Every year, the Bureau of Justice Statistics publishes national crime rates. The rate of occurrence is an informative statistic because it takes into account the population of interest. The rate of occurrence is calculated by dividing the actual number of occurrences by the number of possible times the outcome could have occurred. For the purposes of this resource guide, we will use 1999 numbers for all examples. Sample code can be edited to use with any year's data collection.
We will discuss here two types of rates; personal crime rates and household crime rates, each of which can be calculated using two different data formats.
The hierarchical (or "full" file) can be used in two formats. The first of these is the collection year format, which is based on 12 months (4 quarters) of interviews. This format provides a more expedient measure of crime, and contains all crime reported in the year of interview, regardless of the year the crime actually occurred. This data format is used in BJS annual publications, such as Criminal Victimization 1999: Changes 1998-99 with Trends 1993-99.
The second available format is the data year format. It is based on 18 months (6 quarters) of interviews. Given the six month retrospective reference period used by the NCVS, it takes 18 months to record a complete victimization experience. Under the data year format, crimes are counted in the year they occurred, not the year they were reported. Consequently, the data year format is considered more accurate. When crime is relatively stable, however, the difference between crime rates calculated using collection year and data year is small. This data format is used in BJS special reports.
NACJD archives the data year format of the NCVS. The collection year data can be created from the data year file by selecting for records that record an interview in one of the first four quarters of interviews, and writing only these records to a new (collection year) data file.
A rate is simply one number divided by another. The NCVS makes a distinction between crimes against persons and crimes against households. BJS publications typically report crime rates as per 1,000 persons or households.
Personal crime rate = number of person crimes / (N persons / 1,000)
Household crime rate = number of household crimes / (N households / 1,000)
Numerator and denominator are calculated in separate steps, using the NCVS full (hierarchical) file. The full file is the only data file necessary to calculate crime rates. The incident-level file may be used depending on the analyst's preferences. The exact calculation depends on the data format used (collection year or data year.)
Download codebook, full hierarchical data file (6 interview quarters), and SPSS data definitions statements for full hierarchical file.
Create unweighted incident, person and household system (data) files using the hierarchical data file and SPSS statements.
An incident-level file (the numerator file) is the default for the SPSS statements archived by NACJD with the full hierarchical file.
To create a person-level file (a denominator file), comment out all references to incident variables (V4000's).
To create a household-level file (a denominator file), comment out all references to person (V3000's) and incident (V4000's) variables.
Property crime rate = N property crimes / (N households / 1,000)
Sample collection year results:
1999 rate of completed motor vehicle thefts = 807,729 / (107,159,550 / 1,000) = 7.5 per 1,000 households
Personal crime rate = N personal crimes / (N persons / 1,000)
Sample collection year results:
1999 rate of simple assault without injury = 3,662,086 / (224,568,370 / 1,000) = 16.3 per 1,000 persons age 12 and older
The following link connects to code in a text document. The code can be cut and pasted to use with SPSS. Each setup was written for the 1999 data (collection year and data year files) and can be edited to suit your needs.
|SPS File||Used for Generating 1999....|
|count_vic_cy99.sps||Victimization Counts on a Collection Year (4 Quarters) Basis|
|count_vic_dy99.sps||Crime Counts on a Data Year (6 Quarters) Basis|
|count_hh_cy99.sps||Household Count on a Collection Year (4 Quarters) Basis|
|count_hh_dy99.sps||Household Count on a Data Year (6 Quarters) Basis|
|count_per_cy99.sps||Population Count on a Collection Year (4 Quarters) Basis|
|count_per_dy99.sps||Population Count on a Data Year (6 Quarters) Basis|
BJS Annual publications use collection year format (not data year format).