Chapter 3 - Methodology
The research context selected for the study is a public sector care home run by Birmingham City Council. A care home was selected due to close proximity of service users and service givers, through numerous social interactions (Hausman, 2004). A care home within Birmingham City Council was chosen due to the fact business transformation and policy change is directly focused on the adult and communities department; which includes care homes, in the attempt to ‘save £1 billion over ten years through more efficient ways of work’ yet still ‘exceed customer expectation’ to deliver satisfaction (Birmingham City Council, 2010). Therefore the study aims to help understand how brand personality affects satisfaction in order aid public sector departments to manage change better.
There are three types of research approaches possible; quantitative, qualitative, and mixed methods (Dewhurst, 2006). The quantitative approach can be defined as an approach in which the researcher ‘primarily uses post positivist claims for developing knowledge, employs strategies of inquiry such as surveys, and collects data on predetermined instruments that yield statistical data’ (Creswell, 2003, page 18, cited from Sonderguaard, 2009). Creswell (2003, pg 18 cited from Sonderguarrd, 2009) conversely defines a qualitative approach as an approach in which a scholar ‘often makes knowledge claims based primarily on constructivist perspectives in addition to the use of strategies of inquiry such as narratives, phenomenology’ based upon grounded theory. On the other hand, a mixed methods approach is based on pragmatic grounds; Patton (1990) defines mixed methods as an approach that investigates a phenomenon, using the notion of method triangulation as a means of acquiring convergence of the qualitative and quantitative methods. Therefore, triangulation in a mixed methods approach allows a researcher combine both methodologies in an attempt to offset the inherent weakness of an approach used in isolation (Blumberg et al, 2005). Consequently the author has chosen to adopt a mixed methods approach. The author has elected to utilise two form of primary research; semi-structured interviews as a means of producing rich validity and a more in-depth understanding; whilst also implementing a focused questionnaire in an attempt to establish link between stakeholder satisfaction and brand personality. The author does note Remenyi et al (1998) criticism of the triangulation of data as being time consuming, often costly and in practice not effectively utilised to actually counterbalance the weakness of an approach by the strengths of another. Therefore the research will endeavour to unite the two forms of research adopted proficiently.
However before conducting the primary research, the study analysed the existing literature. Secondary research can be defined as the analysis of data gathered by other scholar (Blumberg et al, 2005). Although secondary research is often perceived as; an examination of broad pieces of information, that generally is outdated or lacking accuracy (Malhotra, 1996). The analysis secondary data allowed the author to fulfil two of the three requirements Zikmund (2000, pg 93) stated as being necessary to create a ‘good empirical study’: the first being to provide a foundation on the breadth and depth of the research at hand, by developing a conceptual background using pertinent literature. The second being, to identify the major influences upon independent variable and dependent variable, in the case of the current research quasi-markets in the public sector. The third requirement, to prove or disprove your hypothesis will be tackled by the primary research.
Primary qualitative research is employed on an exploratory basis. Molhorta and Birks (2000, pg 75) delineate exploratory research; as a “research design that has, as it primary objective, the provision of insight into, and comprehension of the problem confronting the researcher”. Demaree and Wolf (1984) suggest that exploratory research is often used synonymously with qualitative research. Creswell (1998, pg 37) suggests ‘qualitative research is the inquiry process of understanding based on distinct methodological traditions of inquiry that explore a social or human problem. The research aims to build a complex holistic picture, analyse words’ and report detail views of informants using a small sample size. The sample size used in the qualitative research is discussed in detail in chapter (NUMBER). The type of the qualitative research the researcher chose to implement to gain a more holistic image of brand personality and satisfaction was semi-structured interviews.
Denzin and Lincon (1994), defined semi structured interviews as a techniques used to accumulate qualitative data by creating a set of circumstances that allows participants the time and scope to talk about their opinions on a particular subject. This method is particularly useful for the research, as it allows the author the ability to focus upon certain topics, yet does not confine the conversation to topics that maybe too narrow in focus to understand the complexity of the phenomenon (Huberman and Miles, 1994). This in turn, reduces the interviewer bias associated with other forms of interviews that focus upon solely on what the interviewers perceives as important topics. Consequently, the author used of open-ended questions such as “Can you give me an example of interaction with a patient?” (appendix D2); to aid in the understanding of how brand personality affect satisfaction rather than making generalised points about interrelationship of behaviours and perspectives (Patton, 1990).
The study utilised questionnaires as another method of amassing primary research for the purpose of statistically assessing correlations and establishing cause and effects (Baker, 2003). Questionnaire were chosen over other forms quantitative data, as the implementation of the Likert scale is the easiest method of enabling the research to ascertain stakeholders opinions of the brand personality (Malhotra, 1996). Secondly, questionnaires were also chosen as a research method because it allowed the author to target a large sample of specific stakeholders quickly, inexpensively and within time frame available for the research (Alverson and Deetz, 2000). Which lead the author to conclude questionnaires lends itself more to the objectives of the research, more than other forms of quantitative data.
Two separate surveys were designed; one for external stakeholders and the other for internal stakeholders (Appendix B and C). Both surveys contained the Saucier (1994) standardised and validated measure of brand personality, following a similar method used by Bourke et al (2009), Brown (2002) and Davies et al (2004) . By using the validated scale, items such as, agreeableness, conscientiousness, emotional stability, extraversion and intellect could all be assessed effectively (Aaker, 1997); the measure and the construct of the questionnaire can be found in appendix A. The construct of the satisfaction dimension was developed from previous research conduct by Naver and Slater (1990) and Davies et al (2004), to make them more specific to the research study at hand, and shorter to reduce participant fatigue (Bryn, 1998). All items were analysed on a 7 point Likert scale to provide consistency throughout the questionnaire and reduce the possibility of participants’ confusion.
The reliability of the questionnaire will be certified by using Cronbach (1951) Alpha Co-efficient. Zikmund (1997, pg 92) defines reliability as the as the ‘degree in which measure are free from random error and therefore yield consistent results’. Sekeran (1992) stated that reliability is fundamentally concerned with the uniformity and truthfulness of data. The Cronbach (1951) Alpha Co-efficient is the most extensively used, statistical measure of internal reliability. For research to be considered reliable the minimum value of the Alpha Co-effient should be 0.7 (Malhotra , 2002).
A self selecting convenience form of non-probability sampling was employed to select participants to partake in the questionnaire. Malhotra and Birks (1999, pg 353) describes self-selecting convenience sampling as a “non- probability sampling technique that attempts to obtain a sample of convenient elements”. This method was used as the author could not physically hand out the questionnaires to visitors and front-line employees, but relied upon secretary staff to encourage visitors of residents alike to participate in the study when. This method was also chosen as a probability sample, as the researcher was not able to gain a register of the employees, due to confidentiality reasons. Therefore, of 50 questionnaires available for front-line employees, the study had and effective return rate of 74 %, however due to numerous missing values on one questionnaire, the questionnaire was rejected. 25 questionnaires were also given to visitors of patients and 25 questionnaires were given to able patients; there was a slightly lower return rate of 66%, the proportion of return for customer can be seen in figure 2. In accordance to research in the same field conducted by Davies et al (2004) the return rate was judged to be adequate (above 50%).
The Sobel test, was utilised to test analyse the mediating influence of the individual dimensions of brand personality on satisfaction. The Sobel test is significantly affected by different sample size, therefore, is appropriate for this study (Baron and Kenny, 1989). One of the reasons the Sobel test was adopted over the Aroian or Goodman test was due to Baron and Kenny (1989) research which found over 11,000 studies that employed the Sobel test, therefore adding significant weighting to the assumption the test is an accepted method by academics. Pierce (2003) explains that the Sobel test, analysis the whether the independent variable notably affects the mediator: it also analyses whether the independent variable affects the dependant variable in the absence of the mediator: the third examination the Sobel test conducts is if the mediator has a unique effect on the dependant variable; the final analysis the test conducts is whether the effect of the independent variable on the dependant variable decreases in the nonexistence of the mediator as can be seen in figure (). Chisnal (2001), states for the results of mediation to be considered significant, the sobel calculator has to compute a figure greater or less than 1.96; Bourke et al (2009) and Brown (2002) both used 1.96 in their studies.
A limitation of conducting semi-structured interviews with a limited sample; is the question of whether the information gained can be generalised in traditional sense (Myers, 2000). Although the author accepts this as an issue, the purpose of qualitative research in this study is not to establish correlations or definite causal links but as way of adding concurrent validity to quantitative data (Alveerson and Deetz, 2000). To overcome the representativeness of qualitative data, the study combines the data with the finding of questionnaire, which has a significantly larger sample size (Strauss and Corbin, 1990).
Singh (2000) condemn the use self-reporting measure, due to their susceptibility to bias. Singh (2000) suggested that a participant would answer in a socially desirable way. However, Schneider and Bowen (1995) found after investigating numerous studies using meta-analysis, that self-reporting measure are likely to increase the mean values, yet are not likely bias correlations with other constructs as all means are likely to increase. Straus and Corbin (1990) suggest that if self reported measures are anonymous, social desirability bias is significantly reduced; which in turn increases the validity of the questionnaires. Therefore, the study gave all participants anonymity and as the questionnaires were returned into a relevant drop box, participants could have confidence that the confidentiality process would be adhered to.