Methodology & Definitions

Data Source

FBI Return A Masterfiles serve as the source of the SOCDS crime data system. These files consist of Uniform Crime Reports (UCRs) which include monthly crime data submitted by law enforcement agencies to the FBI. The UCR crime category, offenses known to law enforcement agencies, serves as the basis for SOCDS crime data. SOCDS divides these offenses into two categories: violent crime (murder, rape, robbery, robbery with gun, aggravated assault, assault with gun), and property crime (burglary, larceny, and motor vehicle theft).

Estimation

A significant portion of law inforcement agencies do not submit crime data for the full year (22% of CBSA agencies in 2005). SOCDS provides annual estimates of crime for partial responding or non-responding agencies. SOCDS crime estimation is adopted from a technique developed by the ICPSR (Inter-university Consortium for Political and Social Research) in 1994 and is comparable to the process created by the FBI . In this approach, the monthly frequency of response determines the type of estimate used: response type 1 (3-11 months reported), response type 2 (0-2 months reported).

Response type 1

Response type 1 estimates are simple annual proportional projections of crime based on reported monthly crime data.

Offense X:

Response type 1 annual crime = (12/number of months reported)*(sum of crime for months reported)

Response Type 2

Response type 2 estimates for incomplete responding agencies are calculated from the annual crime activity reported by 12 month responding agencies of similar population group size for each state. Reported crime for the incomplete responders is not used in the estimation process. The following denotes the four population groups used in estimation:

1) city agencies with greater than or equal to 250,000 inhabitants
2) city agencies with less than 250,000 inhabitants
3) non-metropolitan county agencies
4) metropolitan county agencies

First, the estimation algorithm aggregates crime for 12 month responding agencies within each state population group. Crime rates (crime/population) are then calculated for these state population group aggregations. These crime rates, applied to the population of those individual incomplete reporting agencies within the same state population group, yield agency crime estimates for each offense.

Offense X:

Response type 2 annual crime = state population group crime rate * population of incomplete reporting agency

Response type (2) estimates were not calculated for agencies in the following states since few agencies in these states provided the FBI with crime data for all 12 months:

2005: Illinois*
2004: Illinois*
2003: Illinois*, Kentucky*
2002: Illinois*, Kentucky*
2001: Illinois*, Kentucky*
2000: Illinois*, Kansas*, and Kentucky*
1999: Illinois*, Kansas*, Kentucky*, Montana*, and New Hampshire*
1998: Illinois*, Kansas*, Kentucky*, Montana*, New Hampshire, and Wisconsin
1997: Illinois*, Kansas*, Kentucky*, Montana*, New Hampshire*, and Vermont*
1992: Alabama, Kentucky

* many agencies within these states did not provide the FBI with data in accordance to UCR guidelines.

Gun Crimes

Gun crime calculations do not include estimates and thus represent complete (12 month) response. Please note that many agencies, defined in all other crime categories as complete responders, do not provide a gun crime breakdown of criminal activity.

Data Comparisons

It is important to note the agency response rate of metropolitan areas and central cities presented in this web data system when comparing crime across time. Varying response rates seen in a particular area could inaccurately distort the area's crime time trend. The metropolitan area response rate, shown in the SOCDS crime output page, is divided into four classes of response: complete response (12 months), incomplete response (3-11 months reported) or response type 1, incomplete response (0-2 months reported) or response type 2, and not available. The total agency population and total agency count serve as the basis for calculating the response rate of the metropolitan area. SOCDS includes a response rate for gun crimes based on agency population. The SOCDS crime output page also provides the type of response received by the central city agency of the MSA. Suburban response, not illustrated in SOCDS, should be viewed as a composite of the metropolitan area and central city response.

Data Differences

The data presented in the SOCDS crime webpages may differ from crime statistics published in FBI documents due to several factors:

1) FBI staff update UCR agency data on a continual basis even after data publication.
2) HUD employs a static 1999 definition of MSAs and their components for all years presented on this website. FBI's "Crime in the United States" publications provide MSA definitions and components contemporaneous with the publication year.
3) HUD's estimation technique for incomplete agency response approximates but does not completely replicate the ICPSR and the FBI estimation procedures.
4) The FBI UCR Return A Masterfile Data serves as the sole source of data used in the development of SOCDS Crime Data. Data, external to the Return A Masterfile, is not included in SOCDS Crime.

Data Notes

SOCDS reports crime rates as not available "N/A" in the cases where agencies did not provide the FBI with crime data and estimation was not performed.

SOCDS Crime Home

Send comments or report errors to: John_I._Carruthers@hud.gov