13 / 762
Chapter 1: Introduction and Graphical Displays
13
It allows use of different sampling techniques for the subgroups.
It reduces the error for estimation. (The concept of estimation will
be covered later in the e-book).
It is efficient.
Disadvantages of a Stratified Sample
It may be difficult to select the appropriate variables for
stratifications.
It is not a useful sampling technique when there are no
homogeneous subgroups.
The stratified sampling procedure can be expensive.
It requires accurate information about the population.
Cluster Sample
Another type of sampling procedure is called cluster sampling. It is often
used to reduce the cost of sampling the population which is scattered across
a large geographic area. For example, if the researcher wanted to estimate
the average cell phone usage in terms of monthly minutes used, in the state
of Kentucky, the researcher may “cluster” the state in terms of counties and
then samples of these clusters are taken.
Definition: Cluster Sample
A cluster sample is obtained by dividing the population into clusters using
naturally occurring geographic or other boundaries. Next, one or more of
the clusters are selected at random and all or random samples of the
members in the sampled cluster or clusters are used for the sample.
Figure 1-7
shows an illustration for a cluster sample. The counties in the
state of Kentucky are considered the clusters. Next, a random sample of
eight counties is selected, and random samples from these eight counties are
then selected. The cluster sample is obtained by combining these eight
random samples.