Probability Sampling Techniques

• The family of probability sampling methods commonly used by

quantitative researchers includes, simple random sampling,
systematic sampling, stratified random sampling, multi-stage sampling
and cluster sampling which are discussed below.
• A common trait which runs through all these techniques is that they
seek to achieve a ‘perfect’ sample.
Simple Random Sampling
• This method of sampling attaches an equal chance of being selected to every member
of the population (Oakshott, 2014).
• Another requirement attached to this method is that a sampling frame must be
available and a method to be used to randomly select the sample should be in place.
• According to Saunders et al (2009), this is a probability sampling procedure that ensures
that each case in the population has an equal chance of being included in the sample. It
is the simplest of all probability sampling techniques.
• A basic and commonly cited example of this sampling technique is a ruffle or lotto draw.
• In more formal processes, random numbers (randomization) are used and are
considered equally likely.
• This approach is however prone to sampling error which may compromise the
representativeness of the sample.
Steps followed when using simple random
sampling
• Step 1: Define the population
• Know every member of the population under study.
• Step 2: Determine the sample size
• Once the population is known, decide on the size of the sample.
Larger samples provide certainty and confidence in results
• Step 3: Randomly select your sample
• Step 4: Collect data from your sample
Stratified Sampling
• This sampling method according to Oakshott (2014) draws from the
reasoning that populations can be divided into different categories.
• This technique, according to Saunders et al (2009) is a modification of
random sampling in which you divide the population into two or more
relevant and significant strata based on one or several attributes and
draw a random sample by either simple or systematic sampling.
• Ideally, stratified sampling co-joins the advantages and disadvantages
of simple random or systematic sampling.
• The population of a province for example may consist of voters
represented by both females and males, with different employment
statuses, belonging to different ethnic groups and located in different
geographical locations among other categories.
Stratified Sampling Continued
• Dividing the population into different categories emerges from the thinking that these
categories have potential to influence the responses hence the need to contain all
responses from each category in the correct proportion.
• Implied in this reasoning is that if a sample represents a segment of the population,
then the views drawn from the sample should reflect as close as possible what the
entire population would have presented as its views.
• This argument is located in the assumption of a ‘perfect sample’ that probabilistic
researchers hold. It is held that to obtain a ‘perfect’ or representative sample, requires
one to separate the categories and draw a random sample from each category.
• This method is acknowledged as a very reliable method but is criticized for assuming
that the researcher has knowledge about the categories of the population, which is
not the case in many instances.
Multi – Stage Sampling

• This sampling method is reliable when studying wide geographical

areas where respondents have to be picked from different parts of
the area under study.
• According to Saunders et al (2009) this technique is also known as
multi-stage cluster sampling and it is a development/improvement of
cluster sampling.
• Researchers use this technique to overcome problems associated with
a geographically dispersed population when face-to-face contact is
needed or where it is expensive and time consuming to construct a
sampling frame for a large geographical area.
Multi – Stage Sampling Continued
• Oakshott (2014) notes that this usually complicates the selection of the sample
and to overcome the challenges presented by a large geographical area, the area is
divided into smaller areas for example into provinces, which are then randomly
selected.
• The selected areas can also be further divided into smaller areas such as districts
and the random selection procedure is repeated again.
• This process can be repeated until the area is small enough for a random sample
to be selected. The random samples chosen from each area must be the same
proportion of the population to avoid bias.
• Similarly, Saunders et al (2009) proposes a four phased approach to multi stage
sampling to achieve the best results. It is however noted that bias is inevitable due
to the concentration of responses from the people in the same area.
Cluster Sampling

• This method is also widely seen as similar to multi-stage sampling (Oakshott, 2014).
• Henry (1990) cited in Saunders et al (2009), also accepts that cluster sampling is, on
the surface, similar to stratified sampling as one needs to divide the population into
discrete groups prior to sampling.
• Usually, the clusters can be based on naturally occurring grouping in the geographical
area under study.
• According to Oakshott (2014), this sampling method is used when a sampling frame is
not available. It is also a reliable sampling method when a researcher is dealing with a
large geographical area.
• When using cluster sampling in such a scenario, the researcher has to divide the large
area into smaller areas called clusters which are small enough for all members to be
surveyed.
Cluster Sampling Continued
• It is further noted that where cluster sampling is used, all members of the cluster
constitute the sample because it is not possible to identify all members of the
cluster in advance hence random sampling is not possible.
• Saunders et al (2009) note that for cluster sampling the sampling frame is the
complete list of clusters rather than a complete list of individual cases within the
population.
• The researcher has to select a few clusters, normally using simple random
sampling and data are then collected from every case within the selected clusters.
• Accordingly, this technique has three main stages which include: choosing the
cluster grouping for the sampling frame, numbering each of the clusters with a
unique number starting from 0, then 1 and so on and selecting the sample using
some form of random sampling
Systematic Sampling

• According to Saunders et al (2009) Systematic sampling involves selecting the sample at

regular intervals from the sampling frame.
• This method of sampling is used with a sampling frame but applies even in situations where
the sampling frame is not available, but the size of the population must be known.
• Accordingly, Oakshott (2014) notes that with systematic sampling, every n th member of a
population is selected, and the value of ‘n’ is determined by the size of the population and
the required sample size using a formula that determines the frequency of selection.
• A good example would be that if a 10% sample is to be selected from a population of 2000
university students, then every 10th person will be selected.
• However, this method requires that the first member of the sample is randomly selected
from the first 10 members of the population such that if the 3 rd member is picked, then
using an interval of 10, the 13th member would be selected. This method is a very quick and
efficient way of obtaining a random sample.