e for its content.
Appendix C-Estimation Procedures Appendix C

Estimation Procedure


The Michigan estimation procedure was developed based on the process used on the national level by the
Federal Bureau of Investigation (FBI). The following decisions were made and followed in the estimation
procedure:
-Crime will be estimated for both offenses and arrests
-Crime will be estimated to the law enforcement agency level only for the purpose of calculating
state totals, not for use at the agency level
-Only state total estimated figures will be published in the annual report
-Population figures used in cross sectional estimating will come from the U.S. Census
-Arrest data will be estimated only for the total, not for the breakdowns of age, race, and sex
-Agencies that are classified as zero population (see Appendix B for definition) that fall into the
cross sectional estimation method will not be included in the estimation


Three different methods are used, depending on the extent of missing data:

Averaging Method
Definition: Taking the existing months of data, averaging it, and using the averaged figure for the
remaining months of missing data.

When Used: When six or more months of actual data exists.

Steps:
-Identify agencies that have reported six or more months of data for the year.
-For each arrest and each offense take the numbers reported, add them together, and divide by the number
of months reported.
-Multiply this averaged figure for each arrest and offense by 12.


Longitudinal Method
Definition: Taking the same agencys data from last year, calculating the rate of change between the two
previous years, applying this rate of change to last years data, using these calculated figures for the current
year.

When Used: When an agency has reported less than six months of data for the year, but reported 12
months of data the previous two years.

Steps:
-Identify all the agencies that have reported less than six months of data for the current year but reported 12
months of data the previous two years.
-Calculate the rate of change between the two complete years.
-Disregard any data reported during the current year.
-Apply the rate of change from the two complete years to the most recent complete years data to calculate
the current years estimate.







2003 Michigan Uniform Crime Report
Page 164
Appendix C

Estimation Procedure



Cross Sectional Method
Definition: Taking averaged crime rate data from a group of similar agencies and applying this figure to
agencies with missing data.

When Used: When an agency has reported less than six months of data for the year, and did not report 12
months of data for the previous year.

Steps:
-Identify all agencies with 12 months of data reported for the current year.
-Place all agencies with full data for this year into the appropriate population group, based on their
population.
a)

up to 999
b)

1,000 2,499
c)

2,500 4,999
d)

5,000 9,999
e)

10,000 24,999
f)

25,000 49,999
g)

50,000 99,999
h)

100,000 249,999
i)

250,000 +
-For each agency, calculate the crime rate for each offense and arrest.
-For each agency, for each offense and arrest, add the total rates within each group and divide by the
number of agencies to get an average.
-Apply the averaged figure for each arrest and each offense for those agencies missing data and fitting into
the grouping.





2003 Michigan Uniform Crime Report
Page 165