misdemeanants (mean age of 26). Likewise, the modal age category for farebeaters and trespassers was 40 and above as opposed to 30-39 for serious arrestees and 21-29 for MJ misdemeanants.

Table 2.

Variation in Demographic Characteristics Between QOL and Serious Arrestees

	Individual QOL Offenses			All
	Farebeating	Trespassing	MJ Misdemeanor	QOL Offenses	Serious
Male (%)	91	83	91	88	88
Female (%)	9	17	9	12	12
Black (%)	71	63	59	64	64
Hispanic (%)	26	25	28	26	25
White (%)	3	12	12	10	11
Mean age (years)	35 a	35 a	26 a	32	30
Age 18-20 (%)	6	6	27	14	17
Age 21-29 (%)	20	15	47 a	29	26
Age 30-39 (%)	26	36	15 a	25	35
Age 40+ (%)	48 a	43 a	10	32	21

Note. Estimates weighted to control for overrepresentation of females.

^aDiffers from serious at the a = .01 level.

Prior Arrests

Overall, QOL and serious arrestees’ prior criminal histories were serious, extensive, and remarkably similar (see Table 3). QOL and serious arrestees were comparably likely to have been previously arrested (82% and 81%, respectively). Of central interest to this analysis, roughly the same small percentage lacked a prior record of arrest (under 20%). Many had also been arrested in the last six months (both 45%). On average, farebeaters had more prior arrests than serious arrestees (17 vs. 9), though this difference was not statistically significant after controlling for age (see Appendix). In other words, older serious arrestees and older farebeaters tended to have about the same number of prior arrests.

Table 3.

Variation in New York State Official Criminal Histories Between QOL and Serious Arrestees

	Individual QOL Offenses			All
	Farebeating	Trespassing	MJ Misdemeanor	QOL Offenses	Serious
Arrest ever (%)	92	87	70	82	81
Arrest 6 mo. (%)	42	51	39	45	45
Mean # arrests	17 a	13	7	11	9
Index (%)	73	58	40 a	55	60
Drug offense (%)	80	71	60	69	64
Violent index (%)	21	22	22	22	26
Robbery (%)	51	38	25	37	33

Note. Estimates weighted to control for overrepresentation of females.

^aDiffers from serious at the a = .01 level.

A majority of QOL and serious arrestees had a prior record for an index offense (55% and 60%, respectively) and for a drug offense (69% and 64%, respectively). MJ misdemeanants were less likely than serious arrestees to have a prior arrest for an index offense (40% vs. 60%, respectively) but this difference was attributable to age (see Appendix). Substantially fewer QOL and serious arrestees had a prior arrest for a violent offense (22% and 26%, respectively; murder, rape or aggravated assault) or robbery (37% and 33%, respectively).

QOL Offending

Overall, QOL and serious arrestees were about as likely to self-report engaging in QOL behaviors in the last year (see Table 4). The greatest variation was associated with the criterion behaviors related to farebeating, trespassing, and MJ misdemeanors. Not surprisingly, farebeaters were significantly more likely than serious arrestees to report having engaged in farebeating (82% vs. 41%), even after controlling for age (see Appendix). The chance that all of the arrestees charged with farebeating had committed the behavior is quite high. The NYPD QOL Enforcement Options Reference Guide (as presented in Erzen, 2001) specifies that an officer must personally observe someone farebeating in order to take action. Accordingly, the 18% of farebeaters (based on their arrest charge) who did not report farebeating in the last year were likely hiding information.

Trespassers were significantly more likely than serious arrestees to report having engaged in trespassing (55% vs. 20%), even after controlling for age (see Appendix). Almost half (45%) of the trespassers reported that they had not engaged in trespassing in the past year. This may indicate either that arrestees were unwilling to report that they trespassed or that there was a difference of opinion. Arrestees may have maintained that they had a right to spend time in an abandoned building, whereas the police may have defined this activity as trespassing. The data do not provide sufficient information to test this hypothesis.

MJ misdemeanants were significantly more likely than serious arrestees to report having engaged in public marijuana-related behaviors: smoked in public (71% vs. 41%), bought or carried in public (63% vs. 35%), and sold in public (32% vs. 14%). Note that two of the three relationships (smoking and buying) were not statistically significant after controlling for age (see Appendix). MJ misdemeanants and trespassers were more likely to report that they sold marijuana in public than serious arrestees and farebeaters, even after controlling for age (see Appendix).

The percentages of QOL and serious arrestees who engaged in each of remaining QOL behaviors were comparable and mostly small; typically 5-20%. The most common QOL behaviors were hanging out in the street (42% and 50%), drinking alcohol in public (33% and 36%) and urinating in public (both 32%). Only one difference was statistically significant. Farebeaters were slightly more likely than serious arrestees to have failed to pick up after their dog (12% vs. 3%), although this relationship was not statistically significant after controlling for age (see Appendix).

Mainstream Status Attainment

QOL and serious arrestees were comparably unlikely to have gone to college, to be married, or to hold a job (see Table 5). Many QOL and serious arrestees reported having not completed a high school education (40% and 44%, respectively). MJ misdemeanants were significantly more likely to be single than serious arrestees (80% vs. 63%); however, this difference was attributable to age (see Appendix). Farebeaters (34%) and trespassers (30%) were significantly more likely than serious arrestees (16%) to report that welfare was their primary source of income. Again, however, this difference was attributable to age (see Appendix).

Table 5.

Variation in Education, Marriage, Employment, and Illicit Drug Use Between QOL and Serious Arrestees

	Individual QOL Offenses			All
	Farebeating	Trespassing	MJ misdemeanor	QOL Offenses	Serious
< H.S.	35%	44%	39%	40%	44%
H.S. grad	49	48	46	47	36
Any college	16	8	16	13	20
Single	66	67	80 a	72	63
Married	3	10	6	7 a	15
Sep/wid/div	14	8	1	7	10
Live w/someone	17	14	13	15	13
Full-time	19	18	36	25	32
Part-time	12	12	13	12	18
Welfare	34 a	30 a	13	25	16
Other legal	14	22	26	21	17
Illegal	21	19	12	17	18
Marijuana	40	39	90 a	58	48
Cocaine/crack	46	73 a	16 a	45	48
Heroin	19	21	2 a	13	12

Note. Estimates weighted to control for overrepresentation of females.

^aDiffers from serious at the a = .01 level.

Drug use differed substantially across arrest categories (see Table 5). Not surprisingly, urinalysis identified nearly all of the MJ misdemeanants (90%) as recent marijuana users in contrast to just under half of the serious arrestees (48%). MJ misdemeanants were also significantly less likely than serious arrestees to be detected as recent cocaine/crack (16% vs. 48%) and heroin users (2% vs. 12%). Trespassers were substantially more likely than serious arrestees to be detected as cocaine/crack users (73% vs. 48%). Multivariate analyses identified that much of this variation was associated with age (see Appendix). This finding is consistent with Golub and Johnson’s (1999, 2001) observation that marijuana was the drug of choice among youthful arrestees in the 1990s as opposed to crack or heroin, which were popular among previous birth cohorts. However even after controlling for age, MJ misdemeanants were still significantly more likely to be detected as recent marijuana users and less likely as cocaine/crack users.

Discussion

QOL policing by definition broadens the range of misbehaviors that can result in arrest. It was part of a number of NYPD initiatives in the 1990s that increased the intensity of law enforcement. This study used a uniquely detailed dataset to explore the extent to which QOL policing widened the net, bringing individuals into the criminal justice system who would not have been arrested otherwise. The characteristics of arrestees for several QOL offenses enforced in NYC in 1999 were compared to those of arrestees for index and drug felonies. To the extent that net widening occurred, the QOL arrestees should have had less extensive criminal records and perhaps stronger participation in conventional institutions.

Several important limitations to the study restrict the generalizability of findings. The analysis does not include baseline data collected before the NYPD implemented QOL policing. These data would have been used to establish the similarity between serious arrestees before and after implementation. They would have also been useful for establishing the extent to which individuals engaged in QOL offenses before the policy change and the extent to which they were or were not being arrested for these activities. Additionally, the evaluation examined only one year of data. The extent of net widening may have differed in NYC’s experience with QOL policing prior to 1999 (or after). In spite of these limitations, the findings do contribute new and advanced empirical insight into whether QOL policing causes net widening. Moreover, these limited findings highlight the potential for future use of ADAM to monitor and evaluate policing innovations, their impacts, and their collateral effects, perhaps more comprehensively than in this study.

The comparisons suggest that QOL policing did not systematically broaden the target population for arrest as practiced in NYC in 1999. On average, the primary difference between QOL and serious arrestees was only their current arrest charge, and not their demographic composition, prior record, past-year participation in a variety of QOL offenses, drug use, educational attainment, marital status, or employment. QOL arrestees did differ by age across offenses; farebeaters and trespassers tended to be older than serious arrestees and MJ misdemeanants tended to be younger. Thus, increased enforcement for any one of these offenses could systematically widen the net for older or younger arrestees. The analysis suggests, however, these arrestees for QOL offenses would be similar to comparably aged more serious arrestees with regard to their demographic composition, prior record, and other characteristics.

The similarity between serious and QOL arrestees may reflect either a lack of crime specialization among offenders or, alternatively, the impact of police discretion in arrest. Prior empirical research has found some evidence of crime specialization, especially among adult arrestees (Blumstein, Cohen, Das, & Moitra, 1988), but even more evidence of a lack of specialization (Farrington, Snyder, & Finnegan, 1988; Klein, 1984; Mazerolle, Brame, Paternoster, Piquero, & Dean, 2000). Thus, QOL policing might simply provide additional opportunities to arrest members of a population that engage in serious crimes, less serious crimes, and QOL misbehaviors.

On the other hand, the lack of observed net widening identified in this analysis could have resulted from the manner in which the NYPD performed QOL policing in 1999. The NYPD sought to send a message to serious criminal offenders. This focus was possibly reflected in choice of locations for patrols and in discretionary use of arrest. In this manner, arresting similar individuals for QOL offenses and serious crimes was part of NYPD’s larger strategy. The results of this analysis partially confirm the general overall success of this element to their approach. Further research is needed to establish the individual-level results of law enforcement efforts, whether offenders got the message, and whether this led to individual behavioral changes as well as crime reduction.

A major concern regarding QOL policing is whether it disproportionately targets minorities. This study found that most (90%) of the QOL arrestees in NYC in 1999 were black or Hispanic. However, most (89%) of the serious arrestees during this same time period were also black or Hispanic. This suggests that QOL policing did not expand the targeting of minorities for arrest, nor did it reduce it. NYPD’s law enforcement generally draws minorities more than whites into the criminal justice system through arrest. This broader racial disproportionality could be the possible result of other aspects of NYPD law enforcement, disproportionate involvement in crime (if any) among minorities, potentially criminogenic conditions faced by minorities, or a combination of these and other factors.

These findings have important implications for policing in NYC. Most centrally, they provide partial support to NYC’s decision to continue QOL policing by countering the accusations of possible net widening and racial bias associated with the policy. The analysis provides some evidence to suggest that QOL policing has not been widening the net for arrest in NYC, has not been systematically drawing large numbers of otherwise law-abiding individuals into the criminal justice system, and has not been the primary cause of racial disproportionality among arrestees. Additional research is necessary to identify the effectiveness and cost effectiveness of NYC’s continued QOL policing and to determine whether the program should be maintained on a limited or citywide basis. Also, additional research should examine the complex topic of continued racial disproportionality among NYC arrestees. However, this research should focus much more broadly than on QOL policing.

The findings also have important implications for other jurisdictions considering the use of QOL policing. The study provides some evidence to suggest that QOL policing does not inevitably result in net widening nor increased racial disproportionality. Of course, any jurisdiction’s experiences with QOL policing will depend on the nature of its crime problem, its implementation of QOL policing, and the synergy (or conflict) of QOL policing with other ongoing programs. Moreover, this study suggests the potential for monitoring policing innovations, perhaps through ADAM as in this study. Monitoring can contemporaneously establish the extent of any collateral impact, such as net widening. To the extent that monitoring unveils net widening, it could provide a strong empirical justification for the modification or even abandonment of a program like QOL policing. To the extent that monitoring indicates a lack of net widening, it could provide support for continuation of QOL policing, help silence accusations that QOL policing causes net widening, and suggest that police officials, analysts, critics, and others could more productively focus their efforts on improving the criminal justice system in other ways.

Appendix.

Logistic regression models were estimated to examine the association between various arrestee characteristics (e.g., being single as dependent variable) with two independent variables: arrest charge (coded as serious, trespassing, farebeating, MJ misdemeanor) and age (coded as 18–20, 21–29, 30–39, 40+). The Wald statistic was used to test whether the variation explained by each independent variable was statistically significant. This provided a systematic test for whether the variation observed across arrestee types for some characteristics could be reasonably attributed to age differences across arrestee types. An analogous procedure using weighted least-squares regression was employed to test the sources of variation in the number of prior arrests, a non-binary variable. In most cases, the variation across arrest charges was not statistically significant after controlling for variation with age (see Table 6, p.18). In the model for “did not pick up after dog,” the variation associated with both variables was not statistically significant, suggesting that the significant finding in the binary comparison might have been the result of chance.

Table 6.

Logistic Regression Analyses of Variation in Select Demographic Factors Across Arrest Charge and Age

	Odds Ratios by Dependent Variable
	Arrest Recorda		QOL Offending						Mainstream Status		Recent Drug Useb
	# of arrestsc	Index ever	Farebeating	Trespassing	Smoked MJ in public	Buy/carry MJ in public	Sold MJ in public	Failed to pick up after dog	Single	Welfare	Marijuana	Cocaine/crack	Heroin
ARREST CHARGE
Farebeating	—	—	5.1	1.0	—	—	0.2	—	—	—	1.1	0.6	—
Trespassing	—	—	1.3	4.3	—	—	2.3	—	—	—	0.9	2.8	—
MJ misdemeanor	—	—	1.3	2.3	—	—	2.2	—	—	—	10.5	0.3	—
Seriousd	—	—	1.0	1.0	—	—	1.0	—	—	—	1.0	1.0	—
AGE
18–20	−9.7	0.3	—	—	5.0	3.5	—	—	15.3	0	11.0	0.1	0.4
21–29	−7.3	0.3	—	—	2.5	2.4	—	—	2.1	0.6	4.8	0.2	0.1
30–39d	1.0	1.0	—	—	1.0	1.0	—	—	1.0	1.0	1.0	1.0	1.0
40+	3.2	0.9	—	—	0.8	0.9	—	—	1.0	3.2	0.6	0.8	0.8
Base odds	12.9	3.1	0.8	0.3	0.5	0.4	0.1	0	0.9	0.2	0.5	3.0	0.2

Note. Estimates are weighted to control for overrepresentation of females. Parameter estimates are reported for variables associated with statistically significant variation at the a = .01 level, based on Wald test.

— Not statistically significant

^aBased on NYS official arrest history

^bAs detected by urinalysis

^cWeighted least squares analysis: constant entered in the line marked base odds; additive parameter estimates entered in the lines for odds ratios; F-test of significance used

^dReference category