e-government for uniform crime reporting data and how to make it more accessible

Upload: dblau2

Post on 02-Apr-2018

221 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    1/27

    1

    E-Government for Uniform Crime Reporting Data: How to Make it More Accessible

    The Value of Ensuring OpenGovernmentData is Truly Accessible

    Transparency and availability of government data is a critical goal. Jaeger, Bertot, and

    Grimes describe a survey of benefits of government transparency, including democratic

    participation, prevention of corruption, informed decision-making, and accuracy of government

    information.1 However, they note that, If the Obama administration's renewed commitments to

    transparency and the focus on the capacities of e-government and the web as an avenue to

    advance transparency are to succeed, their efforts must ultimately result in citizen-centered

    approaches that are available to and usable by all members of the publicfuture policy will need

    to focus on the human dimensions of transparency, not just the technological dimensionsto be

    embraced by the public, it will need to not only be available to all, but be designed to be usable

    by all. They go on to describe the technical skill gaps most often encountered by the general

    public in attempting to fully interpret and use government data:2

    1) Technology literacy: the ability to use and understand technologies2) Usability: the design of technologies in such ways that are intuitive and allow users to

    engage in the content embedded within the technology

    3) Accessibility: the ability of persons with disabilities (or any person lacking a certaintechnical background) to be able to access the content

    By: David Blau, George Mason University Graduate School of Public Policy, 11/26/2013

    Contact: [email protected]

    1Bertot, John Carlo, and Jaeger, Paul T. Transparency and Technological Change: Ensuring Equal and Sustained

    Public Access to Government Information. Government Information Quarterly, Vol. 27, pp. 371, 2010.

    2Bertot, John Carlo, Grimes, Justin M., and Jarger, Paul T. Using ICTs to Create a Culture of Transparency: E-

    Government and Social Media as Openness and Anti-Corruption Tools for Societies. Government Information

    Quarterly, Vol. 27, pp. 268, 2010.

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    2/27

    2

    4) Functionality: the design of the technologies to include features (e.g., search, e-government service tracking; accountability measures, etc.) that users desire

    In order to bridge the gap, and make data truly accessible to the public, government should

    provide training for users and testing that each aspect of the technical gap is minimized. But,

    accommodating all users, regardless of technical background, is often neglected. Bertot et al.

    find that government transparency websites likewww.data.govare directed toward the more

    technically-inclined.3 In addition, e-government services generally are limited by difficulties

    in organization, structure, search, metadata, and other factors.4

    The National Incident-Based Reporting System: Intelligent Federal Data on Crime

    The Uniform Crime Reporting (UCR) system was developed in 1929 to meet the need for

    reliable standardized crime statistics. According to the Federal Bureau of Investigation, the

    Uniform Crime Reporting (UCR) system was, conceived in 1929 by the International

    Association of Chiefs of Police to meet the need for reliable uniform crime statistics for the

    nation. In 1930, the Federal Bureau of Investigation was tasked with collecting, publishing, and

    archiving those statistics.5

    3See Bertot et al., Using ICTs to Create a Culture of Transparency: E-Government and Social Media as Openness

    and Anti-Corruption Tools for Societies. This approach is typified by the nascent and ambitious plan by the

    Obama administration to make vast amounts of government data available through the www.data.gov site. These

    types of transparency initiatives are directed toward the more technically inclined citizen: researchers,

    technologists, and civic-minded geeks. (page 268)

    4See Bertot et al., Transparency and Technological Change: Ensuring Equal and Sustained Public Access to

    Government Information. Even for members of the public with internet access, e-government services generally

    are limited by difficulties in organization, structure, search, metadata, and other factors. Consider the unrefined

    scope and disorganization of the typical results of a search on www.usa.gov. E-government in the United States

    simply has not been designed to account for the needs of the users of e-government, particularly the members of

    the public seeking information or engagement. (page 373)

    5http://www.fbi.gov/about-us/cjis/ucr/ucrFBI Uniform Crime Reports website, retrieved June 30, 2013

    http://www.data.gov/http://www.data.gov/http://www.data.gov/http://www.fbi.gov/about-us/cjis/ucr/ucrhttp://www.fbi.gov/about-us/cjis/ucr/ucrhttp://www.fbi.gov/about-us/cjis/ucr/ucrhttp://www.fbi.gov/about-us/cjis/ucr/ucrhttp://www.data.gov/
  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    3/27

    3

    In 1985, Abt Associates6recommended critical changes to the UCR program in the

    Blueprint for the Future of the Uniform Crime Reporting Program.7 Instead of just collecting

    summary data, they proposed that the FBI should collect detailed data on each individual

    criminal incident, and make the data more meaningful by attaching demographic information, in

    order to provide better insight into the causes and effects of crime. By collecting and providing

    detailed information on each incident, the National Incident-Based Reporting System (NIBRS)

    would be able to enhance the quantity, quality, and timeliness of data collected by the law

    enforcement community, and to improve the methodology used for compiling, analyzing,

    auditing, and publishing the collected crime data.

    From the start of the program, NIBRS was intended for a variety of types of users. As

    the FBI NIBRS program history states8:

    While the Programs primary objective is to generate a reliable set of criminal

    statistics for use in law enforcement administration, operation, and management, its data

    have over the years become one of the countrys leading social indicators. The American

    public looks to Uniform Crime Reports for information on fluctuations in the level of

    crime, while criminologists, sociologists, legislators, municipal planners, the press, and

    other students of criminal justice use the statistics for varied research and planning

    purposes.

    6http://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/1995/95sec1.pdfFBI Summary of the Uniform Crime

    Reporting Program, retrieved June 30, 2013

    7http://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/2011/crime-in-the-u.s.-2011/aboutucrmain FBI - About

    The Uniform Crime Reporting Program, retrieved June 30, 2013

    8http://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/1995/95sec1.pdfIbid Summary

    http://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/1995/95sec1.pdfhttp://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/1995/95sec1.pdfhttp://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/1995/95sec1.pdfhttp://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/2011/crime-in-the-u.s.-2011/aboutucrmainhttp://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/2011/crime-in-the-u.s.-2011/aboutucrmainhttp://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/2011/crime-in-the-u.s.-2011/aboutucrmainhttp://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/1995/95sec1.pdfhttp://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/1995/95sec1.pdfhttp://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/1995/95sec1.pdfhttp://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/1995/95sec1.pdfhttp://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/2011/crime-in-the-u.s.-2011/aboutucrmainhttp://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/1995/95sec1.pdf
  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    4/27

    4

    However, if the version of the data made available to the American public does not

    include all of this demographic data, or at least requires significant additional technical expertise

    and time to make the detailed information easily understandable, then the process of integrating

    and presenting the data could be improved. In order for data to be effective for research, one

    would want both historical data and a sufficient sample size.

    This paper focuses on NIBRS/UCR processes for data gathering and analysis regarding

    hate crimes. This is because this data is far less available than other UCR data, and does not

    fulfill the open e-government accessibility criteria discussed above.9 The original NIBRS/UCR

    data for hate crimes is available to the public in two formats. One format is summary data that

    omits many of the data fields from the original NIBRS entry, separated into spreadsheets based

    on types of victims and offenses, found online.10 This format has three limitations. First, most

    of the NIBRS-gathered additional data intended to provide a greater degree of specificity and to

    enable more productive research and analysis is not included, in order to present a simplified

    view.11 The FBI summary data gives totals each year, whereas in the original NIBRS, each

    incident is its own data point with an incident number and detailed info. Second, the data is

    presented for each year in fourteen different spreadsheets, requiring a manual combination of the

    data across spreadsheets to analyze and compare hate crime rates between regions and across

    9The National Archive of Criminal Justice Data (NACJD) website does not contain any of the original full incident

    UCR hate crime data. Seewww.icpsr.umich.edu/icpsrweb/NACJD/.

    10http://www.fbi.gov/stats-services/crimestatsFBI UCR statistics, including hate crime statistics. Retrieved June30, 2013.

    11Prior research into UCR Hate Crime data often relies on the FBI summary data, which is far removed from the

    raw data, and omits the accompanying demographic data. For example, see Rubenstein, William B., The Real

    Story of U.S. Hate Crimes Statistics: An Empirical Analysis.Tulane Law Review, Vol. 78, pp. 1213-1246, 2004.

    Available at SSRN: http://ssrn.com/abstract=547883

    http://c/Users/David/Downloads/www.icpsr.umich.edu/icpsrweb/NACJD/http://c/Users/David/Downloads/www.icpsr.umich.edu/icpsrweb/NACJD/http://c/Users/David/Downloads/www.icpsr.umich.edu/icpsrweb/NACJD/http://www.fbi.gov/stats-services/crimestatshttp://www.fbi.gov/stats-services/crimestatshttp://www.fbi.gov/stats-services/crimestatshttp://www.fbi.gov/stats-services/crimestatshttp://c/Users/David/Downloads/www.icpsr.umich.edu/icpsrweb/NACJD/
  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    5/27

    5

    years. Third, FBI hate crime Tables 1-11 show totals for incidents, offenses and victims by bias

    type or offender nationwide; and Tables 12-14 show much more summarized data by individual

    cities, counties, other agencies such as universities, and by state totals. This separation and

    simplification of location data results in the user being able, for example, to see the total of

    religious bias motivated crimes in Atlanta, Georgia, but not the bias-motived crimes against a

    more detailed breakdown of religion in Atlanta - not, in other words, how many crimes were

    allegedly committed against Jews, Catholics, Protestants or Muslims specifically within a county

    or city because of their religious beliefs. The data is summarized and separated to such an

    extent, it limits the ability of civil society or law enforcement organizations to understand the

    nature of hate crimes in their local communities, and to work with local resources to reduce or

    eliminate the incidence of those crimes. This is clearly an example of where the Obama

    administrationsprogress towards truly open e-government is admirable, but the goals of full

    functionality and detail of the data versus accessibility and understandability for all public

    citizens are not being jointly met. Even the FBI summary totals only go back to 2004, meaning

    that the entirety of hate crime data gathered from the period 19912003 must be obtained and

    interpreted in ASCII. Without the process described here of reformatting that data, it would be

    unavailable to any researcher, law enforcement personnel, policy maker, or average citizen

    lacking the requisite technical background.

    In the second format, the public can obtain detailed, incident-by-incident data including

    geographic location from the FBI, but it is in ASCII, a format requiring technical expertise and

    translation to make it usable for research and analysis. In order to translate from ASCII into a

    more easily understood format like that for Microsoft Excel (.xls or .xlsx), one needs statistical

    software like STATA, SPSS, or SAS. These three programs, however, require substantial

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    6/27

    6

    technical training in order to teach the user how to translate and analyze data, creating a major

    barrier to access to any citizen lacking this training. They are also prohibitively expensive, if the

    user does not have an educational discount. The cheapest version of STATA, STATA 13/IC,

    costs $595 for a single year license, or $1,195 for a perpetual license without technical support.12

    The cheapest version of SPSS, SPSS Standard, costs $2,320 for a fixed term license with a year

    of technical support. The cheapest version of SAS is $8,700 for an annual license with a year of

    technical support. This again is a prohibitive barrier to the average citizen being able to access

    this data.

    The full original ASCII dataset is only available through an email request. This second

    format provides more detailed data (e.g., the subcategory breakdowns of bias motivations, the

    known offenders races, and the victim types for each agency submitting hate crime data to the

    FBI UCR Program) from the FBI UCR Programs Hate Crime Master Files at the Criminal

    Justice Information Services Division.13 While this data is comprehensive, it is not provided in a

    format that is easy to understand or use. A single-page of documentation is provided, which

    states in entirety:

    What you should know before you download the Uniform Crime Reporting

    (UCR) Programs master files

    12SPSS on the IBM website:https://www-

    112.ibm.com/software/howtobuy/buyingtools/paexpress/Express?part_number=D0EKZLL%2CD0EEMLL%2CD0EK0

    LL%2CD0EEJLL&catalogLocale=en_US&Locale=en_US&country=USA&PT=jsp&CC=USA&VP=&TACTICS=%26S_TACT

    %3D%26S_CMP%3D%26brand%3Dnone&ibm-submit=View+US+prices+%26+buy. SAS:

    https://www.sas.com/order/product.jsp?code=PERSANLBNDL. And STATA:

    http://www.stata.com/order/new/bus/single-user-licenses/dl/.

    13http://www.fbi.gov/about-us/cjis/ucr/hate-crime/data-collection-manual/viewUniform Crime Reports Hate

    Crime Data Collection Guidelines and Training Manual, p. 57, retrieved June 30, 2013.

    https://www-112.ibm.com/software/howtobuy/buyingtools/paexpress/Express?part_number=D0EKZLL%2CD0EEMLL%2CD0EK0LL%2CD0EEJLL&catalogLocale=en_US&Locale=en_US&country=USA&PT=jsp&CC=USA&VP=&TACTICS=%26S_TACT%3D%26S_CMP%3D%26brand%3Dnone&ibm-submit=View+US+prices+%26+buyhttps://www-112.ibm.com/software/howtobuy/buyingtools/paexpress/Express?part_number=D0EKZLL%2CD0EEMLL%2CD0EK0LL%2CD0EEJLL&catalogLocale=en_US&Locale=en_US&country=USA&PT=jsp&CC=USA&VP=&TACTICS=%26S_TACT%3D%26S_CMP%3D%26brand%3Dnone&ibm-submit=View+US+prices+%26+buyhttps://www-112.ibm.com/software/howtobuy/buyingtools/paexpress/Express?part_number=D0EKZLL%2CD0EEMLL%2CD0EK0LL%2CD0EEJLL&catalogLocale=en_US&Locale=en_US&country=USA&PT=jsp&CC=USA&VP=&TACTICS=%26S_TACT%3D%26S_CMP%3D%26brand%3Dnone&ibm-submit=View+US+prices+%26+buyhttps://www-112.ibm.com/software/howtobuy/buyingtools/paexpress/Express?part_number=D0EKZLL%2CD0EEMLL%2CD0EK0LL%2CD0EEJLL&catalogLocale=en_US&Locale=en_US&country=USA&PT=jsp&CC=USA&VP=&TACTICS=%26S_TACT%3D%26S_CMP%3D%26brand%3Dnone&ibm-submit=View+US+prices+%26+buyhttps://www-112.ibm.com/software/howtobuy/buyingtools/paexpress/Express?part_number=D0EKZLL%2CD0EEMLL%2CD0EK0LL%2CD0EEJLL&catalogLocale=en_US&Locale=en_US&country=USA&PT=jsp&CC=USA&VP=&TACTICS=%26S_TACT%3D%26S_CMP%3D%26brand%3Dnone&ibm-submit=View+US+prices+%26+buyhttps://www-112.ibm.com/software/howtobuy/buyingtools/paexpress/Express?part_number=D0EKZLL%2CD0EEMLL%2CD0EK0LL%2CD0EEJLL&catalogLocale=en_US&Locale=en_US&country=USA&PT=jsp&CC=USA&VP=&TACTICS=%26S_TACT%3D%26S_CMP%3D%26brand%3Dnone&ibm-submit=View+US+prices+%26+buyhttps://www.sas.com/order/product.jsp?code=PERSANLBNDLhttps://www.sas.com/order/product.jsp?code=PERSANLBNDLhttp://www.stata.com/order/new/bus/single-user-licenses/dl/http://www.stata.com/order/new/bus/single-user-licenses/dl/http://www.fbi.gov/about-us/cjis/ucr/hate-crime/data-collection-manual/viewhttp://www.fbi.gov/about-us/cjis/ucr/hate-crime/data-collection-manual/viewhttp://www.fbi.gov/about-us/cjis/ucr/hate-crime/data-collection-manual/viewhttp://www.fbi.gov/about-us/cjis/ucr/hate-crime/data-collection-manual/viewhttp://www.stata.com/order/new/bus/single-user-licenses/dl/https://www.sas.com/order/product.jsp?code=PERSANLBNDLhttps://www-112.ibm.com/software/howtobuy/buyingtools/paexpress/Express?part_number=D0EKZLL%2CD0EEMLL%2CD0EK0LL%2CD0EEJLL&catalogLocale=en_US&Locale=en_US&country=USA&PT=jsp&CC=USA&VP=&TACTICS=%26S_TACT%3D%26S_CMP%3D%26brand%3Dnone&ibm-submit=View+US+prices+%26+buyhttps://www-112.ibm.com/software/howtobuy/buyingtools/paexpress/Express?part_number=D0EKZLL%2CD0EEMLL%2CD0EK0LL%2CD0EEJLL&catalogLocale=en_US&Locale=en_US&country=USA&PT=jsp&CC=USA&VP=&TACTICS=%26S_TACT%3D%26S_CMP%3D%26brand%3Dnone&ibm-submit=View+US+prices+%26+buyhttps://www-112.ibm.com/software/howtobuy/buyingtools/paexpress/Express?part_number=D0EKZLL%2CD0EEMLL%2CD0EK0LL%2CD0EEJLL&catalogLocale=en_US&Locale=en_US&country=USA&PT=jsp&CC=USA&VP=&TACTICS=%26S_TACT%3D%26S_CMP%3D%26brand%3Dnone&ibm-submit=View+US+prices+%26+buyhttps://www-112.ibm.com/software/howtobuy/buyingtools/paexpress/Express?part_number=D0EKZLL%2CD0EEMLL%2CD0EK0LL%2CD0EEJLL&catalogLocale=en_US&Locale=en_US&country=USA&PT=jsp&CC=USA&VP=&TACTICS=%26S_TACT%3D%26S_CMP%3D%26brand%3Dnone&ibm-submit=View+US+prices+%26+buy
  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    7/27

    7

    The UCR Program provides its master files (i.e., raw, unpublished data), data sets

    extracted from these master files (as electronic text files), and printouts (in the form of

    paper or portable document format files [PDFs])also extracted from the master files

    upon request. Anyone wishing to obtain any of these data may do so via e-mail at

    [email protected].

    Before you download master files from this site, you should be aware of the

    following facts:

    UCR master files are formatted in ASCII only. (Please be aware they are

    not available in Microsoft Excel or Access.)

    UCR master files must be imported into statistical software, such as SAS

    or SPSS, in order to be read properly. The UCR Program does not provide statistical

    software in conjunction with the master files; therefore, the data requester must have

    access to the appropriate software.

    The UCR Program does not provide technical assistance for the master

    files. We will provide record descriptions, which will be disseminated with the files, but

    staff is not available to assist you with technical glitches such as problems importing

    data, etc.

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    8/27

    8

    Users should be aware that some of the files (most notably the National

    Incident-Based Reporting System [NIBRS] master files) are quite large and typically are

    used in a mainframe environment.14

    Two university research centers have taken the whole of the original FBI UCR data and

    reformatted it, but have normalized the data according to certain assumptions and present the

    data in a format that still requires technical expertise and use of expensive statistical modeling

    software. The National Archive of Criminal Justice Data (NACJD) of the Inter-University

    Consortium on Political and Social Research (ICPSR) at the University of Michigan houses an

    alternative version general UCR data, albeit aggregated to the county-level, isolated by year, and

    which still requires knowledge of software like STATA or SPSS. Michael Maltz and Joseph

    Targonski note the comments of a reviewer of their research on the inconsistencies of UCR data

    in ICPSR format, One reviewer of an earlier version of this article noted that [t]he weakness

    of the ICPSR county-level data file is [obvious] to anyone who carefully reads the ICPSR

    codebook. Yetnone of the critics of MGLC (referring to a 1998 paper using UCR data, called

    More Guns, Less Crime) noted or mentioned it; apparently many users of the data do not read

    the ICPSR codebook carefully.15 The NACJD/ICPSR data is not the raw original FBI data, but

    comes from the Crime-by-CountyFBI summary file. This crime-by-county file distributes

    crime totals from an agency that lies in multiple counties to each county based on the percentage

    of the total agency population that lies in each county. In addition, county population figures in

    14Uniform Crime Reporting What you should know before you download the Uniform Crime Reporting (UCR)

    Programs master files, text document provided with ascii file database, received by email January 2013 from the

    Criminal Justice Information Services Division.

    15Maltz, Michael D., and Targonski, Joseph. A Note on the Use of County-Level UCR Data. Journal of

    Quantitative Criminology, Vol. 18, No. 3, pp. 298-300, September 2002.

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    9/27

    9

    this data do not include the populations from jurisdictions that failed to report crime data, so the

    total county population figures can be incorrect.16 Maltz and Targonski note that the FBI is

    aware of the shortcomings of UCR data, and on the basis of the above, The FBI publishes

    warnings about comparing cities and counties; they can be found in every issue of their annual

    report, Crime in the United States. The warning reads, These rankings [of cities and counties

    based on their Crime Index figures] lead to simplistic and or incomplete analyses which often

    create misleading perceptions.17

    The National Consortium on Violence Research (NCOVR) at Carnegie Melon also

    houses UCR data, but it is only available via requests to an Oracle database, requiring knowledge

    of SQL. Maltz and Targonski go so far as to say, Such factors can often be accounted for by

    analyzing the data using sophisticated computer models. But these models cannot compensate

    for missing data of the type and magnitude encountered in the UCRDue to the Problems

    caused by the imputed data, we conclude that county-level crime data, as they are currently

    constituted, should not be used, especially in policy studies.18 But no source provides

    comprehensive data in the form of the original raw incident reports in the NIBRS system in a

    way that an average citizen can use.

    On behalf of the public, journalists and academic researchers who may not have access to

    the mainframe environment recommended by the Criminal Justice Information Services

    Division, we reorganized and streamlined the data, and conducted an assessment of its

    availability and quality using commonly available software: a combination of Microsoft Excel,

    16ibid., p. 300.

    17ibid., p. 300.

    18ibid., p. 300.

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    10/27

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    11/27

    11

    to contact us so that we can correct them. The file is also very large, approximately 25

    megabytes, and so can take a long time to load and save on many computers.

    Step 1: Make ASCII Code Readable for Anyone

    The Uniform Crime Reporting ASCII dataset for hate crimes obtained from the FBI

    consists of 155,338 data records, separated by year, for 1991 through 2011. The data must be

    translated using the FBI-provided codebook, the Hate Crime Yearly Master Record Description

    (1995). This document explains which digits of the string of numbers and letters pertain to what

    information. Otherwise, the data appears as shown below:

    BH01ALAST0000 19910430 MONTGOMERY, AL AL8 D632N

    BH01AL0010000 1991043019910101BIRMINGHAM, AL AL9 A632N

    BH01AL0010100 1991043019910101BESSEMER, AL AL4 631N

    Individual data within each cell appears meaningful, and indeed it is. For example, the first eight

    digits of the second cell are the date, in a year/month/day format. However, the columns are not

    separated in the document according to where each data field should end, as one can see above

    from the date/date/city name combination in the second column.

    One method to format the large UCR hate crimes ASCII file into a readable document

    requires creating a dictionary file within statistical software such as STATA, a schema in

    Microsoft Access, or the function to manually adjust column widths in the Text Import Wizard

    function in Microsoft Excel. We relied primarily on STATA and Excel, based on the data

    definitions provided by the UCR Hate Crime Yearly Master Record Description (or codebook).

    Other researchers may be able to provide even simpler approaches. While the use of a dictionary

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    12/27

    12

    file would assist in translation of such large datasets in STATA, some individual years stray from

    the attached codebook dictionary file (given as a picture pdf, not text, so it also must be manually

    re-created), making that approach arguably more time-consuming than simply readjusting

    column widths on a year-by-year basis. But, it should be noted that individual years can be off

    from the codebook dictionary by a digit or two, meaning each year has to be visually inspected

    digit-by-digit in the translation and integration process.

    Here is a detailed description of the process used to integrate these disparate documents

    into a single, streamlined database for the general user with a less technical background in ASCII

    or STATA. Below is a graphic of the Text Import Wizard in Microsoft Excel, which we used

    for the first step in the reformatting process.

    The process must be repeated once for the Batch Header data(rows containing

    information about a reporting agency), and once for the Incident Report data (rows containing

    information about a particular incident), as the column definitions differ between the two.

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    13/27

    13

    Incidents are arranged in the rows below the corresponding Batch (agency that reported the

    incident), rather than in the same row. The latter format is more useful for the researcher,

    because it would link batch (geographic and demographic) information with the details of the

    incident in the same record. An agency can be any of several types of local reporting

    organizationcity, county, university etc. Given the current configuration of the file, using the

    codebook alone to define the columns in order to make the data readable is insufficient, because

    correcting the columns for the Batch Header data renders the Incident Report data

    meaningless, as the columns are defined in different places for each type of entry. For example,

    see below for hate crime data reported for the year 1991 for Hoover, AL:

    BH 1 AL0011200 1991 430 19910101 HOOVER AL 3 6 3 1

    IR 1 AL0011200 K0Y L6XY ED4K2007 1010N400101B13

    IR 1 AL0011200 3A0G Q-FO 771G2007 1226N400100U29

    Two hate crime incidents were reported by the Hoover, Alabama agency. However,

    when formatting columns for the BH data type, the IR data underneath loses its meaning, as

    the column breaks are different for each data type. Therefore, in order to compare trends in

    incident reports (hate crimes) across the data containedwithin the batch header (such as

    population, reporting agency name, or locality type, such as county or city), the Incident

    Report data must be manually re-arranged left-to-right in the same row with its corresponding

    locality information, rather than in a separate row beneath it.20

    20Again, a dictionary file would solve this issue, but some years are off by a digit or two.

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    14/27

    14

    This was accomplished by using a lookup function in Microsoft Excel using the one

    common denominator between the IR and BH data, the Originating Agency Identifier

    (ORI) number. The ORI number is the unique identifying federal number pertaining to each

    reporting local agency. The end result is a spreadsheet where an individual row entry contains

    both information on the details of each individual hate crime (incidents) as well as data

    describing the agency and associated locality that reported it.21

    Step 3: Reformatting Codes into Text

    Next, certain data fields, even after formatting, still require the codebook to interpret their

    meaning. The location where the hate crime occurred, for example, is a two-digit code:

    21Using a dictionary file in STATA is likely the most efficient way to do this. This requires knowledge of STATA

    programming using a dictionary file. Doing so would still require using the ORI (the only common denominator

    between the IR and BH data).

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    15/27

    15

    (Reproduced from Hate Crime Yearly Master Record Description, p. 22)

    We translated the values for this variable and for similar variables across the entirety of

    the reported incidents from 1991-2011, using STATA and value labels based on the Master

    Record Description. Researchers working with this data need some sort of program (like

    STATA) to automate this renaming process, because there are more than 150,000 records in the

    hate crime UCR data, each with over 50 data fields. So, while an incident originally appeared as

    follows:

    loccode01 populationgroup ucroffensecode01 biasmotivation01

    12 9A 290 25

    It now appears in our version as this:

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    16/27

    16

    Location Population

    Group

    UCR Offense Code Bias Motivation

    Grocery/Supermarket MSA Counties

    100,000 or over

    Destruction/Damage/Vandalism

    of Property

    Anti-Other Religion

    These simple translation methods fulfill the programs intention of making this data more

    accessible to the media, researchers, local officials and the general public.

    Data Label (column) Definitions for Final Recompiled MS Excel Database

    From left to right (with page in the codebook where the value label is provided in

    parentheses):

    1) Incident Date: the date the incident occurred (page 18 of the codebook)2) Incident Number: the unique identifying number given to each hate crime incident by the

    local reporting agency (pg. 17)

    3) Data Source: the media by which the FBI received the incident data (pg. 17)4) Agency Name (brought over from Batch Header data): the name of the local reporting

    agency (note that this is not always identical to City Name) (pg. 15)

    5) ORI: Originating Reporting Identifier (brought over from Batch Header data): uniquenumber assigned to each local reporting agency (pg. 6)

    6) City Name (brought over from Batch Header data): name of the city covered by the ORI(pg. 7)

    7) Population (brought over from Batch Header data, Current Population) (pg. 14)

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    17/27

    17

    8) Agency Indicator (brought over from Batch Header data): distinguishes amongcounties/colleges/cities for local reporting agency (pg. 12)

    9) Core City (brought over from Batch Header data): indicates whether or not the ORI is thecore city of a Metropolitan Statistical Area (pg. 12)

    10)Country Division (brought over from Batch Header data): geographic division in whichthe state of the local reporting agency is located (pg. 9)

    11)Country Region (brought over from Batch Header data): a higher-level agglomeration ofthe Country Division field above, the geographic region in which the above

    geographic division is located (pg. 11)

    12)Population Group (brought over from Batch Header data): the population category of theORI (pg. 8)

    13)Original ASCII Victims Total: the Total Number of Individual Victims field in theincident data (pg. 18)

    14)New Victims Total: Sum of Victims for Each Offense: As noted below, the TotalNumber of Individual Victims for a given year is often smaller in this detailed data than

    it is in the value for the similarly titled variable presented in FBI summary data at the

    UCR website. Field 13) is also less for a given year than the total derived by simply

    adding the individual victims for each individual offense within an incident. The latter

    approach may overstate the true number of victims, but we have calculated it and

    included it side-by-side with the original total number of victims for comparison and

    audit purposes.

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    18/27

    18

    15)Vics Difference: the difference between the Original ASCII Victims Total (or TotalNumber of Individual Victims, or TotalVics in the codebook and original ASCII

    code) and the newly recalculated NewVictims Total

    16)Total Offenders: the total number of offenders in the incident (pg. 19)17)Offenders Race (pg. 19)

    The following 5 fields repeat for each individual offense reporting within a single

    incident:

    18)Bias Motivation: for example: Anti-Black, the nature of the hate crime bias (pg. 23)

    19)Types of Victims: for example: individual, business, or society/public (pg. 24)20)UCR (Uniform Crime Reporting) Offense: the UCR criminal offense type (pg. 20-22)21)Location Code: the location where the offense occurred (pg. 22)22)Number of Victims: the number of victims pertaining to that individual offense within the

    reported incident (pg. 22)

    The last 3 fields appear once:

    23)State Code (brought over from Batch Header data): the 2-digit code used by the FBI toidentify the state (pg. 5)

    24)FIPS Code (brought over from Batch Header data): Federal Information ProcessingSystem County Code (the codebook refers the user to FIPS Publication 55 for a legend

    for these county codes) (pg. 16)

    25)State Abbreviation

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    19/27

    19

    Mini-Tutorial For the Excel Novice: Using Filters for the Beta-Version Hate Crime Incident

    Report Spreadsheet

    The combined dataset of all hate crimes from 1991 to 2011 is large. Academic

    researchers and interested citizens analyzing crime may benefit from isolating the data by

    attribute. One simple way to sort the data is to use filters in Microsoft Excel. The downloadable

    data spreadsheet we have prepared is pre-sorted first by year. The spreadsheet has the

    Autofilterfunction of Microsoft Excel already applied. This feature can be turned off by

    clicking on the Data tab on the top of the screen (for Excel 2010) and then deselecting the

    Filter option. With filters left on, click on a drop-down arrow and select and deselect data to

    be shown as required. Below is a graphic showing how to restrict the data to show only 2004:

    Simply click on the drop-down arrow for the Incident Date (column A) and check only the

    2004box, and the search criteria is restricted to only those incidents occurring in 2004. The

    data is provided with incidents sorted by year, and within each year, by incident number. It is

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    20/27

    20

    important to remember that if the user has multiple years showing, and then sorts by incident

    number or some other factor, that data for each year will be intermingled, and each year will

    cease to be in contiguous order if the spreadsheet is saved. In order to resort the data by year,

    remove all restrictions on non-date filters, and then resort again by year (sort oldest to newest).

    Using the Tool for Verification of UCR ASCII Source Data Compared to FBI UCR Online

    Summary Reports

    Maltz and Targonski note that the UCR is a voluntary program. Local agencies are not

    required to submit any type of criminal data. Thus, the FBI has no control over the reliability,

    accuracy, consistency, timeliness, or completeness of the data they receive.22 As a

    consequence, some agencies may submit data for only a portion of a year. In order to keep data

    comparable from one year to another, the FBI often imputes full year trends off of a portion of a

    years reported data. That is, they extrapolate the rate for a portion of the year out to the full

    year.

    This combined data tool allows the researcher to compare the detailed incident reports to

    the summary reports posted online at the FBI UCR website for hate crimes. The FBI summary

    reports (available in spreadsheet format only from 2004-2011, and in pdf files for earlier years)

    always state that the number of hate crime victims and hate crime offenses for each year are

    identical.23 In the source data, the number of victims and offenses are not identical.

    22Maltz elaborates why so much data never makes it to the federal database: natural disasters, budgetary

    restrictions, personnel changes, inadequate training, and conversion to new computer or crime reporting systems

    all have affected the ability of police departments to report consistently, on time, completely, or at all. And some

    agencies may not fill out crime reports simply because they rarely have any crime to report. Maltz and Targonski,

    A Note on the Use of County-Level UCR Data, p. 299.

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    21/27

    21

    The data appears to be aligned or normalized before being presented in the online

    summaries, in a process that is not explained to the public user of the online summary data. A

    possible cause of the discrepancy may come from the ambiguity of recording multiple offenses

    within a single incident, and how to count the victims in the total victims figure. The source

    data total victims number is always smaller than the equivalent figure in the FBI summary

    report. For example, in the FBI public summary report for 2004, there were 9.514 total victims.

    However, if one adds the total victimsfield for each incident in 2004, first by selecting only

    2004for Incident Dateand then selecting the Original ASCII Victims Total column

    (originally named Total Victims in the codebook), the sum given at the bottom right of the

    Microsoft Excel window indicates that there were only 5,390 victims.

    Sometimes the total could be understated on the part of the local reporting agency. In the

    source data, some incidents have an Original ASCII Victims Total (total victims) number

    much smaller than the sum resulting from adding the number of victims named in each separate

    offense within that single incident. It is unclear whether or not the true number of total victims

    for an incident should be computed as the sum of the number of victims for each offense within

    that incident, or if some of the number of victims columns are redundant. Pending further

    clarification on the actual process, the data could be interpreted as showing the same victims

    being counted multiple times, if there are multiple offenses, for a single incident. It appears to be

    equally unclear, or at least inconsistent for local reporting agencies as well. Logically, one can

    hypothesize several scenarios where the number of victims and offenses would not be identical

    every year. One offense could have any number of victims, or there could be many offenses

    with a single victim. For example, a man threatened John Doe, kicked a small dent in his car,

    shook him when John Doe exited his car, and called him a racial epithet. There are several

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    22/27

    22

    offenses: intimidation, simple assault, destruction of property, all resulting from a single

    incident. If one is supposed to sum the victim totals for each offense in the incident data, this

    counts John Doe as a victim three times. This is not intended as a normative evaluation of such a

    process, if it is indeed the process being followed, but it is unclear from the data if this is what is

    happening at the local reporting agency level, or if it is what is intended by the FBI as the correct

    methodology, or if some local agencies are recording the total number of victims this way, and

    others are following some other unclear approach.24

    If we instead compute the total number of victims by adding each of the number of

    victims subtotals for each offense, the total is actually the same as the FBI summary data, at

    9,528. As a result of this tracing process for 2004, we decided to add a new column that

    computes the total number of victims by adding the number for each offense.

    The database enables many other types of research. For example, now hate crime totals

    for each locality for 19912011 can be recomputed on a per capita basis from a single source,

    as the data tool incorporates the population totals from the locality as reported by the last census

    within each incident row. Basic population-indexed rates of change for offenses against each

    vulnerable group can now be computed by any user, as well as a more robust set of summary

    statistics.

    Omissions in Provided Documentation

    According to the 1995 edition of the Hate Crime Yearly Master Record Description, or

    codebook, bias motivation fields for each hate crime are coded 11 -15 for Anti-Racial biases,

    24The current form for local agencies to fill out a hate crime incident report is located at:

    http://www.fbi.gov/about-us/cjis/ucr/reporting-forms/hate-crime-incident-report-pdf.

    http://www.fbi.gov/about-us/cjis/ucr/reporting-forms/hate-crime-incident-report-pdfhttp://www.fbi.gov/about-us/cjis/ucr/reporting-forms/hate-crime-incident-report-pdfhttp://www.fbi.gov/about-us/cjis/ucr/reporting-forms/hate-crime-incident-report-pdf
  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    23/27

    23

    21-27 for Anti-Religious, 32-33 for Anti-Ethnicity/National Origin, and 41-45 for Anti-Sexual.

    However, codes for Anti-Disability (51 for Anti-Physical Disability and 52 for Anti-Mental

    Disability) are used in the Hate Crime data, but no explanation of these codes is provided in the

    codebook (we have recoded them into the appropriate text in the streamlined data tool). The

    actual definitions are included in the more recent Hate Crime Data Collection Guidelines and

    Training Manual, (2012) from the Criminal Justice Information Services Division, FBI, at

    http://www.fbi.gov/about-us/cjis/ucr/hate-crime/data-collection-manual. The FBI Hate Crime

    Technical Specification, Version 2.1 (dated 2012), also provides an explanation for these codes.

    The new technical specification is helpful in solving some of the unexplained discrepancies

    found in the older codebook that the FBI emails in response to public requests. However, the

    new specification is intended for submissions from 2012 and after, where the data is in a slightly

    different format and the strings of digits are slightly longer (and so yet again, it is near-

    impossible to create a dictionary file in STATA). Even the 2012 documentation cannot solve

    some issues related to the original codebook mentioned here.

    For bias motivation, code 31 (Anti-Arab) is still occasionally usedas a bias motivation

    code incorrectly by local reporting agencies, even though it was eliminated by the FBI (and

    presumably replaced with some other bias category). These potentially confusing and

    unexplained overlaps between ethnicity and different religions as bias categories are not

    addressed in the codebook either.

    A third coding inconsistency is the inclusion of the 23* field, presumably a larceny

    offense, as all larceny offenses start with a UCR code of 23. However, there is already a

    separate All Other Larceny category, and the 23* offense is not mentioned in the codebook

    (nor the 2012 Hate Crime Data Collection Guidelines and Training Manual or the 2012

    http://www.fbi.gov/about-us/cjis/ucr/hate-crime/data-collection-manualhttp://www.fbi.gov/about-us/cjis/ucr/hate-crime/data-collection-manualhttp://www.fbi.gov/about-us/cjis/ucr/hate-crime/data-collection-manual
  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    24/27

    24

    Technical Specificationnoted above). We have recoded these entries as Additional Larceny

    Offense.

    Another inconsistency occurs within the Offenders Race data category. According to

    the 1995 codebook,both Black and Unknown are coded as B. This may overstate the

    statistic of African-American hate crime offenders, as entries intended as Unknown (U) may

    have been coded as B for Blackand counted accordingly. While every year (1991-2010) has

    a number of U entries for offenders race, this does not necessarily prove or disprove that some

    B entries may have been intended as U. There remains apossibility that the number of

    African-American hate crime offenders is overstated, if local agencies coded entries based on

    incorrect guidelines provided in the codebook. There is no such typo in the updated

    documentation, but that does not prevent there from being incorrectly coded entries in the old

    data.

    The Hate Crime Record Description (codebook) states on page 7 that for the Batch

    Header (agency and other area demographic information) data, there is a placeholder from digit

    14 to 25 for incident number. This blank space is intended to ensure that the batch and

    incident data, while regrettably organized top-to-bottom, rather than right-to-left, will at least

    line up by column width. Unfortunately, none of the data has this placeholder in the batch

    data. Digit 14 is actually the beginning of the date the agency began submitting information to

    the FBI, while in the codebook, it tells the user that this information begins at digit 26. No

    documentation is provided that addresses this inconsistency. The user has to deduce it without

    assistance. So, if one intended to use a formal dictionary file in a program like STATA, the

    codebook would again give you an incorrect guide. While the 2012 update to the codebook

    (which the user has to find themselves, as the FBI only provides the 1995 version) provides

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    25/27

    25

    definitions for several value labels not defined in the 1995 codebook, it also follows a different

    column-width-digit definition. This is why we had to go through the painstaking process of

    setting column widths manually, because the data often strays slightly from the codebook.

    The agency indicator field, described on page 12 of the codebook, indicates whether

    the reporting agency is a city (1), county (2), university or college (3), state police (4),

    or is covered by another agency (0). However, several agencies have the number 5, 6, or 7 in

    this field. There is also no explanation in the 2012 documentation.

    The type of victim field is an 8-character field with the letter B standing for a

    business, I for an individual, and so on (the list is on page 24 of the codebook. However,

    often the entry for this field consists of multiple applicable initials, separated by different

    numbers of spaces. Once again, no explanation is provided in the codebook. These are most

    likely not instances where the multiple victim types were intended as different individual victims

    associated with different offenses, because there are already separate data entries for separate

    offenses, and these multiple victim types in question are listed within the same single offense.

    Offenses with multiple victim types listed have been recoded as though the multiple initials are

    not an error. So, B, G is now recoded as Business/Government, and I, G, S as

    Individual/Government/Society/Public, and so on. This problem is specific to the 1991-2011

    data, and so the Data Collection Manual and 2012 Technical Specification can provide no

    assistance.

    The Population Group codes (listed on page 8 of the codebook) pertain to the type of

    population district (size and type) covered by the ORI. However, no definition for the code 2C

  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    26/27

    26

    is listed in any document we found, and such a code appears in the data throughout. These

    entries have simply been left as 2C.

    The Location Codes in the codebook only cover 01 25, but there are entries from 37

    50 in the data. In order to correctly interpret these entries, one needs to track down the current

    form for recording hate crime incidents, which includes the correct codes for 3750, and is

    located at:http://www.fbi.gov/about-us/cjis/ucr/reporting-forms/hate-crime-incident-report-pdf.

    Looking Ahead: The UCR Redevelopment Project

    Many of the issues created by using a difficult to interpret format like ASCII for

    submissions of UCR data from source agencies can be solved with a faster, cheaper, less paper-

    dependent format, such as Extensible Markup Language, or XML. The FBI has begun to

    implement an XML-based reporting solution to help check submissions efficiently for basic

    logical tests (such as the total of the number of victims agreeing with the sum of the number of

    victims for each individual offense). An XML-based system will also enable easier creation of a

    more powerful data query tool similar to the one which we have created here for the legacy data.

    However, there is no mention on the FBI website as to how the legacy data will be converted or

    integrated with the new format. Failure to do so will result in the loss of the ability to research

    crime data for those years (1991-2011), at least within a single tool, and with the format

    standardized. Also, streamlining the data filing/collection, verification, and publication

    processes does not address any ambiguity for local source agencies as to how certain data should

    be counted (validation).

    Regrettably, a March 4, 2013 update notes that work on the contract has stopped. The

    System Test Readiness Review (S-TRR), planned for 10/29/2012 has not been completed, and

    http://www.fbi.gov/about-us/cjis/ucr/reporting-forms/hate-crime-incident-report-pdfhttp://www.fbi.gov/about-us/cjis/ucr/reporting-forms/hate-crime-incident-report-pdfhttp://www.fbi.gov/about-us/cjis/ucr/reporting-forms/hate-crime-incident-report-pdfhttp://www.fbi.gov/about-us/cjis/ucr/reporting-forms/hate-crime-incident-report-pdf
  • 7/27/2019 E-Government for Uniform Crime Reporting Data and How to Make it More Accessible

    27/27

    delays in the delivery schedule may conflict with the 11/18/2013 contract expiration date. These

    delays have led the FBI to abandon for the moment the concurrent project of ensuring that the

    XML-based filing solution currently under development would be usable for both state and

    federal agencies. Researchers should look for future updates for the new UCR system at the

    UCR Redevelopment Project website.25

    Additional Resources

    The Hate Crime Yearly Master Record Description (codebook)26provided by the FBI

    can be requested from the Criminal Justice Information Services Division. This document also

    contains definitions of the types of data collected by the FBI. More detailed information on each

    of these categories of data presented can be found in the U.S. Department of Justice, Office of

    Justice Programs, Bureau of Justice Statistics, Special Report: Hate Crimes Reported in NIBRS,

    1997-1999 athttp://bjs.gov/content/pub/ascii/hcrn99.txt. Other useful sources include the Hate

    Crime Data Collection Guidelines and Training Manual, (2012) from the Criminal Justice

    Information Services Division, FBI, athttp://www.fbi.gov/about-us/cjis/ucr/hate-crime/data-

    collection-manual. The data submission specifications for local reporting agencies can be found

    atwww.fbi.gov/about-us/cjis/ucr. Finally, the original report establishing the purpose of the

    modern NIBRS system is located athttps://www.ncjrs.gov/pdffiles1/bjs/98348.pdf.

    25http://www.fbi.gov/about-us/cjis/ucr/ucr-redevelopment-project. UCR Redevelopment Project. Retrieved July

    15, 2013.

    26http://www.asucrp.net/FBI%20Manuals%20and%20Addendums.html

    http://bjs.gov/content/pub/ascii/hcrn99.txthttp://bjs.gov/content/pub/ascii/hcrn99.txthttp://bjs.gov/content/pub/ascii/hcrn99.txthttp://www.fbi.gov/about-us/cjis/ucr/hate-crime/data-collection-manualhttp://www.fbi.gov/about-us/cjis/ucr/hate-crime/data-collection-manualhttp://www.fbi.gov/about-us/cjis/ucr/hate-crime/data-collection-manualhttp://www.fbi.gov/about-us/cjis/ucr/hate-crime/data-collection-manualhttp://www.fbi.gov/about-us/cjis/ucrhttp://www.fbi.gov/about-us/cjis/ucrhttp://www.fbi.gov/about-us/cjis/ucrhttps://www.ncjrs.gov/pdffiles1/bjs/98348.pdfhttps://www.ncjrs.gov/pdffiles1/bjs/98348.pdfhttp://www.fbi.gov/about-us/cjis/ucr/ucr-redevelopment-projecthttp://www.fbi.gov/about-us/cjis/ucr/ucr-redevelopment-projecthttp://www.fbi.gov/about-us/cjis/ucr/ucr-redevelopment-projecthttp://www.asucrp.net/FBI%20Manuals%20and%20Addendums.htmlhttp://www.asucrp.net/FBI%20Manuals%20and%20Addendums.htmlhttp://www.asucrp.net/FBI%20Manuals%20and%20Addendums.htmlhttp://www.asucrp.net/FBI%20Manuals%20and%20Addendums.htmlhttp://www.fbi.gov/about-us/cjis/ucr/ucr-redevelopment-projecthttps://www.ncjrs.gov/pdffiles1/bjs/98348.pdfhttp://www.fbi.gov/about-us/cjis/ucrhttp://www.fbi.gov/about-us/cjis/ucr/hate-crime/data-collection-manualhttp://www.fbi.gov/about-us/cjis/ucr/hate-crime/data-collection-manualhttp://bjs.gov/content/pub/ascii/hcrn99.txt