• No results found

4.4 DATA COLLECTION METHODS

4.4.5 Sampling

Sampling is a major step in quantitative research due to the fact that for most research questions it is not possible to collect data from an entire population. Sample refers to the subset of the population from which evidence is gathered (Easterby – Smith et al, 2012). Sampling is the process of selecting a small number of cases (a sample) from a bigger group (the population) as the basis for estimating or predicting the prevalence of an unknown piece of information, situation or outcome regarding the bigger group (Kumar, 2014). Although the focus of the study is to find answers to the research question as they relate to the total study population rather than the sample, the researcher employs the process of sampling in an attempt to estimate what is likely to be the situation in the total study population. Saunders et al (2012) suggest that sampling provides a valid alternative to a census when it would be impracticable for the researcher to survey the entire population, when the available budget constrains the researcher from surveying the entire population and when time constraints prevents the researcher from surveying the entire population.

157

Although it is possible to collect data from every international student at the university for this study, the cost and time required might be substantial. Sampling allows the researcher to collect data from a certain proportion of students and use the evidence to draw conclusions about the population. Sampling is equally important where interviews, observation and other data collection methods are used. The main benefits of sampling rather than a census include lower cost, greater accuracy of results and greater speed of data collection (Blumberg et al, 2011). Saunders et al (2012) claim that the organisation of data collection is more manageable as fewer people are involved and the smaller number of cases for which the researcher needs to collect data means that more time can be spent designing and piloting the means of collecting the data. The fewer number of data to be prepared for analysis also means that the results can be available more quickly.

4.4.5.1 Probability Sampling

This is a sampling technique in which the chance of each case being selected from the population is known. The bias inherent in non – probability sampling procedures is eliminated because the selection process is random. The main types of probability sampling are:

• Simple random sampling: This is a sample selected in such a way that each unit of the population has an equal chance of being selected. It is a straightforward process that involves only one stage of sample selection. In small scale research studies, a sample can be selected by drawing names or numbers out of a fish bowl, using a spinner, rolling dice or turning a roulette wheel. In large studies, printed random numbers tables or computers can be used to draw up a list of random numbers as a basis for selecting a sample. Although the method is easy to use, the researcher requires a sampling frame to work from.

• Systematic sampling: This is a method where sample units are selected from the sampling frame at a uniform rate (for example, every twentieth item from a chosen start point in a list of names). This method relies on the population list being organised randomly, so that selecting the sample in a systematic way does not produce bias by reducing the chances of some cases being selected. It is simple to draw a sample with this method. However, increased variability may be introduced if sampling interval is related to a periodic ordering of the population.

158

• Stratified sampling: This is a method where the population is divided into mutually exclusive groups and a random sample taken from each group. The results may be weighted and combined in some situations. This method ensures that the resulting sample will be distributed in the same way as the population in terms of the stratifying criterion. However, the method is only really feasible in situations when the information required to identify the members of the population in terms of the stratifying criterion is available. The amount of work required to identify members of the population for stratification purposes makes the method uneconomical in most situations.

• Cluster sampling: This method involves making a random selection from a sampling frame that contains groups of units rather than individual units. It is based on the ability of the researcher to divide the sampling population into groups and then select elements within each cluster (Kumar, 2014). Depending on the level of the clustering, sampling may be conducted at different levels. This method is economically more efficient than simple random sampling and provides an unbiased estimate of the population parameters if properly applied. However, it often produces lower statistical efficiency due to sub – groups being homogeneous rather than heterogeneous.

To determine the sample of the African student population to be interviewed, a probability sampling method was used. The researcher used stratified sampling by dividing the African students into different nationalities and taking a random sample from each nationality. The number of respondents selected from each nationality was proportional to the size of the group of students of that nationality enrolled at the university in September 2014. A breakdown of the nationalities of the 122 African students enrolled at the university in September 2014 is shown in the table below along with the sample size (total of 20) used for the survey. The Research Randomizer at

159 Country Number of Students Proportion Sample Size Sample Nigeria 72 59% 12 4, 14, 19, 24, 25, 27, 31, 37, 45, 51, 56 & 68 Kenya 12 10% 2 3 & 5 Zimbabwe 9 7% 1 5 Libya 5 4% 1 2 Zambia 4 3% 1 2 Mauritius 4 3% 1 2 Ghana 4 3% 1 2 Egypt 3 3% 1 1 Others 9 7% Total 122 20

Table 4.5: A breakdown of nationalities, proportions and sampling.

The students that were numbered in the position selected by the Randomizer on the list of enrolled students for the intake (grouped by nationality) were selected to participate in the in-depth interviews. The stratified sampling method was used in order to ensure that the resulting sample will be distributed in the same way as the African student population in terms of nationality and to ensure that every member of each of the major nationalities had an equal chance of being selected to participate in the study. Statistically this increases the likelihood of getting responses that are representative of the African student population at the university.

4.4.5.2 Non – Probability Sampling

This is a sampling technique where the chance of each case being selected is not known. They are normally used when either the number of elements in a population is unknown or cannot be individually identified. They can never provide the researcher with the same level of confidence as probability based sampling does when drawing inferences about the population of interest from a specific sample (Easterby – Smith et al, 2012). The main types of non – probability sampling are:

• Judgemental sampling: This method involves selecting respondents who possess specific characteristics which the researcher believes are representative of the population as a whole. The researcher selects the sample to fulfil a purpose or selects those that can provide the best information needed to achieve the objectives of the study. This method can be extremely useful in constructing a historical reality, describing a phenomenon or describing something about which

160

only a little is known. However, the bias due to expert belief may make the sample unrepresentative.

• Snowball sampling: This method involves the respondent suggesting other cases for selection because they are similar to themselves. Initial respondents can be selected by probability samples and asked to identify other people that can provide the information required. It is a useful technique in situations where the researcher knows very little about the group or organisation being studied and needs to make contact with a few people, who can then direct them to other members. It is difficult to use the method when the sample becomes fairly large. • Quota sampling: This method involves giving interviewers quotas of different

groups of people to be questioned. The population is divided up into categories (like male / female, age groups, etc.) and then selection continues until a sample of a specific size is achieved within each category (Easterby – Smith et al, 2012). The objective of the method is to ensure that each of the categories is adequately represented in the sample. It is a very cost – effective way of selecting a sample. However, it introduces bias in the researcher’s classification of subjects and the resulting sample is not a probability one.

• Convenience sampling: This method involves selecting cases that are most easily available for inclusion in the sample. It involves choosing the nearest and most convenient persons to act as respondent, with the process continuing until the required sample size is reached (Robson, 2002). The sample is guided primarily by the convenience of the researcher, which might be easy accessibility, geographical proximity, known contacts or being part of a specific group. Researchers generally use this method to obtain a large number of completed questionnaires quickly and economically. Although they are not proper probability samples and it is impossible to guarantee that any sample achieved represents a specific population that may be of interest, their value depends on the purpose for collecting the data (Easterby – Smith et al, 2012). The time / cost constraints and operational difficulties in the application of most of the methods discussed above made convenience sampling the most practical option for the collection of data for the survey. Saunders et al (2012) argued that samples ostensibly chosen for convenience often meet purposive sampling criteria that are relevant to the research aim. The group of students that completed the questionnaire for this study can

161

be described as typical of international students from their respective countries and as capable as any other in terms of their ability to provide the information required to address the research questions. Bryman (1989) stated that in the field of business & management, convenience samples are very common and indeed more common than probability – based samples.

The structure of the population and the availability of the information required to identify the members of the population in terms of a stratifying criterion made stratified sampling the ideal choice for the in-depth interviews. The researcher also wanted to generate responses that were fairly representative of the African student population at the university by using a probability sampling method.