Microsoft Word Bachelor-Thesis-bl-pi-klar docx

4.6.1 Sampling population and sampling procedure
 
The first step is to determine the relevant target group with following time- and boundaries 
constraints for the sampling procedure. This in order to get as correct reliable data gathered as 
possible. The sample population should be as similar in characteristics to the actual 
population as possible as deficiencies otherwise will lead to non-representative results. It is 
therefore important to be as comprehensive as possible when setting up the frame for which 
the people are going to be chosen by. This is especially important in research were the goal is 
to examine something where there is known to be differences between the ones involved in 

---

25
the research area and those not. Researchers must be precise and make sure that the 
interviewees know the area that they are talking about. This kind of research is considered to 
be more purposive than random. Though this is considered a more strategic way of sampling 
data gathering, it is not to be considered completely non-random (Bryman & Bell, 2011). 
The researchers conducting the study applied the concept of snowball sampling when the 
interviews were set-up. The authors established contact with a group of people whom were 
relevant for the topic, and then used these people to establish contact with others that as well 
were relevant for the topic. Snowball sampling is not considered to be random since it is 
difficult to know the exact extent of the population. By large, snowball sampling is not used 
within quantitative studies, but within qualitative. The choice of sampling application was 
selected due to the fact that there is no available framework for sampling the population from 
which it should be taken, and the trouble of establishing such a framework means that an 
approach as snowball sampling is the only feasible one (Bryman & Bell, 2011). 
Since the purpose of thesis was to explore the relationship between brand loyalty and low 
product involvement, the researchers needed products established for being regarded as low 
involvement products. Kapferer and Laurent’s (1985) research mentioned such products, 
many of them being household products. Thereby, the researchers used three examples of low 
involvement products when conducting the semi-structured interviews for the sampling 
process. This in order to assure that the interviewees had a product to connect and consider to 
when answering the questions, which would allow for a better flow in the interview. Hence, 
all the interviewees had to be at least partly responsible for making purchases of low 
involvement products for their household. This was done in order to assure that those 
interviewed would represent the whole population of those purchasing such household 
products (Bryman & Bell, 2011). Though the choice in sample population had to be partly 
responsible for household purchases, the choice of respondents depended on the snowball 
sampling. The researchers made sure that there were no attachments towards the interviewed 
either, as this has the possibility to affect the gathered data. This will in greater extent lead to 
more adequate gathered data, freer from deficiencies (Bryman & Bell, 2011). Once ten 
interviews were conducted (list of respondents can be seen in appendix three), the researchers 
determined that theoretical saturation was reached as the answers were considered similar and 
repetitive with deep enough data to conduct a reliable analysis (Bryman & Bell, 2011). 

---

26
The interviews took place face-to-face rather than over telephone as it is considered more 
personal and enabled the interviewees to talk more freely. The interviewers also made sure 
that it was the interviewees that decided where they wanted to be interviewed as this meant 
they would, in greater extent, be interviewed in their more natural environment. The 
interviewers also made sure that there were only two interviewers at each interview as it was 
considered that more could perhaps create a more tense setting as the interviewee might feel 
like there were too many people watching and observing. It was neither practicable to conduct 
an interview by oneself, this since the answers from the respondents had to be recorded, both 
by notes and audio. This at the same time as the actual interview was conducted. Therefore, 
the interviews were conducted with two interviewers in order to ensure that nothing of the 
discussed content would be missed or misinterpreted (Jacobsen, 2002).