CAPÍTLUO 3: ANÁLISIS Y DISEÑO DEL SISTEMA
3.6. Conclusiones del capítulo
Bloomberg and Volpe (2008, pp. 78–79) maintain that “…typical delimitations refer to selected aspects of the research problem, the time and location of the study, the sample selected…” Early in the study I asked myself the following questions: Which data sources will be regarded as rich in information? How would I obtain the relevant data? To whom should I direct my request for assistance? Which settings should I go into? Which combination of data collection techniques should I apply? Who should I approach for peer reviews? As the study proceeded, I experienced what Urquhart (2013) pointed out, namely, that the nature of questions such as these changed.
Typical questions then included: What data is the TES firm or client business concealing? Which data sources may confirm or challenge my understanding?
In search for data, qualitative researchers turn to people in real-world situations. As Yin (2011) points out, there are a variety of institutional and everyday settings in which people perform real-life roles.
First, field settings can include small groups of people who share a common bond, such as a gang or a work group. Second, they can cover residents of the same small geographical area. Third, field settings can focus on institutional scenes. Everyday life in many different kinds of institutions, such as clinical settings or schools, can be the topics of study.
Fourth, field settings may be defined as unrelated groups of people. They may share some common condition, such as a similar health problem or medical ailment, but they do not interact as a group, reside in geographically proximal areas, or serve as members of similar institutional settings. This fourth definition has been prominent in grounded theory research (Yin, 2011, pp. 111–112).
In view of the fact that I did not know any persons who were in temporary employment service, I set out to identify potential sites where potential research participants could be recruited. Guided by the concept of triangular employment relationship inherent in the TES phenomenon, I located a number of sites. Such sites included a hospitality entity in Hazyview, a public service institution, a retail business and a bakery in Nelspruit. While it was convenient to select settings in Nelspruit and Hazyview because of cost effectiveness, accessibility and time, being mindful that convenience alone has been considered the least desirable sampling strategy (Suri, 2011) and is neither purposeful nor strategic (Patton, 2002).
The retail entity comprised a large grocery chain shop while the manufacturing entity was a food manufacturing outlet. The public service institution was a provincial government department with a lengthy record of recruiting temporary employees. I also took advantage of my work commitments in Pretoria to locate other possible RPs. My colleagues referred me to one ex-TES employee who was working in a banking institution in Pretoria at the time of the study and this individual, in turn, put me in touch with a RP who was working in a state-owned communication setting in Pretoria.
Thus, I used primarily purposive sampling (Foley, 2010; Suri, 2011) to identify access points or settings where I could find people involved in temporary employment service. I was also persuaded by Patton‟s (1990, p. 169) contention that
“the logic and power of purposeful sampling lie in selecting information-rich cases for study in depth. Information-rich cases are those from which one can learn a great deal about issues of central importance to the purpose of the inquiry, thus the term purposeful sampling”. Studying information-rich cases, therefore, yields insights and in-depth understanding rather than empirical generalizations.
In fact, I combined purposeful sampling with snowball sampling (see, for example, Patton, 2002; Suri, 2011) as I requested informants, contacts, colleagues and friends to provide me with the names and contact information of people they thought could be willing to participate in the study. According it is the referral sampling technique where existing participants recruit future participants from among their acquaintances. As Suri (2011, 69) defines it, “Snowball sampling involves seeking information from key informants about the details of other information-rich cases in the field.” I noted the caution against the “expert bias” involved in snowballing but, nevertheless, I proceeded as the cases identified were selected more for their inside knowledge than their expertise.
Gaining access to potential research participates constitutes the next critical step in the process of data collection (Denzin & Lincoln, 2005. This was not easy! After the lack of success of my initial attempts to gain access via TES firms and client business organisations, I was fortunate to make contact with potential research participants after I had approached certain trade unions for information about appropriate settings. I regarded the difficulty I had encountered in persuading the TES firms and client business organisation to grant me permission or to participate in the study as a challenge. I got a strong impression that there is something I could discover that was rich enough for answering the research questions in this study.
I interviewed labour representatives, the Department of Labour administrative centre manager, and Commissioners at the Commission for Conciliation, Mediation and Arbitration (CCMA) for information about people involved in temporary employment and where I might be able to locate them. In addition, I acquired valuable information during participant observation with some TES employees. In order to obtain a
broader contextual understanding, I gathered information on the dispute resolution processes pertaining to TES employees from the regional office of the CCMA. Also, interviewing the Department of Labour administrative centre manager provided valuable insights into the monitoring and enforcement of compliance with labour-related policies and regulations.
I identified potential participants according to their experience and knowledge of the triangular nature of the employment relationship inherent in TES. In addition, in view of the complexity of the phenomenon, I deemed it appropriate to sample from multiple settings to gain a wider perspective of the impact of TES as had emerged from current debates between various stakeholders as well as the preliminary research I had undertaken. I intended testing the assertion that TES employees were motivated particularly in accepting TES employment; such as job satisfaction, life satisfaction, and turnover intention (De Cuyper & De Witte, 2008; Burgess & Connell, 2006). Accordingly, I opted to use theoretical sampling. Finally, I took heed of Krueger and Cassey (2009) who emphasise that when sampling the purpose of the study should be taken into account.
While there is no consensus among qualitative researchers on the size of the sample inclusion in a qualitative research, there does appear to be agreement that this is determined by the study‟s purpose and also that the samples in a qualitative study are usually relatively small. Ritchie and Lewis (2003) cite three reasons for this state of affairs. Firstly, if the data analysis is to be conducted properly, the sample should make it possible for the researcher to reach a point where additional cases do not provide any new insights. In other words, the principle of data saturation determines sample size. Secondly, as statistical analysis is not required in qualitative research, sample sizes do not have to be statistically valid. Finally, since qualitative data is rich in detail and facilitates description, relatively small samples provide sufficient information to justify the conclusions reached. In fact, even one case is justified when the case is critical for confirming a premise, when it is rare or extreme, or when it provides unusual access for the purposes of academic research. In these circumstances, Mason (2010) and Suri (2011) concluded that in a nutshell, sample size in qualitative studies is determined more by the researcher than data saturation per se.
I considered the fact that sample size is critically important for researchers who needed statistical power to generalise. However, for qualitative research context is more important than numbers. A small number of interviews would result in shallow and inconclusive contributions. At the same time, a large number of interviewees would result in an unworkable amount of data, which would make transcription very difficult and insightful interpretations doubtful.
Silverman (2005) observes that sample size may need to be changed as a study progresses and the researcher applies theoretical sampling. Broadly speaking, specific new cases are selected based on the insights emerging during the study.
For example, it may be necessary to increase the size of the sample in order to learn more about the emergent concept, or else the researcher may seek out “deviant cases” – cases which differ from what has been found up to that point. In addition, the researcher may include some cases at a certain point and other cases at other points during the study. For example, Johnson (1998) analysed thirteen cases in order to build his theory and then added another six “negative” cases.
Carson, Gilmore, Perry, and Gronhaug (2001) and Bowen (2008) support the view that cases should be included until the researcher reaches theoretical saturation.
The decision as to the point at which this point is reached should be left to the discretion of the researcher. They (Carson et al.) furthermore suggest that if the researcher is not able to justify the use of a single case, 2 to 4 cases should be the minimum with 10 sufficient, and 15 the absolute maximum.
When I approached leaders of organised labour to identify the abovementioned setting, i.e. where employees were deployed in a triangular relationship set-up, the first person with whom I had contact provided me with the contact details of all his fellow TES employees. This made it easy for me to approach them directly.
Sustained interaction with the participants made it possible for me to eventually develop the TES employee conceptual framework. I first used convenience sampling as I searched for the most obvious cases of TES, that is, persons whose work settings were identified by organised labour. As I developed categories from the data
obtained from these participants, I applied purposeful sampling. Finally, focusing on cases essential to the emerging theory, I utilised theoretical sampling.7
Initially I had thought that an in-depth interview of a sample of 5 participants would be sufficient for the purposes of the study. However, guided by Silverman‟s advice (2005) to “theoretically” include cases as the study progressed, I ended up with eight cases.
In setting the criteria for the research participants I made use of practitioners with expert knowledge in the field of ER and dispute resolution in South Africa. They were required to (i) engage in peer discourses on the research topic, (ii) preside over labour disputes (iii) monitor the enforcement of labour and environmental law, and (iv) be involved in the deployment to a client business organisation by a TES firm.
Particularly important was who to include in in the sample. I also noted the views of Kirkevold and Bergland (2007, pp. 70–71), namely, (a) being a member of the population or group under investigation; (b) having the ability to articulate experiences in respect of the phenomenon or situation under investigation; (c) being in a state of health that permits participation in study; and (d) being willing to share one‟s experiences.
Finally, I similarly decided that participants had to be (i) 18 years or older, (ii) able to understand and read either English or an African language, (iii) in a position to provide informed consent, (iv) currently or previously employed at a TES firm, (v) able to participate in an interview, and (vi) willing to speak about their experiences.
My sampling strategy complied with the inclusion criteria as suggested by Kirkevold and Bergland (2007).