Qualitative researchers favour non-probability sampling techniques (Berg 2007: p.43; Neuman 2007: p.141). These techniques are utilised when a sampling frame for a population is unknown or when a probabilistic approach is not required (Blaxter, Hughes and Tight 2004: p.163) and are beneficial for accessing difficult to reach study populations (Berg 2007: p.43). According to Kumar (2011: p.206), there are five types of non-probability designs: quota, accidental, expert, purposive and snowball. Each is outlined, in turn.
'Quota' involves the selection of a sample that contains pre-specified characteristics e.g. race, gender and deciding how many people will be placed into each category (Babbie 2007: p.185; Neuman 2007: p.142). 'Accidental' sampling is simply based upon availability of research subjects e.g. those who are close at hand or are easily accessible (such as a class of students) - there are no other defining characteristics of the sample population (Kumar 2011: p.207; Berg 2007: p.43). 'Expert' involves the researcher using his / her own ability to judge who constitutes being an 'expert' in a field of study and thus selects them as part of a sample on this basis (Kumar 2011: p.206).
The sampling techniques selected for this research are 'purposive' and 'snowballing'. The aim of purposive sampling is to obtain specialised information from participants - which is similar to expert - but where the desired study population is difficult to reach (Kumar 2011: p.206; Neuman 2007: p.143). There are two main reasons for choosing purposive sampling. The first is that the approach is consistent with phenomenological research in that interviewees will provide reliable information about the phenomena being studied (Sanders 1982), and secondly, the approach enables the researcher to source participants from the literature and case study references as well as from personal, professional networks. 'Snowball' entails the use of networks - this is where a number of people are selected by the researcher that share characteristics of interest, data is collected from them, and then they are asked to identify other people for interview that have these same characteristics (Babbie 2007: p.184; Berg 2007: p.44). Snowball will be used to some extent since interviewees will be asked to suggest other suitable contacts.
The sample size will be determined retrospectively, when 'saturation' is reached in terms of information concerning the issues explored, i.e. when no further significant issues are revealed. The point of absorption will therefore depend on the diversity of the answers provided (Richards 2005: p.135-136; O'Leary 2005: p.114).