As mentioned earlier, the purpose of taking measurements from a sample of a population is so that you
can use it to perform inferential statistics, which enables you to make estimates about the population
without having to measure the entire population. Theoretically, you want the statistics from your
sample to be as close as possible to the population parameters you are trying to estimate. To increase
the likelihood that this happens, you should try your best to draw a sample that is representative of the
population.
You may be wondering, “What is the best way to draw a sample that is representative of the
background population?” The honest answer is, “It depends on your resources.” If you are a
government agency, you can invest a lot of resources in conducting representative sampling from a
population for your studies. But if you are a graduate student working on a dissertation, then based on
resources available, you probably have to settle for a sample that is not as representative of the
population as a government agency could afford. Nevertheless, you can still use your judgment to make
the wisest decisions possible about your sampling approach.
Taking a simple random sample
Taking a simple random sample (SRS) is considered a representative approach to sampling from a
background population. In an SRS, every member of the population has an equal chance of being
selected randomly and included in the sample. As an example, recall the printout of the current patient
list from a clinic discussed in the previous section. Considering that list a clinical population, imagine
that you used scissors to cut the list up so that each name was on its own slip of paper, and then you put
all the slips of paper into a hat. If you want to take an SRS of 20 patients, you could randomly remove
20 names from the hat. The SRS would be seen as a highly representative sample.
In practice, an SRS is usually taken using a computer so that you can take advantage of a
random number generator (RNG) (and do not have to cut up all that paper). Imagine that the
patient list from which you were sampling was not printed on paper, but was instead stored in a
column in a spreadsheet in Microsoft Excel. You could use the following steps to take an SRS of
20 patients from this list using the computer:
1. Create a column containing random numbers.
You could create another column in the spreadsheet called “Random” and enter the following
formula into the top cell in the column: =RAND(). If you drag that cell down so that the entire
column contains this command, you will see that Excel populates each cell with a random number
between 0 and 1. Each time Excel evaluates, the random number gets recalculated.
2. Sort the list by the random number column.
3. Select the top 20 rows from the list.
This process ensures that your sample of 20 patients was taken completely at random. Statistical
packages like those described in Chapter 4 have RNG commands similar to the one in Excel.