Method & Data
In this part of the thesis the design of the research process is outlined, this describing how the research was layout, conducted and designed. Sections followed will further explain more descriptive how the process for gathering information and knowledge were conducted, with reasons for the decisions.
The term methodology can be separated from the term method (Svenning, 2003). Meth-od is more seen as the actual practical way in which data is gathered and later on ar-ranged (Svenning, 2003). Methodology is constructed from the underlying assumptions and beliefs of the researchers, this to inform the reader about the nature of the selected research approach and fundamental strategies used within the research process (Black-shaw & Crawford, 2009; Svenning, 2003). For the research philosophy, there are two commonly used dimensions that can be implemented for the guidance and nature of the research, these two are; positivism and interpretivism. The positivism philosophy is as-sociated with studying the observable ‘law-like’ reality, this by collecting data to search for regularities and casual relationships within it (Saunders et al., 2012). Interpretivism is in contrary focusing on the complexity of humans and their role as social actors with no ‘law-like’ generalizations (Saunders et al., 2012). Further, Saunders et al. (2012) in-troduces a mix of the positivism- and interpretivism philosophies, this mixed philoso-phy is known as; pragmatism. The pragmatism philosophy does recognize that there is no single philosophy that can explain the entire picture due to the complexity of the na-ture. The pragmatists often use multiple methods or methods that enable reliable and relevant data to create a credible research.
In this thesis we implement a pragmatism philosophy. The underlying assumption for this choice is that we believe the nature of the research to be complex and thus not be interpreted through only one of the methodological philosophies. Due to the fact that this research will contain both quantitative and qualitative method approaches we argue for that the pragmatist philosophy will in the best way possible support the coming parts of the thesis and provide us with as a reliable and honest result as possible. However we want to state that there is no guarantee that a definite reality will exist since the reality may vary, but the pragmatism philosophy will in the best way possible contribute to an honest result.
According to Saunders et al. (2012) it is significant for the researcher to recognize the nature of the research design, this to decide upon which one of the three research de-signs that is the most suitable for the study in mind. The three approaches that a re-searcher can go about with are; exploratory, descriptive or explanatory.
Firstly, an exploratory study is a suitable approach when the aim of the research is to clarify the understanding of a problem, especially when the researcher is not fully se-cure about the specific nature of the problem. The study is often conducted by open-ended questions for the researcher to easily gain more than one insight of the topic. Secondly, a descriptive study can be declared as an opposite to the mentioned explorato-ry study. For a descriptive study it is significant for the researcher to “have a clear pic-ture of the phenomenon on which you wish to collect data prior to the collection of the data” (Saunders et al., 2012, p.171). Thirdly, an explanatory study will define and estab-lish a relationship between variables through studies from a situation or a problem (Saunders et al., 2012). Ghauri and Gronhaug (2010) refer to this study under the term ‘casual research’ as the aim of the research is to seclude the cause/s and figure out to what extent the cause/s leads to the proven effect/s.
For the nature of this research our thesis will take on an exploratory perspective, both in the questionnaire and in the focus groups. The questionnaire is based on theories as a framework, however it is explaining the yet unexplored nature of the implications of age and gender, and how they affect celebrity endorsement. The questionnaire take on a quantitative observation at what the opinions are towards the source effects of celebrity endorsement. The focus groups further explain the reasons and factors behind the find-ings from the questionnaire. We believe that a quantitative study is not enough in-depth to solely explain the phenomenon of celebrity endorsement, thus the focus groups pro-vide the possibility for an in-depth understanding of the complex phenomenon. Imple-menting this kind of study it helps us to recognize and understand the relationship be-tween the variables of age, gender and appeals towards celebrity endorsement.
Saunders et al. (2012) state that strategy/approach is in general terms a plan of actions structured to achieve a specific goal. Further they argue that a research strategy is the plan of actions for a researcher for how to go about with the process of answering the stated research question. The key idea of a research strategy is to build the foundation for the rest of the research layout so that it aims to answer the particular research ques-tion examined. Thus doing so by being supported by the previous stated ideas of the re-search questions, objectives, approach and purpose (Saunders et al., 2012). Creswell (2003) presents three ideas through which research methods can be investigated; quanti-tative, qualitative and mixed method. Creswell (2003) defines the three strategies; by firstly arguing that a quantitative approach is based on the interests of the investigator as a post-positivist (i.e., using specific variables as measurements and observations to test theories) who claims for expanding the knowledge by using strategies of inquiries that result in statistical data, this by using experiments and questionnaires on predetermined instruments. Secondly, Creswell (2003) argues that the concept of qualitative research is based on that the investigator claims knowledge alternatively on a constructivist per-spective (i.e., multiple understanding with the idea of developing a repeated pattern) or advocacy/participatory perspective (i.e., political and change oriented) or both com-bined together. A qualitative research is performed when the researcher wants to gain open-ended answers, and this is implemented by the usage of strategies like; ethnogra-phy and action research (Saunders et al., 2012). There exist also a mix of the quantita-tive- and the qualitative approach, involving strategies like; case studies and archival re-search, this approach is called the mixed method. Hesse-Biber (2010) argues that the mixed method approach complements the quantitative and qualitative approaches and is used when the researcher wants to gain a fuller understanding of the topic in question, this since it utilizes both the aspects of quantitative and qualitative approaches.
In this thesis we use a mixed method approach. This decision is built upon the argument that we want to conduct our research in form of a quantitative questionnaire, since it fur-ther allow us to receive a large extent of samples and to get a general understanding for the opinions of the respondents. Also we will by the usage of focus groups add qualita-tive stability to support the propositions, this by getting a deeper understanding for the overall problem. Using the mixed method is the best way possible for us to receive the information needed to understand our propositions. Only using one of the approaches would not be sufficient for this research. This research will take on an approach with fo-cuses on the qualitative aspects of it, thus our quantitative research will be interpreted in more of a qualitative way since we do not include the typical statistical methods that are sufficient for conducting a research with a quantitative nature. Thus our research will take on the ideas of the figure stated below.
Saunders et al. (2012) state that there are three approaches that can influence the design of the overall research; inductive, deductive and abductive. The choice of approach is based on how the researcher is going about with literature, data collection and theory (Saunders et al., 2012). When the research starts off with theory that is based on the ex-isting literature a deductive research approach is implemented. Conversely, an inductive approach starts off with the gathering of data, which is conducted to later on generate new theory. Lastly the abductive approach can be used when exploring new as well as existing theory to identify themes and patterns between the data. Thus one can say that the abductive approach is a mix of the deductive- and inductive approach and is rather in an unstructured way elaborating between theory and data back and forth.
To best fulfill the purpose of this thesis we implement an abductive approach. This is a result of that we are using already existing theory from literature which we want to fur-ther develop by moving back and forth between theory and data. By the usage of our questionnaire, we get an overview of the scientific nature of the problem, which is asso-ciated with a deductive approach, but we also gain further knowledge by the usage of focus groups, which is seen as an inductive approach, thus resulting in a mix of the two and making our thesis taking on an abductive approach. This since we do not want to exclude any aspects of the two approaches since a mixture of the two give us a holistic overview of the phenomenon.
Method is the technique and process of how to obtain and analyze data, this by gather-ing information via for example questionnaires, observations and interviews (Saunders et al., 2012). The method describes the overall idea of “how shall I conduct my re-search” (Saunders et al., 2012, p.54). The method part of this thesis is divided into sub-sections in which each step of the procedure is described more in detail; the layout for the rest of the method part is as follows:
There are two kinds of data to be collected when researchers want to answer their re-search question, these two approaches are; primary data and secondary data (Saunders et al., 2012). Primary data is associated with the data collected by the researchers them-selves, which is new data specifically collected for the purpose of the research (Saun-ders et al., 2012). Other data that can be used by the researchers for answering their re-search question is existing data, also called secondary data, conducted by other re-searchers in previous studies. Saunders et al. (2012) argue that the overall idea of data collection is to firstly review secondary data and from that find new and additional theo-ries that can be explained by collecting primary data.
Primary data can be collected through a variety of ways, among others through inter-views, questionnaires and observations (Saunders et al., 2012). The reason for collecting primary data is to get specific and relevant data for the purpose of the research. Ques-tionnaires are according to Saunders et al. (2012) a general and common tool for re-searchers to go about with to collect primary data. Interviews are used when collecting data that answers purposeful questions about the topic in question (Saunders et al., 2012). In this thesis we collect a wide range of primary data, thus focusing on an in-depth perspective as well. Based on this argument we use two ways for collecting our primary data; a quantitative questionnaire and qualitative semi-structured focus groups. This is a decision taken due to that we want to be able to get as a large extent of data to analyze, thus trying to get as a wide picture of the generalized problem as possible.
According to Saunders et al. (2012) it is suggested that one should consider to a further extent analyze already conducted data by reviewing existing literature. In this thesis we have mainly gathered our secondary data from the Jönköping University Library and the online Library. To make our research as reliable as possible we will gather peer-reviewed literature which is highly cited and of high quality. However, we would like to reserve the fact that due to limited time and resources, there could be a chance of us, in the large jungle of literature, missing out on literature that could be of further assistance for this study.
“Sampling occurs when researchers examine a portion or sample of a larger group of potential participants and use the results to make statements that apply to this broader group or population” (Fritz & Morgan, 2010, p.3). Sanders et al. (2012) state that a sample can be as reliable as using the whole population within the research if the sam-ple is representative enough to the whole population. They further argue that using a sample can be more effective since using the whole population might be impractical due to the size, the cost and the time needed for reaching out to the whole population.
Ghauri and Gronhaug (2010) and Saunders et al. (2012) recognize that there exist two different kinds of sampling techniques; probability sampling and non-probability sam-pling.
Saunders et al. (2007) present the probability sampling as being ‘representative sam-pling’, meaning “in which the chance, or probability, of each case being selected from the population is known and is not zero” (Saunders et al., 2007, p.607). Thus, results from a probability sampling should be similar as results from the whole population. Ghauri and Gronhaug (2010) explain further the advantages with such sampling and how equal probability as sampling can allow the researcher to estimate unknown pa-rameters or make valid statistical inferences about the population (Bryman & Bell, 2010; Ghauri & Gronhaug, 2010; Saunders et al., 2007). Saunders et al. (2007) state that a non-probability sampling is one “in which the chance or probability of each case being selected is not known” (Saunders et al., 2007, p.604). Based on those words the charac-teristics of a sampling can only be generalized without a statistical background (Saun-ders et al., 2007). Ghauri and Gronhaug (2010) suggest that all researchers should prefer a probability sample since a non-probability sample may present the researcher with misleading results, as it does not fully represent the population.
In this thesis we adopt a non-probability sampling approach as the limitations of time and possibility to connect with all respondents within Bryngfjorden and the students of JIBS are restricting us on various levels. Both the questionnaire and the focus groups will be selected by using this approach, this so that we get as a representable sample as possible considering age and gender among the respondents in the questionnaire as well as representable in-depth opinions among the participants of the focus groups.
The actual sample for our questionnaire consists of two different groups of sample; the members of the Golf Club Bryngfjorden and the students of Jönköping International Business School. This to get as a large variety of ages and genders among the respond-ents as possible to make the sample representable for the whole population and thus re-liable.
Bryngfjorden Golf Club
The sample from Bryngfjorden Golf Club is as of March 2014 consisting of 854 mem-bers, 639 of these members are men and 215 are women. The members have an age range from 2 years to 82 years old.
After thoroughly analyzing the data and statistics of the members of Bryngfjorden Golf Club, which was originally supposed to be the overall sample for the questionnaire in this research, we recognize that the largest proportion of the members were older thus between the ages of 40 and 65 years old. Due to this issue we decide to send out the questionnaire to younger people to complement the lack of younger participants. In hope of expanding the proportion of younger respondents we decide to target the stu-dents of Jönköping International Business School. Thus the sample that we have been using from Jönköping International Business School is consisting of 82 International Management students. This sample is consisting of the members of a Facebook group, out of the 82 members; 41 are female and 41 are male students. Thus presenting us with a 50/50 situation among the gender, and the majority of the students are taking on an age in the range of 20 to 30 years old.
Focus Group Sampling
For this research four focus groups are constructed in the following way; two heteroge-neous focus groups and two homogenous focus groups. The heterogeneous focus groups will implicate diversity among the participants in aspect of gender but not in the aspect of age. For the homogeneous focus groups the aspects of gender and age provide us with participants of both the same gender as well as the same generation.
The participants of the groups consists of participants as shown in the table below:
Design in context of a questionnaire/survey is described as “a way of arranging the en-vironment in which a survey takes place” (Fink, 2003, p.31). Fink (2003) proceeds by explaining the term environment as the objects of the questionnaire, which often con-sists of individuals or groups of people, activities, places etc.
The design of our questionnaire is constructed as a structures standardized interview with closed questions. This approach gives the respondents the same context of ques-tioning and a limited choice of possible answers, thus provide us with a result with less chance of misinterpretation and ease the process with measuring and decoding the data retrieved (Saunders et al., 2012; Bryman & Bell, 2010). The questionnaire is construct-ed as a self-completed questionnaire via the Internet by using Qualtrics, this since it is a suitable approach for closed questions and it will help us to further enlarge the size of the sample (Saunders et al., 2012).
The questionnaire consists of a brief introduction and explanation of the purpose of the questionnaire and the reason to why we want the respondent to answer it, this to in-crease the validity of the response. Due to the stated argument by Langmeyer and Walker (1991) we decide to put all of the questions concerning celebrities and products into a visual context in the questionnaire, this to further extent eliminate the risk of that the respondent is not familiar with the product or the celebrity. Also this is due to the fact that the respondent will have a possibility to associate the context as if it was a real life example and answer the question as reliable and valid as possible.
The first part of the questionnaire is based on the gender of the respondent. The first questions are categorical questions covering the ground of the independent variables (i.e. age and gender). Following this there are two questions (question 3 and 4, i.e. edu-cation and income level of the respondent) that work as a complement to question 1 and 2, regarding control variables for the generalizability of the sample. Based on what the respondent answer in the question concerning gender, the male respondents will have questions regarding male stereotypical products (question 5-9) and female respondents will have female stereotypical products (question 10-14). Questions 15-19 are about gender neutral products and are answered by both female and male respondents.
The second part of the questionnaire (Question 19-24) includes questions with regard of the Source effects model (see figure 2.2 by Ohanian, 1990). The questions are based on the source effects in combination with a non-gender stereotypical and low involvement product. From the source models only the main words/characteristic from each of the three effects are applied, thus resulting in three different possible response alternatives for the respondent. We believe that more words/characteristics would only confuse the respondent and decrease the interest of them completing and submitting the question-naire. The questions are clearly visible for the decoder on what source effect they are regarding, however the encoder is not be able to distinguish a resemblance between them, this in order for us to keep the questionnaire unbiased from pre-assumed thoughts from the respondents. Thus keeping a high reliability and validity of the measurements of the concepts.
Celebrity Endorser Selection
When creating the questionnaire to suit our propositions we decide to use hypothetical cases of celebrities endorsing products, which we ourselves constructed and created. The celebrities were chosen due to their recognition, firstly as celebrities but also due to their equal knowledge within the product area in which the hypothetical product that they are endorsing belongs to. We decide to only implement Swedish celebrities in all of the questions within the questionnaire. This is a result of that we want to increase the chance of that all of the people within the sample will have the same chance for recog-nizing the celebrities, thus creating as an honest result as possible. We have also used celebrities with different ages by creating a range of ages it increases the likeliness and familiarity of that the respondents recognizing the celebrity.
The choices of the female and male celebrity endorsers are supposed to be within the same or similar field for the respondents to achieve a similar perception of the celebrity endorsers as high as possible. With regards of respondents’ own connections and attrac-tion to each celebrity endorser, there is a possibility that a respondent can choose the ce-lebrity endorser based on his/her persona instead of the combination of celebrity en-dorser and product. We do not want to have an active part in the choosing of what we consider to be a preferable among the respondents of the female and male endorsers. Thus we do not want our own preferences to be a part of the process and consequently create a biased result. For the answers in our questionnaire to become as subjective as possible we therefore only apply each celebrity endorser once into the questionnaire.
For selecting the products in our questionnaire we have summarized the information and different aspects regarding the level of involvement of the products. Petty et al. (1983) argue that source effects in celebrity endorsement are differently significant due to the level of the product; low- or high involvement. Thus meaning that people will re-spond to a celebrity as an endorser differently depending on the degree of product in-volvement demanded, meaning that the effectiveness of a celebrity as an endorser is de-pendent on the product involvement. To state the definition of low- and high involve-ment clearer than what is already done, Holmes and Crocker (1987) states that high in-volvement products are purchased infrequently and low involvement products are pur-chased more frequently. Petty et al. (1983) argue that a celebrity as an endorser plays a more vital role when it comes to low involvement products compared to high involve-ment products.
Due to the above statements we have come to the conclusion to only use the concept of low involvement products as examples within our research, thus meaning that we only use examples with low involvement products in our questionnaire since celebrities play a more significant role (Petty et al., 1983). The choice for using low involvement prod-ucts is to maximize the reliability of the results (Saunders et al., 2012) meaning that the aspect of age and gender of the respondents are what influences the answers and not the level of product involvement, this to make the influence of age and gender as accurate as possible. We think that the usage of both low- and high involvement products would rather eliminate what our original idea is about by rather putting the level of product in-to focus.
An additional reason for us to use low involvement products is due to the price level, this as a high involvement product are purchased infrequently (Holmes & Crocker, 1987) it hypothetically implies that the cost for purchasing it is a bit more expensive. Thus imaginatively eliminating people with lower incomes, here meaning that younger people would most likely be out of the picture for purchasing such a product. The re-spondents who cannot afford buying a high involvement product, such as a car or house, would leave the response unreliable. Thus since there is a high probability that those re-spondents do not know which aspects to consider when purchasing such a products.
Interviews may be highly formalized and structured, this by using the same questions for every respondent, or they might be very informal with unstructured questions vary-ing from interview to interview (Saunders et al., 2012). Thus an interview can have both parts, this depending on the purpose of the overall research. Between the two extremes of a formalized and structured interview and an informal and unstructured interview there are a number of different interview techniques to choose among. A structured in-terview uses highly structured questionnaires with predetermined questions, resulting in a quantitative research interview. In contrary an unstructured interview is informal with no specific guidelines, rather an interview that is shaped along with the interview. As a mix of a structured and an unstructured interview one can do a semi-structured inter-view, which implicates that the researcher has a list of themes or key questions to start off with but then lets each interview take its own overall form.
In this study we implement a semi-structured interview approach, in the form of a focus group, this due to the fact that we want to gain in-depth data and opinions of the inter-viewees. An unstructured interview is not preferred since we have built up key themes and guidelines for our thesis to collect the most interesting and appropriate data but still keeping it on a level so that we gain valuable information from the interviewees. We re-ject the idea of a structured interview based on the reasons that we have a questionnaire constructed in such way and that we need opinions from the interviewees that will not be received through closed questions. The aim is to gain knowledge of the behaviors of the hypothetical consumers here being the interviewees. An additional reason for us choosing a semi-structured interview is that it is preferred when there are a large num-ber of questions to be answered and when these questions are either complex or open-ended.
Table of Contents
2 Frame of Reference
2.1 Celebrity Endorsement
2.2 Endorsement Process
2.3 The Source Models
2.4 Elaboration Likelihood Model
2.5 Match-up hypothesis
2.6 Age of the consumer
2.7 Gender congruity within celebrity endorsement
2.8 Relevance of the theories
2.9 Proposition Development
3 Method & Data
3.4 Quality of the study
3.5 Summary of the Method
4 Empirical Findings
4.1 Findings from the Questionnaire
4.2 Findings from the Focus Groups
5.1 The Endorsement Process and Consumer Preference
5.2 The Source Effects
5.4 Match-Up & Self-image of gender perspective
6.3 Suggestions for future research
GET THE COMPLETE PROJECT
Perception of the Celebrity Endorser A study of how age and gender influences the consumer perception