POPULATION AND SAMPLING

The population and sampling procedures are discussed in this section.

4.5.1 THE TARGET POPULATION

Israel (2008: xvi) defines a study population as the set of all items that are being considered for measuring some characteristics. This is the group of cases the researcher can draw on according to Wilson (2010:306). The target population for this study was all SMMEs in the Naledi Local Municipality in the North West Province of South Africa. It is from this population that a sample of 400 SMMEs was drawn. SMMEs have been defined by different interest groups around the world in diverse ways as has been indicated in Chapter 2 in the literature. Because informal traders have sometimes been classified among SMMEs, it is important to set some criteria for business that have to be included in the study. Such SMMEs should have some degree of permanency and a place for transacting their activities. It is important to define SMMEs in this study as the concept

has been used in diverse ways. Clarifying the concept will ensure that the reader understands what the operational definition is. In this study, SMME refers to a business that employs at least one other person in addition to the owner or manager and has a formal place of operation. The unit of analysis is the SMME.

4.5.2 SAMPLING

Sampling is a process of systematically selecting cases or respondents for inclusion in a research study which will be a representative of the entire population. As will be explained in a later section, the study used the stratified random sampling technique to select 400 respondents for the questionnaire.

4.5.2.1 The sample frame

Sample frame refers to the list of accessible population members from which a researcher can draw a sample (Wilson, 2010:307). Wilson (2010:191) rightly observed that researchers do not always have access to a concrete sample frame and they have to compile their own. The absence of a reliable database for SMMEs in the North West Province was confirmed by Africa (2007:6). To overcome this difficulty, the researcher employed matriculants from the different communities within which the study was conducted to draw up the sample frame. In all, a sample frame of 3522 SMMEs was compiled and used for the study.

4.5.2.2 Sampling procedure/technique

Although there are different sampling techniques for selecting a sample, these can be grouped under two main categories of probability and non-probability sampling. Probability sampling allows the researcher to ensure in advance that each element of the population under study has a known likelihood/chance of being selected (Leedy &Ormrod, 2005:199). Some of the specific types of probability sampling include: simple random

sampling; stratified random sampling; cluster sampling; systematic random sampling; and multistage-sampling. Nonprobability sampling on the other hand does not allow the researcher to ensure that each element of the population has a known chance of being selected (Leedy & Ormrod, 2005:206). Some of the types of nonprobability sampling include: convenience sampling; quota sampling; purposive sampling; snowball sampling; and judgement sampling.

In making the choice between probability and non-probability sampling, Dzansi (2004:183) cites Sekaran (1992) to make the point that the choice is eventually determined by the available time, resources, the population distribution, and the extent of generalisability required.

A non-proportional stratified random sampling technique was used for this study. Blumberg, Coopers and Schindler (2008:244) state that stratified random sampling allows populations to be segmented into several mutually exclusive sub-populations or strata. This technique allows for items to be selected from each sub group to ensure representativity (Wilson, 2010:196). In this study, the strata included four distinct ethnic groups - Tswanas, Afrikaners, Coloureds and Indians. The different ethnic groups differ in numbers, location of their businesses, culture and so on. It is therefore important that a technique that ensures that each segment of the sub-groups is represented is applied. There are two types of stratification namely proportional and non-proportional stratification. As said earlier, non-proportional stratified sampling was used. Was this not the case, there was the possibility that some of the groups could have been over represented while others would have very few or not feature at all. After the SMMEs belonging to the entrepreneurs from the four groups had been stratified with the names of the businesses into different boxes, an assistant was asked to pick a name from the box at a time. Each time a name was selected, it was not returned to the box until the required number for the group was obtained. The same procedure was used for the rest

of the groups in the other boxes until the required sample size for each ethnic group has been reached.

4.5.2.3 The sample size

Determining the sample size for research is complex and depends on a number of factors. Wilson (2010:201) for example points out that sample size to a greater extent is influenced by one's research philosophy. This author argues that if one adopts a positivist stance, then there is a need to select a large sample size enough to allow for statistical analysis. Wilson (2010:202) concludes that one’s choice of sample size depends on:

• The confidence needed in the data, i.e. level of certainty that the characteristics of the data collected will represent the characteristics of the total population;

• The comparative sample size of earlier studies;

• The margin of error that can be tolerated, i.e. the accuracy one requires for any estimates made from the sample;

• The type of analysis to be done; and

• The size of the population from which the sample is being drawn.

Leedy and Ormrod (2005:207) suggest some guidelines when selecting a sample size. They recommend the following:

• Population of less than 100 (one should survey the whole population); • If the population size is around 500, then 50% of it should be sampled;

• If the population size is around 1,500, then 20% is recommended and beyond 5,000 or more units, a sample size of 400 is regarded as enough.

The population in this study is 3522 which makes a sample of 400 to be considered large enough based on the suggestion above.