25
subpopulation. In general, stratified sampling produces samples
that are more representative of the population than simple
random sampling if the stratum information is accurate.
4. Cluster Sampling
a. Cluster sampling addresses two problems: Researchers lack a
good sampling frame for a geographically dispersed population
and the cost to reach a sampled element is very high. Instead of
using a single sampling frame, researchers use a sampling design
that involves multiple stages and clusters. A cluster is a unit that
contains final sampling elements but can be treated temporarily
as a sampling element itself. In other words, the researcher
randomly samples clusters, and then randomly samples elements
from within the selected clusters; this has a big practical
advantage. He or she can create a good sampling frame of
clusters, even if it is impossible to create one for sampling
elements. Once the researcher gets a sample of clusters, creating
a sampling frame for elements within each cluster becomes more
manageable. A second advantage for geographically dispersed
populations is that elements within each cluster are physically
closer to one another. This may produce a savings in locating or
reaching each element.
III. Random Digit Dialing
a. Random-digit-dialing (RDD) is a special sampling technique used in research projects in
which the general public is interviewed by telephone. Here is how RDD works in the United
States. Telephone numbers have three parts: a three-digit area code, a three-digit exchange
number or central office code, and a four-digit number. In RDD, a researcher identifies
active area codes and exchanges, and then randomly selects four digit numbers. After finding
and calling a working residential number, a second stage of sampling is necessary, within
household sampling, to select the person to be interviewed. Remember that the sampling
element in RDD is the phone number, not the person or the household.
IV. How Large Should a Sample Be?
a. The best answer to this question is, “It depends!” What does it depends on?
i. The kind of data analysis the researcher plans (descriptive, multiple regression).
ii. On how accurate the sample has to be for the researcher’s purposes (acceptable
sampling error).
iii. On population characteristics (homogenous or heterogeneous, large or small). On
principle for sample sizes is, the smaller the population, the bigger the sampling
ratio has to be for an accurate sample. Larger populations permit smaller sampling
ratios for equally good samples. This is because as the population size grows, the
returns in accuracy for sample size shrink. For small populations (under 1,000), a
researcher needs a large sampling ratio (about 30%). For moderately large
populations (10,000), a smaller sampling ratio (about 10%) is needed to be equally
accurate. For large populations (over 150,000), smaller sampling ratios (about 1%)
are possible to be very accurate. To sample from very large populations (over
10,000,000), one can achieve accuracy using tiny sampling ratios (0.025%).
V. Drawing Inferences
a. The purpose of sampling is to enable a researcher to draw inferences from the sample to the
population. The thing to remember is: probability samples are more likely when compared to
nonprobability samples to yield representative samples of the population. In other words, a
researcher, who wants to draw inferences about the population from his or her sample,
should always try to produce a sample that is similar to the population. If the sample is not
similar or representative of the population in which it was drawn, the ability to make
accurate inferences is highly impaired.
VI. So, Should I Always Use A Probability Sampling Technique?
a. NO! The answer is a little more complicated than that. Besides, if it were that easy to
determine the sampling technique for a study why would there be so many to choose from?
The short answer to this question is: It depends. It depends on numerous factors. This