HomeThe Trends MonitorLong Term TrendsNew York City Crime TrendsOpinion Page2006 Metro Area Violent Crime Rates2006 Metro Area Homicide RatesResearch Papers

Crime and Victimization in the New York Metropolitan Area, 1980-2003:

Comparing Victims’ Reports of Crime to Police Estimates

Janet L. Lauritsen, Ph.D.i
Brian E. Oliver, M.A.
Robin J. Schaum, Ph.D.
Richard Rosenfeld, Ph.D.
Department of Criminology & Criminal Justice
University of Missouri-St. Louis

September 2008


For more than three decades, the Nation has had two national indicators of crime: the Uniform Crime Reporting (UCR) program and the National Crime Victimization Survey (NCVS). The UCR program contains information that is voluntarily provided by police departments and forwarded to the FBI. The NCVS gathers information from a nationally representative sample of persons age 12 or older to produce estimates of crime that are independent of the recording practices of criminal justice systems. Data from the two programs are routinely used together to provide a more complete assessment of crime in the United States. ii

Annual national estimates of the number of crimes derived from NCVS data have often been higher than the annual counts in the UCR. There are several reasons why this may occur. Most importantly, the NCVS data include crimes that are not reported to the police.iii According to recent estimates, approximately 52% of violent victimizations and 62% of property victimizations are not reported to the police. In addition, NCVS counts may be higher if police departments do not record all of the incidents that come to their attention or do not forward the reports to the national UCR program. Indeed, increased knowledge about crimes not reported to the police and assessments of the reliability of police data were two important reasons for the development of the crime survey in the early 1970s. 

For some types of crimes in the NCVS and the UCR, it is possible to reconcile apparent discrepancies in annual national estimates by taking into account differences in the coverage of both data series and adjusting the NCVS counts to include only those incidents said to have been reported to the police. When such adjustments are made, levels and trends in burglary, robbery, and motor vehicle theft appear similar in the NCVS and UCR. On the other hand, UCR and NCVS levels and trends in aggravated assault and rape continue to exhibit discrepancies after these kinds of adjustments are made. Remaining differences in levels or trends in aggravated assault and rape may reflect broader changes concerning the public’s willingness to report crime to the police, the ways in which police departments record crime, the quality of victimization survey data, or other factors. It is clear that the differences in the methodologies of the UCR and NCVS programs must be considered when assessing both levels and trends of crime in the Nation.

Local governments rarely have access to victim survey data.

When State and local governments are interested in assessing levels or trends in crime in their own areas, they typically rely solely on police data because victimization survey data are rarely available for places other than the Nation as a whole. The collection of reliable crime survey data is costly and most State and local governments have not had the resources to conduct their own victimization surveys, especially on an annual basis. Knowing that crime may not be reported to the police or fully recorded, many wonder whether police-based estimates for local areas provide an accurate foundation on which to assess levels or short- or long-term trends in crime. In addition, many wonder whether conclusions drawn from national police and victim survey data also apply to their local areas.

To examine how police records compare to victimization survey estimates for places other than the Nation as a whole, the Bureau of Justice Statistics (BJS) and the U.S. Census Bureau have developed special subsets of NCVS data that are capable of providing survey-based estimates of crime for the largest metropolitan areas in the country. This report compares NCVS and police estimates of burglary, robbery, and aggravated assault for one of the country’s largest metropolitan areas — New York City — for 1980 through 2003. The features of the NCVS sample limit the UCR-NCVS comparisons to the core counties that comprise this metropolitan area (described below). These comparisons can be used to inform the public about the levels and trends in criminal victimization in these areas. They also provide information about the correspondence between police and victim survey data throughout the 1980s, the 1990s and the early 2000s for the New York City area.

Characteristics of the data

NCVS estimates

The NCVS, and its predecessor, the National Crime Survey (NCS), are household-based surveys designed to gather information about victimization directly from the public. A random sample of U.S. households is generated using a multistage stratified sampling procedure. The result is a representative sample of households and persons ages 12 or older living in households.iv The Census Bureau develops the sample and administers the survey for BJS. The sampling strategy for the NCVS is developed explicitly to provide national estimates of criminal victimization. Generally speaking, the sample design is capable of providing local area estimates of crime only for certain places. Heavily populated metropolitan areas constitute unique self representing portions of the sample; as a result, it is possible to use the data from the largest of those areas to form reliable survey-based estimates of crime for those places.v

Two important limitations of the NCVS metropolitan area data should be noted. First, the boundaries of metropolitan areas can change over time due to population shifts and development patterns. To ensure that metropolitan area victimization rates remain comparable over time, it is necessary to restrict the geographic boundaries of the areas to the core counties that remained continuously part of each metropolitan area from 1980 through 2003. For the New York City core county area, the sample is limited to residents from Bronx, Kings, New York, Putnam, Queens, Richmond, Rockland, and Westchester counties.vi

Second, compared to the national sample, the number of persons interviewed within the metropolitan area in a year is relatively small (see table 1). Sample size places statistical limitations on the types of crimes that can be reliably estimated for each metropolitan area and year. The three crimes that occur with sufficient statistical regularity to permit comparisons with police data are burglary, robbery, and aggravated assault. To maximize the reliability of the NCVS metropolitan area estimates of burglary, robbery, and aggravated assault victimization, 3-year moving averages are used for NCVS crime rate estimates.vii

Table 1. Number of NCVS Interviews in the Nw York Metropolitan Core County Area, 1980-2003
Year N of Interviews
2003 5191
2002 5397
2001 5806
2000 5986
1999 5668
1998 5799
1997 6081
1996 5981
1995 6126
1994 6200
1993 6204
1992 6501
1991 6837
1990 6778
1989 6890
1988 6943
1987 7010
1986 7187
1985 7360
1984 8116
1983 9434
1982 9285
1981 8796
1980 9072

Police estimates

Police counts of burglary, robbery, and aggravated assault in each metropolitan area are based on offenses known to the police departments with jurisdiction in the core county areas.viii Police data for each county and year were gathered from different sources for each metropolitan area. Population data for each county and year were obtained from the Census Bureau so that standardized rates could be used for comparisons.

In each metropolitan area, police-based crime rates rely on summaries from multiple police departments and jurisdictions. As a result, it is not possible to draw conclusions about the reporting practices of any particular police department within the metropolitan area. Nonetheless, it is possible to illustrate the degree of correspondence between the two data series by comparing the crime rates based on police department records to those estimated by the survey.

Nature of the comparisons

Three sets of estimates are provided for each crime type and year. The first set of estimates, based on NCVS data, is designated “NCVS Crime Rates.” They consist of the burglary, robbery, and aggravated assault victimization rates regardless of whether the incident was reported to the police.ix The second set of estimates, also based on NCVS data, is the victimization rates based on incidents that victims said were reported to the police. These estimates are referred to as “NCVS Reported Crime Rates.” The third set of estimates, based on police records, consists of the UCR crime rates for each metropolitan area. These estimates are referred to as “UCR Crime Rates.”

For each crime type, the NCVS Crime Rates are compared with the NCVS Reported Crime Rates to assess whether levels of reported crime to the police have changed appreciably.x The NCVS Reported Crime Rates are compared to the UCR Crime Rates to ascertain the correspondence between police data and victim reports for each metropolitan area.


In the New York metropolitan region, police and survey data agree that the burglary rate declined throughout much of the 1980’s, 1990’s and 2000’s (figure 1). The NCVS data shows a small exception in the mid- to late-1980s when rates stabilized and then briefly increased. Over the full time period, the NCVS burglary rate has been roughly two times greater than the UCR rate. The NCVS reported burglary rate has been very similar to the UCR estimates throughout the full period.

Figure 1. Burglary rates for New York City metropolitain area, 1980-2003

The gap between the NCVS burglary rates and the NCVS reported burglary rates appears to be somewhat reduced in recent years compared to the 1980’s and early 1990’s. The data show that burglary reporting rates in the New York city metropolitan area averaged about 54% between 1980 and 1994, but has averaged about 61% since 1995. For comparison, recent estimates for the Nation show that about 56% of burglaries were reported to the police.xi


Unlike residential burglary, robbery can take place away from home. This fact makes comparisons of UCR and NCVS robbery data more complex. If a nonresident, such as a tourist or commuter, is victimized by robbery while in the metropolitan area, the crime (if reported) is likely to be reported to the police department with jurisdiction in the area where the incident occurred. This means that UCR robbery counts for metropolitan areas include crimes committed against nonresidents. NCVS robbery counts, on the other hand, are based only on interviews with persons who reside in the metropolitan core county areas. If UCR robbery rates for the New York area contains many incidents against persons who live outside these counties, they will be higher than the local NCVS robbery rates.xii

Alternatively, if many of the robberies against metropolitan area residents took place while they were outside of the core counties of their metropolitan areas, the NCVS estimates could be higher than UCR estimates. The NCVS counts do not easily distinguish victimizations that occurred inside the metropolitan core county area from those that occurred outside. Little evidence indicates which of the above concerns is more problematic. To resolve the issue would require additional information about the residence of each victim in UCR records and location information for each incident in the NCVS reports. The UCR-NCVS correspondence in robbery will reflect in part, the extent to which these two concerns are relatively equal.

Police and victim survey data for robbery in the New York area suggest similar trends during the 1980’s. According to victim survey reports, rates fell fairly substantially between 1982 and 1986, with police data showing similar, if smaller, declines. Both police and victim survey data indicate that rates increased during the later 1980’s, and after 1991 began dropping at a fairly continual pace until about 2001 (see figure 2). For the period 2001-2003, both NCVS and UCR robbery rates have been fairly stable. From 1994 to 2000, the levels of robbery reported to the police by victims have closely corresponded to the estimates in the UCR. However, during the late 1980’s, early 1990’s, and the last few years of the series, the UCR rates were somewhat higher than the victim reports.

Figure 2. Robbery rates for the New York City metropolitan area, 1980-2003

The gap between the NCVS robbery rates and the NCVS reported robbery rates was relatively stable from the early 1980’s through the early 2000’s, suggesting no significant change in reporting. The survey data show that robbery reporting rates in the New York city metropolitan area ranged around 48% from 1980 to 1984, 51% from 1985 to 1989, 42% from 1990 to 1994 and 54% from 1995 to 1999. For 2000-2003, the rate has averaged 51%. Estimates for the Nation show a somewhat higher proportion (60%) of robberies reported to the police.


Like robbery, comparisons of police and victim survey data for aggravated assault involve unique complexities. Police data will include an unknown number of crimes committed against nonresidents, and residents may become victims of aggravated assault when they are outside of their metropolitan area. Police data will also include an unknown number of crimes against persons ineligible or unlikely to participate in the NCVS. But the measurement of aggravated assault also is unique in that it is likely to have been affected by broad scale changes in police response to domestic assault incidents. Mandatory arrest policies have replaced informal dispositions in many areas of the Nation. As a result, trends exhibited in police data may not correspond closely with trends based on survey data, but instead reflect changes in police handling of aggravated assault incidents.xiii

Aggravated assault estimates for the New York metropolitan area show a mixed set of trends. The survey data suggest that rates were peaked in the early 1980’s, the late-1980’s and the early- to mid- 1990’s, with steady declines since then (see figure 3). In comparison, police-based estimates of aggravated assault showed steady increases from 1983 to 1989 and steady declines thereafter. However, the police rates suggest that the declines in aggravated assault began around 1988, several years before the rates began to decline in the victim survey data (1993). Also, starting around 1984, the levels of aggravated assault have been higher in the police data than expected based on the number of incidents victims said they reported to the police.

Figure 3. Aggravated assault rates for New York City metropolitan area, 1980-2003

The gap between the NCVS aggravated assault rate and the NCVS reported aggravated assault rate has fluctuated over time. Between 1980 and 1984, approximately 59 percent of NCVS aggravated assault victims reported the incident to the police. This average then dropped to 48 percent between 1985 and 1989. Beginning in 1990, however, the percentage reporting to the police increased above 50 percent and averaged around 53 percent between 1990 and 2003. Estimates for the Nation show asomewhat higher proportion (62%) of aggravated assaults reported to the police.

In the New York city metropolitan area, the correspondence between the police and survey estimates of aggravated assault was somewhat less than the correspondence for the crimes of burglary and robbery. More fluctuations also were exhibited in the survey estimates than in the UCR data. The lower degree of correspondence between the NCVS aggravated assault rates and the UCR rates makes it more difficult to draw conclusions about long-term trends in aggravated assault. However, both data sources agree that rates of aggravated assault generally declined from the mid-1990’s through 2003.


As noted earlier, local governments often are interested in assessing levels or trends in crime, but they typically draw conclusions about crime in their areas solely on the basis of police data. These analyses have examined how police data compare to victim survey data for burglary, robbery, and aggravated assault for the New York City metropolitan area. The main findings are as follows:

  • For burglary and robbery, UCR crime trends were generally similar to NCVS reported crime trends for 1980 through 2003. Levels of victim-reported burglary and police estimates of burglary were fairly similar as well. However, levels of NCVS victim-reported robbery were slightly lower than police estimates.

  • For aggravated assault, there were more discrepancies in both levels and trends between UCR crime rates and NCVS reported crime rates. This suggests that the police data are likely to have been affected by broader policy changes in police handling of domestic violence incidents and other serious assaults.

  • The reporting rates in the New York metropolitan area remain slightly higher for burglary, but slightly lower for robbery and aggravated assault, than reporting rates for the Nation as a whole.

i This report is one in a series of metropolitan area reports comparing victim estimates of crime to police estimate. It is an updated and revised version of an earlier publication prepared when the first author was Visiting Research Fellow at the Bureau of Justice Statistics (see Lauritsen and Schaum, Crime and Victimization in the Three Largest Metropolitan Areas, 1980-98, Bureau of Justice Statistics, 2005, NCJ 208075).

ii In this report the terms “police data” and “UCR data” are used interchangeably. For more detail about the UCR and NCVS programs, see “The Nation’s Two Crime Measures,” Bureau of Justice Statistics, 2003, NCJ 122705.

iii For more details, see Hart and Rennison (2003), Reporting Crime to the Police, 1992-2000, Bureau of Justice Statistics, NCJ 195710.

iv Persons not living in households, such as homeless, institutionalized or incarcerated persons, are not included in the sample design.

v The NCVS sample is generated in two stages. During the first stage, primary sampling units (PSU's) are designated to reflect metropolitan areas, counties, and groups of counties. Larger PSU's (reflecting metropolitan areas) are termed self-representing (SR) and are automatically included in the sample. Smaller PSU's (non-self-representing or NSR PSU's) are grouped together in similar strata based on known Census geographic and demographic characteristics. From the NSR PSU's, sample PSU's are selected by probability proportionate to population size.

vi For an extensive analysis comparing UCR and NCVS rates for New York City through 1999, see also Langan and Durose, "The Remarkable Drop in Crime in New York City," ISTAT (Italy's National Institute of Statistics) meeting “per una societa pie secura,” Rome, December 3-5, 2004.

vii The NCVS data cover the period from 1979 to 2004. The use of 3-year moving averages results in the loss of the first and last year estimates, however it improves the stability of the survey estimates. Each estimate in this report is based on approximately 16,000 to 27,000 interviews with persons ages 12 and over

viii Because police departments voluntarily submit data to the UCR, often times a large amount of data is missing across time and places. This missing data issue made it difficult to collect consistent data across all of the counties over the 20-year period. For this reason, the police data used here for the metropolitan core county areas of New York city were gathered from the State criminal justice website rather than from the UCR reports. The New York city metropolitan area police data were retrieved from the New York State Division of Criminal Justice Services at: http://www.criminaljustice.state.ny.us/

ix Because the NCVS is a household-based survey, it includes residential burglaries but not commercial burglaries. The UCR data that were available to us included both residential and commercial burglaries. Because we were unable to determine how many of the police recorded burglaries were against commercial establishments in each county and year, the UCR burglary rates are expected to be higher than the NCVS reported burglary rates. Also, unlike the NCVS burglary rate which is typically based on the number of households, the UCR burglary rate is measured by taking the total number of burglaries known to the police and dividing by the total number of persons. To make the NCVS and UCR burglary rates as comparable as possible, the NCVS rates were estimated using persons age 12 or over in the denominator. In addition, it was not possible to determine how many of the police recorded robberies and aggravated assaults were against persons under age 12 (incidents excluded from the NCVS because persons under age 12 are not interviewed). As a result, the NCVS robbery and aggravated assault rates were created by dividing the number of victimizations by the population age 12 or over, while the UCR rates were created by dividing the total number of offenses known divided by the total population for each county and year. UCR rates based on the total population age 12 or over in the denominator also were calculated to check the sensitivity of the findings to differences in population coverage. Of course, the UCR rates increased when the denominator was restricted to the population age 12 or over. However, the substantive conclusions reported here were similar regardless of which denominator was used for the UCR and NCVS rates.

x Significant changes were made to the methodology of the NCVS in 1992. To make the NCVS rates comparable before and after 1992, the earlier rates were weighted by their crime specific adjustment factors. These weights were derived from assessments of how national estimates changed following a phase-in of the new methodology. There is little reason to suspect that the effects of the new methodology varied across these three metropolitan areas. For more information on the weighting of crimes, see Kindermann, C., Lynch, J., and Cantor, D. (1997), Effects of the Redesign on Victimization Estimates, Bureau of Justice Statistics, NCJ 164381.

xi Catalano, S. (2006). Criminal Victimization, 2005, Bureau of Justice Statistics, NCJ 214644.

xii UCR robbery rates also will be higher than NCVS robbery rates if police robbery data include many incidents against persons who are not part of households (for example, homeless persons). Such persons are ineligible to participate in household-based surveys.

xiii For an analysis comparing national estimates of these data, see Rosenfeld, R. (2007), “Explaining the divergence between UCR and NCVS aggravated assault trends,” pp. 251-68 in Lynch, J. and Addington, L. (eds.), Understanding Crime Statistics: Revisiting the Divergence of the NCVS and the UCR, Cambridge University Press.

Enter supporting content here