Method & Methodology
In this section the chosen method and methodology is presented. Additionally, the process of the data collection and interpretation are covered.
A research philosophy is a belief about the way data could be collected. It encompasses beliefs, assumptions, perceptions and the nature of reality and truth, and will unconsciously influence the research design (Saunders, Lewis & Thornhill, 2009). Two predominant philosophical approaches to gaining knowledge within the social sciences exist. These are known as positivism and interpretivism. Saunders, Lewis and Thornhill (2009) define the act of seeking to understand and explain a particular set of circumstances such as feelings and attitudes in unique, complex business situations as interpretivism, whereas positivism is the act of attempting to test theory in order to try to increase the predictive understanding of phenomena. In this study we mostly adopt the positivist approach since formal propositions, quantifiable measures of variables, hypothesis testing, and drawing of inferences about a phenomenon from a sample of a population is present, which according to Saunders, Lewis and Thornhill (2009) is a requirement for conducting a positivist research. However, we cannot commit entirely to the paradigm of positivism as our research philosophy, for the reason that our questionnaire is structured to find out people’s attitudes to gain a deeper understanding from our findings. Hence, a part of an interpretivist approach will be present in the discussion and recommendation part.
The two philosophical research approaches are associated with three different styles of reasoning. Saunders, Lewis and Thornhill (2009) describe these as inductive, deductive and abductive. Inductive is defined as observing a phenomenon and further developing a theory by analyzing and interpreting results, deductive is defined as the development and testing of theories and expect the outcomes of certain phenomena, which are then reworked or confirmed, and abductive involves combining developed theory with empirical findings, which allows the researcher to understand both empirical findings and theory. As this study focuses on the latter, an abductive reasoning will be applied. The positivist approach and abductive reasoning are mostly connected with quantitative research (Williamson, 2002), hence our study will adopt a quantitative research design with large enough sample size to make assumptions, however with the inability to draw generalizations.
Choosing the most suitable research design is essential in order to gather valid results that enable the researcher to transform the research question into an entire research project. Therefore, the research design creates the leading way on how the research question will be answered and how the objectives of this study will be satisfied since the answer of the research question is related to the choice of the research strategy, the choice of collection techniques and the analysis procedures, and the time horizon. It is significant for a study, that the research design emphasis validity and reliability, which in turn is based on consistent findings (Saunders, Lewis & Thornhill, 2009). The purpose of this study is mainly related to descriptive, but includes parts of the explanatory research.
Descriptive research describes population characteristics in order to portray a clear picture of a particular phenomenon on which the data will be collected. The current situation, its properties and conditions will be described by answering the “who”, “what”, “when” and “where” questions, rather than explaining the “why” and “how” (Williamson, 2002; Saunders, Lewis & Thornhill, 2009). Hence, the descriptive research is also known as the status research, since it displays the status quo of a phenomenon. Furthermore, descriptive research is relatively straightforward and easy to implement by novice researchers (Williamson, 2002). Prior research asserts that a descriptive research method emerges new questions, which require research besides describing the data. Further conclusions, evaluated data and synthesizing ideas will be encouraged due to the outcomes of the descriptive data analysis. Explanatory research emphasizes on the causal relationships between variables. It encourages explaining the relationship between variables based on a specific situation. Eventually, hypothesis will be tested and correlations between the variables can be detected and further defined. In order to implement explanatory research probability sampling is required (Williamson, 2002). The purpose of this study is to gather facts about the impact on consumer’s behavioral intention to use mobile banking. As mentioned before, this study has descriptive characteristics, revealing the answers to “what”, “when”, “where”, and “who” questions, and will encourage further explanation beyond the data description. Due to limited resources an explanatory research cannot entirely be implemented in this study. However, the collected data can be further evaluated, the relationships between the various factors and the consumer’s behavioral intention can be explained, and therefore recommendations to the most important players within the financial sector in Germany can be generated.
Since this study will mainly implement a descriptive research, numerical data is required. Thus, a quantitative research will be conducted since it allows access to a great amount of data from a sizable population (Saunders, Lewis & Thornhill, 2009). It will strengthen the ability to assume opinions and attitudes of a chosen sample to the entire German population. To carry out the quantitative research a questionnaire in form of a survey was chosen as an appropriate approach. The questionnaire supports gathering numerical and representative data within a dispersed sample (Saunders, Lewis & Thornhill, 2009). Furthermore, this approach enables conducting the data despite limited resources.
A time horizon has to be determined that is suitable to the research question, regardless the chosen research strategy and data collecting technique. The time horizon that captures a “snapshot” of a specific situation and problem is called cross-sectional. In order to create a “diary” perspective, a longitudinal time horizon needs to be implemented (Saunders, Lewis & Thornhill, 2009). This research will be cross-sectional, since only a particular phenomenon within a specific time period will be described. In addition, limitations in time and resources, only enables a cross-sectional research. Among cross-sectional studies, survey strategies are often employed, according to Robson (2002), which this study will follow as well.
According to Saunders, Lewis and Thornhill (2009) it is vital to behave in an ethical manner when conducting research. Ethics is a broad term to define, however, Collins (2011) mentions that ethics is based on two aspects; right and wrong norms for humans as well as loyalty and honesty. We attempt to work according to society’s ethics. In order to satisfy as many of the ethical aspects we ensure that information presented in the thesis will put no individual or organization to harm. The questionnaire was conducted anonymously so that no data can be traced to any individual respondents.
According to Saunders, Lewis and Thornhill (2009), in descriptive research it is vital to have a well-defined picture of the phenomena on which a researcher strives to collect data, before starting the data collection process. Therefore, we began by constructing an insightful frame of references on the subject, in order to gain deeper knowledge in the area of research. This improved the objective of the research, and gaps in existing literature were found. The secondary data was conducted in the form of academic articles, books, reports and policy papers from the university library in Jönköping, as well as online databases such as Primo and Google Scholar. Search words for the literature included mobile banking, technology acceptance model, financial technology and financial innovation. The secondary data made it possible to answer the first part of the research question. After a deep analysis of the secondary data, primary data was conducted in the form of a questionnaire.
Among explanatory research, questionnaires is a widely used research method since it provides an efficient way of numerical data collection from a large sample prior to quantitative analysis. Furthermore, it encourages enquiring the variability of opinions and attitudes of the respondents (Saunders, Lewis & Thornhill, 2009). Questionnaires are mainly executed as a survey, which can be conducted through the Internet. This enables the respondent to remain anonymous, which has a positive effect of the response rate since the respondent does not feel pressure to answer in a socially desirable way. The anonymity in turn leads to a likelihood of validity and reliability maximization (Saunders, Lewis & Thornhill, 2009). In this case, a self-administrative questionnaire was conducted, which contains a predetermined order of a set of standardized questions in order to be confident that each response can be interpreted in the same way (Saunders, Lewis & Thornhill, 2009). The entire survey is divided into nine different sections, which display the seven factors that influence the consumer’s behavioral intention of mobile banking usage, and a section that provides information about the current situation of the respondent regarding their mobile banking usage.
The questionnaire comprises a total of 20 questions, which have been proven in prior studies to obtain an essential data set that is based on the proposed factors (Appendix A, table 21). The questions were designed in various types like as ranking questions, list questions, and rating questions. Questions were adapted to a 5-point Likert scale, ranging from (1) strongly agree to strongly disagree. Since this study is based on the consumer’s opinion toward the factors that might influence the acceptance of mobile banking, most of the questions represent this type of design. The list questions secure that the respondent takes all possible answers into consideration. These were mainly used in relation to demographical aspects. Ranking type questions have been used in order to obtain greater insights towards the specific factors. The questions were asked in a specific series of statements to inhibit confusion and lead the respondent through the different categories that are significant for this study. Through this design of the questionnaire information about the respondent’s opinion and attitude toward mobile banking usage could be conducted. Initially, the questionnaire was conducted in English since this study refers to international research (Appendix A, table 21). The questionnaire was additionally translated into German since the chosen sample consists only of Germans (Appendix A, table 22). A back-translation was chosen as the translation technique, which means the source questionnaire (English) was translated to the target questionnaire (German) and back to the source questionnaire (English) (Saunders, Lewis & Thornhill, 2009). This translation technique inhibited errors of validity and reliability due to language barriers. Furthermore, it increased the response rate.
Population and Sample
In order to answer the research question and to meet the objectives inference is needed from a chosen sample about a population, which is defined as the full set of individuals from which a sample is taken from (Saunders, Lewis and Thornhill, 2009). The population of this study comprises the entire German population, regardless the current geographical location of each individual.
Regarding the purpose of this study, a probability sample would be an appropriate sampling technique since it enables inferring from the sample about a population through generalization. Especially, survey-based research is mainly conducted through a probability sample technique (Saunders, Lewis & Thornhill, 2009). However, due to restrictions of time, economic recourses and access to a suitable sample frame, the sample could not be selected statistically at random. Thus, a non-probability sample, which is based on subjective judgment, was applied. This sampling technique enables specifying the probability that the chosen sample includes any individual and enhances answering the research question. However, the possibility exists that the sample will not represent the entire population due to a lack of the suitable sampling frame, and therefore, the outcomes may not lead to an appropriate answer (Saunders, Lewis & Thornhill, 2009).
Non-probability sampling is often used among researchers when only a sufficient number of responses are required to satisfy their research objectives (Saunders, Lewis & Thornhill, 2009). The most suitable type of non-probability sampling under the giving conditions is called convenient sampling. It is based on convenient selection of samples that are easiest to obtain while the selection process continues until the required sample size has been conducted. There is a possibility that convenience sampling could lead to biases due to the little variance among the individuals and subsequently to an error regarding generalizing the sample to a population (Saunders, Lewis & Thornhill, 2009). The population is assumed to be little variant, which renders the chosen sampling technique appropriate for this study, even though the limitation of this sampling technique has to be taken into consideration.
Prior to releasing an appropriate survey, a pilot test was taken in order to ensure that the collected data will meet the objectives of the study. A pilot test helps refining the questionnaire so that possible problems emerged by the respondents while answering the questions could be detected and enhanced immediately. Furthermore, the pilot test can help to emerge problems that refer to the collection of data. Subsequently, the problems can be solved and errors within the final data collection can be inhibited (Saunders, Lewis & Thornhill, 2009). Regarding this study, the questionnaire was pilot tested among a few chosen respondents. Based on the feedback of each respondent, the questionnaire was enhanced. The length and layout of the questionnaire were adjusted, questions were rephrased due to problems of understanding, and further instructions were added. These adjustments lead to the final version of the survey, which appeared to be appropriate to meet the study’s objectives.
The survey was created through the world’s leading provider of online-survey solutions, called SurveyMonkey. SurveyMonkey is a platform that provides an easy set-up of an online-survey for free in several languages (SurveyMonkey website, 2015). The questionnaire, which was developed in English and German, was administrated via Internet in order to generate a high level of response among a dispersed sample, despite the lack of resources. According to Sue and Ritter (2007) online administration is a suitable option if the sample is fairly large and widely distributed geographically. Furthermore, the constant access to the Internet creates a great possibility of a large number of respondents (McBurney & White, 2007). Another important aspect is the anonymity of the respondent while answering the questionnaire, which let the respondent feel comfortable (Saunders, Lewis & Thornhill, 2009).
The survey was distributed via a link provided by SurveyMonkey on March 21st, 2016. The link was posted on Facebook in several groups that addressed only Germans, who are likely to participate in questionnaires. The Facebook-users were invited to participate in the survey and to share it in order to increase awareness. A cover letter at the beginning of the survey contains all important information regarding the purpose of the study, the specific target group, and legal information, which protect each respondent by using the gathered information confidentially and only for the purpose of this survey.
To analyze the data conducted from our questionnaire, a statistical data software called IBM SPSS Statistics Version 21 was employed to test the hypotheses and further drawing conclusions. In addition, SurveyMonkey provided tools for an initial analysis of the data.
Setting up data in IBM SPSS Statistics Version 21
In order to be able to analyze the data in IBM SPSS, the data set had to be re-coded, which means to transform all answers into values, which can be later used for the analyzing process. Furthermore, all questions that represent demographical aspects could be initially analyzed based on the existing data set. Subsequently, all incomplete data sets will be excluded of the entire analyzing process. Incomplete data sets are for instance responses that have not been completed by the respondents. Excluding these data sets will ensure that the outcomes will not be adulterated. Since the questionnaire contains several questions related to one factor, these questions need to be compounded into one factor. With the created factors in the IBM SPSS data set, a Cronbach’s Alpha Analysis, linear regression analyses, and a Spearmen’s rank correlation coefficients (Spearmen’s rho) were be conducted. These tests will examine, if the data set can be used, if the chosen model is useful for the data, if and to what degree the variables are related.
Reliability and Validity
Reliability refers to the degree of consistency of the results of the data, when the same test is conducted again (Saunders, Lewis & Thornhill, 2009). The most used statistical test used to ensure reliability is the Cronbach’s alpha, which was employed in this study (Appendix B, table 23-39). The Cronbach’s alpha provides the internal consistency of responses to the rating scale that is administered in the research (George & Mallery, 2003).
Validity stands for how thoroughly and correctly the research is conducted according to the phenomenon tested (Saunders, Lewis & Thornhill, 2009). To ensure the degree of which our study provides sufficient coverage of the research question, we mainly concern ourselves with content validity. To confirm content validity of the questionnaire used in this study we selected questions used in previous studies that have been shown to be significant in research regarding the model used in this study, originally proposed by Davis (1989).
Linear Regression Analyses
A multiple linear regression analysis provides detailed information about the model used, the conducted data set, and the form of the linear association between two variables. The multiple linear regression analysis is based on one dependent variable (Y), and two or more independent variable (X) (Anderson, Sweeney, Williams, Freeman & Shoesmith, 2010). This study mainly focused on three outcomes of the multiple linear regression analysis. First, the coefficients show if the two tested variables are linearly related, and if the independent variables are economical significant toward the dependent variable. Second, the ANOVA table shows if the chosen model is significant since if a model is not significant, the model is useless and the data outcome cannot be used. Third, the model summary includes the coefficient of determination, which reveals the goodness of fit of the model used (Anderson et al., 2010). To obtain detailed information about the strength of the relationship between the factors and the consumer’s intention behavioral, we conducted a Spearmen’s rho.
In addition, a simple linear regression analysis was conducted in order to analyze the relationship between PU and PEU. The outcomes, the analysis as well as the interpretation are based on the same criteria as in the multiple linear regression analysis.
A Spearman’s rank correlation coefficients, also known as the Spearman’s rho, are a nonparametric statistical test, which statistically measures the degree of the association between two variables. In order to perform a Spearman’s rho, the two variables have to be discrete variables, which include ordinal variables (Anderson et al., 2010). Since this study is based on ordinal variables, and aims to discover the relationship between each of the seven different factors that may or may not influence the decision of accepting mobile banking, the Spearman’s rho appears to be a suitable statistical test. Based on the strength of the relationship between one of the seven factors and the consumer’s behavioral intention to use mobile banking, we can reject or accept the hypotheses proposed in the frame of reference.
When performing a multiple linear regression analysis, one needs to examine if the problem of multicollinearity is detected. Multicollinearity can be explained as when two or more independent variables in a multiple regression are highly correlated, and can be linearly predicted by the other variables with a substantial degree of accuracy (Berry & Feldman,1985). This does not affect the model as a whole, but the independent variables appear as insignificant when actually being significant. Multicollinearity can be detected by the value of the Variance Inflation Factor (VIF), which measures the impact of collinearity among variables in a regression model (Anderson et al., 2010).
Table of Contents
1.2 Problem discussion
1.4 Research question
2 Frame of References
2.1 Mobile banking
2.2 Mobile banking providers
2.3 Mobile banking services
2.4 Challenges in mobile banking
2.5 Consumer’s behavioral intention and acceptance of mobile banking
2.6 Technology Acceptance Model
2.8 Conclusion of the frame of reference
3 Method & Methodology
3.2 Research method
3.3 Ethical research
3.4 Data collection
3.5 Data Analysis
4 Empirical findings
4.1 Descriptive analysis
4.3 Multiple Linear Regression Analysis
4.4 Simple Linear Regression Analysis
4.6 Spearmen’s rho
5 Hypothesis testing
6.2 Perceived Usefulness
6.3 Perceived Ease of Use
6.4 Personal Innovativeness
6.5 Relative Advantages
6.6 Perceived Risks
6.7 Perceived Costs
6.8 Social Norms
6.9 Behavioral Intention prediction
6.10 Conclusion of analysis
7.1 Contribution to literature
8 Discussion and recommendations
8.1 Managerial recommendations
8.2 Limitations and further research
GET THE COMPLETE PROJECT
Consumer acceptance of mobile banking in Germany