Chapter 3: Organizational Management
4.1 Research Process
4.1.7 Samples Selection
To make a research accurate and economical, quite often representative subset of the population are selected by the researchers. This technique is called sampling and selected representatives are called sample (Showkat and Parveen, 2017). A sample has been defined by Field (2005) as “a smaller (but hopefully representative) collection of units from a population used to determine truths about that population” (Sharma, 2016, p. 15). Sample mainly represents the whole
population. Sometimes it is not possible to collect data from the entire population. Thus, sampling can be an alternative if it is impossible to collect data from the whole population and the budget and time is limited (Saunders et al., 2015). Sampling techniques are divided into two main types, they are; Probability and Non-probability sampling. In probability sampling, every single member of the population has a chance to be picked as a sample. Which implies, in this technique, samples are chosen without any bias, randomly and independently. On the other hand, non-probability sampling method allows to pick samples in a non-randomised way and based on researcher’s subjective judgement (Saunders et al., 2015; Chaudhuri and Stenger, 2005).
Methods of probability sampling: There are five main techniques of probability sampling (Saunders et al., 2015) are discussed below;
• Simple random sampling: This is a method of probability sampling where each unit of the population has known and equal chance to be selected as a sample. This method is easy to conduct and there is a high probability to achieve a representative sample. Many statistical procedures can be used in this technique. However, in simple random sampling, it is difficult to identify and contact all members of the population (Sharma, 2016).
• Systematic Sampling: This method first makes a sampling frame based on some variable and number them from 1 to N then, pick a starting point from 1 to N and finally select the sample at regular intervals such as every k units after the starting place in the sampling frame (Shama, 2015). Although, the process is very easy
114 but, there is a possibility to miss equal chances of members of the population to include (Hajimia, 2014; Chaudhuri and Stenger, 2005).
• Stratified random sampling: This method of random sampling divides the population into two or more strata-based groups according to their different characteristics, for example, age, ethnicity, income, etc. Then samples are
randomly selected from these groups in simple or systematic way (Sharma, 2016). Thus, this includes advantages and disadvantages of both simple and systematic random sampling method. Patton (2002) advocated this method is more accurate of sampling as it represents the subgroups. This also can be used for proportional as well as non-proportional samples. Only disadvantage of this method is to identify all members of the subgroups and the population which is a common problem of probability sampling (Hajimia, 2014).
• Cluster sampling: This method is similar to stratified sampling as it also divides the population into several groups before sampling where groups are called
clusters and they are based on any naturally occurring grouping such as,
geographical area (Henry 1990). This sampling technique represents the complete list of clusters not the complete list of individuals. Therefor it may be easier and useful process of sampling when the population is large and spread in a large area, but the representation of total population is less accurate in this sampling than stratified random sampling (Saunders et al., 2015).
• Multi-stage sampling: This method is also called multi-stage cluster sampling as, it expands cluster sampling and normally overcomes problems associated with the population which is geographically spread such as, face-to-face contact and expenses. A series of cluster samples involving some form of random sampling are involved in this method. Multi-stage sampling method significantly saves time and money depending on how appropriate and available all the series of different sampling frames (Saunders et al., 2015; Chaudhuri and Stenger, 2005). Methods of non-probability sampling: There are five main techniques of non-probability sampling (Saunders et al., 2015) are discussed below;
• Quota sampling: This is entirely non-random method of sampling where population is segregated into cells based on related control characteristics. A quota of sample units is established for each cell such as, 20 boys and 20 girls. Then samples are selected to fill quota and thus it is a type of stratified where units are selected within strata is completely non-random (Barnett, 1991; Sharma, 2016). Quota sampling is less costly and very quick process, moreover, it does not
115 involve a sampling frame. On the other hand, quota boundary system of this method involves more biasness. Also, samples may not be represented which are not easy to access and unwilling to participate 39.
• Purposive sampling: In this process, samples are selected based on researcher’s experience or knowledge to find the most appropriate sample for the study. This is also called “judgment” sampling. Purposive sampling is taken for working with small size but very informative samples (Neuman 2005). This method is fairly less expensive, more suitable and easily approachable, and select only those units that are more related to answer the research questions and address the research objectives. But the disadvantage is inaccuracy of the researcher’s judgement of selecting samples will not represent the population (Showkat and Parveen, 2017; Saunders et al., 2015).
• Snowball sampling: This process is also called "chain referral sampling,” where samples are collected in several steps by considering the collection of added respondents is on the basis of referrals from the primary respondents and the process stops when no more referrals are found, or the samples are too large to manage. Snowball sampling is suitable when no other methods are available or when the population is incomprehensible. Once the initial contacts are found then they identify more members, who introduce other members of the population and so on. But the problem is to make the initial contact and respondents biasness is a big issue as most likely they identify similar to themselves for further respondents which can cause a homogeneous sample.40
• Self-selection sampling: This is the method of non-probability sampling which selects samples by publicising researcher’s need for samples, then collect data from those who want to participate (Saunders et al., 2015). This is very useful technique for the researcher to collect data from appropriate respondents but there is a possibility to get not enough samples as the topic might not relevant or
appealing to them.
• Convenience (or haphazard) sampling: This method allows researchers to select respondents according to their own convenience by collecting data haphazardly from the closest available respondents who are willingly participate to the process and it continues until the required sample size has been achieved. This method is used extensively but bias is expected and there are possible impacts that which are
39 Saunders et al., 2015; Hajimia, 2014; Chaudhuri and Stenger, 2005. 40 Showkat and Parveen, 2017; Saunders et al., 2015; Lee, 1993.
116 out of control, as the respondents emerge in the sample only because they were easy to obtain (Saunders et al., 2015).
Although, particular sampling methods are used for particular research strategy, for many research projects combination of different sampling techniques can be used depending on the capability of researcher to get access to organizations (Saunders et al., 2015).
Field of this research is very specific as it is investigating the effect of culture and leadership behaviour on ethical labour management in RMG industry of UK and Bangladesh, so the choice of research strategy will be guided by the research questions and objectives which will bear out in the following part of this chapter. The extent of existing knowledge depicted in literature review chapter is also important for choosing the research strategy followed by the amount of time and other resources such as primary data from fieldworks, as well as research philosophy and research approach. The next part of this chapter below, will give a clear idea of preferring our research strategy in light of philosophical positions of this study.