According to Bless et al. (2006), sample size is the part of the whole population used by a researcher and whose features and characteristics can be inferred to the whole population. This is supported by Fellows and Liu (2015) who add that the purpose and context of the research largely determines the sample size. The authors go on to add that sampling allows the collection of data and subsequent parts of the research to be done with ease while still providing a good representation of the population (Fellows and Liu, 2015). In this study, a total of 138 participants were selected as the sample size.

Several sampling techniques are available to researchers and these can be largely reduced to two areas, namely probability and non-probability sampling techniques (Chen, 2003). In probability methods, such as random sampling, all members of the population have an equal chance of being selected and taking part in the study (Chen, 2011). Usually, random sampling occurs without replacement, hence, the members of the population can only be selected once (Fellows and Liu, 2015).

**3.6.1 Probability sampling **

Kothari, 2008 defines probability sampling as random sampling and or chance sampling. In this kind of sampling each item of the universe forms an identical chance of being included in the sample (ibid). Cooper and Schindler (2008) argue that this based on the conception of random selection and organised technique which guarantees that each population element is given a known non-zero chance of selection. Bryman and Bell (2011) further argue that the

aim of such technique is generally to keep sampling error to a minimum. According to Chen (2011), the motivation for probability sampling is that it allows for optimum representation especially when there are cost, time and logistical constraints. Probability sampling has various techniques such as:

Systematic sampling: According to Taherdoost (2016), stratified sampling is used when the researcher selects every nth case after the random start has been chosen. On the contrary, Etikan and Bala (2017) argue that in systematic sampling, the researcher selects only the first unit randomly, while the remaining units of the sample to be selected are fixed. This sampling method has confident points of having improvement over other sampling techniques, as ample the systematic sample is feast more equally completed to the entire population. A systematic sampling is “spreads the population more evenly over the population” (Sharma, 2017: 750).

**Simple random sampling: Taherdoost (2016) postulates that the simple **
random sampling allows every case of the population to have an equal
probability of inclusion in the sample. Thus, a simple random permits every
single item in the universe the probability of being part of the sample.

**Stratified sampling: The stratified sampling is the complex technique among **
all the probability sampling techniques. Sharma (2017) explains that stratified
sampling involves the process of dividing the population into smaller groups
called “strata”. A random sample is then taken from each of the stratum. A
stratified sampling reduces the potential of human error or bias in the selection
of the individual cases to be included in the sample.

**Cluster sampling: In this sampling method, all the natural occurring groups are **
chosen as samples (Sharma, 2017). The scholar suggests that all probability
sampling techniques require sampling frames of all the sampling units with the
exception of the cluster sampling.

Out of the various probability sampling techniques, the systematic sampling was used to select 130 respondents for the quantitative study. The systematic sampling technique was used because it was easier and simple to conduct when compared to the simple random sampling.

**3.6.2 Non-probability sampling **

Cooper and Schindler (2008) argue that non probability sampling is different in that it is a non- random and subjective. Such sampling does not give basis for approximating the probability that each item in the population has of being included in the sample. Bryman and Bell (2011) argues that such a sampling generally surveys one individual in the organisation.

• **Convenience sampling: Cooper and Schindler (2008) indicates that this form of **
sampling is easy to conduct and the cheapest. The researchers have the freedom to
choose whoever they can find. Convenience samplings are the slightest trustworthy
design.

• Purposive sampling: Judgement sampling is used commonly in qualitative research where the wish happens to develop hypotheses rather than to simplify to larger populations (Kothari, 2008). There are two major types of purposive sampling, judgement sampling and quota sampling. Judgement sampling happens when a researcher sample members to follow to some standard (Cooper and Schindler, 2008). The researcher’s decision is used for choosing items which he considers as representative of the population (Kothari, 2008).On the other hand, quota sampling is used to advance representativeness (Cooper and Schindler, 2008). When using quota sampling, the researcher has an ease access to the sample population (Kumar, 2011).

• Snowball sampling: Snowball sampling is the procedure of choosing a sample by means of networks (Kumar, 2011). Snowball sampling is convenient when the researcher wants to sample subjects that are difficult to identify (Cooper and Schindler, 2008).

From the various non-probability sampling technique, the purposive sampling technique was used to select eight participants for the qualitative study.