For example, the United States Census combines
their population

data with crime data compiled from the FBI's Uniform Crime
Reporting system (UCR)

to produce crime rates. These rates are calculated
with a constant of 100,000

and are broken down by type of crime. As described
by Sacco and Kennedy in The

Criminal Event, the constant in the above
equation is designed so the crime rate

can be expressed per c (the constant)
units of exposure. The selection of the

constant is largely arbitrary, but
rates are traditionally expressed per 1,000

or per 100,000 units of exposure.
It is, although, most important to choose a

value of c so the resulting rates
will be meaningful to the reader. For example,

if you are calculating crime
rates for a town of 10,000 people, and you choose a

value of c * 10,000, the
results will be expressed in terms of a population

greater than the town.
Consequently, the results will not be meaningful to the

town residents. On
the other hand, if a value of 100 were chosen, a resident

reading the rates
would see how many people in a group of 100 would be likely to

be a victim of
a given crime. In contrast to the constant, which is arbitrary,

and the
numerator which is provided by observations and reports, the denominator

must
be carefully determined for each type of crime. To determine this

denominator
the target population must be established. The target population can

be
defined as "the population we want to talk about" (Mosteller,

Fienberg,
and Rourke, 1983). "The usual practice is to use the total

population
residing within the jurisdiction in which the count of events has

taken
place" (Sacco and Kennedy, 1996). Although, this method does not take

into
account "the demographic characteristics of the population or the

empirical
fact that different segments of the population are involved in

criminal
events with different levels of frequency" (Sacco and Kennedy,

1996).
Segments of the population can be created by looking at cross sections
of

age, sex and potential for victimization. Members of these segments may
have a

higher probability to being exposed to a particular crime then members
of other

segments. For example, people who spend a large amount of time
outside their

homes in the evening may have a higher potential of being a
victim of a crime

such as aggravated assault. If the entire population were
used in the

denominator, the calculated rate of aggravated assaults would be
artificially

high for the general population and artificially low for the
segment of people

who go out in the evening. In summary, when reviewing crime
rates, the value of

the denominator should be considered when drawing
conclusions. If the

denominator reflects the size of the total population
residing within the

measured jurisdiction, conclusions should reflect the
possibility that subsets

of the population may be incorrectly included in the
calculated crime rate. If a

subset such as males aged 16 - 30, were used in
the calculations, conclusions

should consider the subset of the population
which may be excluded.