This chapter presents the methods used in this research. The respondents of the research and sampling techniques, instruments and methods adopted, the data gathering procedures, and statistical methods implemented on data.
There are several methods which could be used to carry out the research ant it is dependent on research problem area. When dealing with research problem, one can use any of the three categorization of research (Yin, 1994).
Exploratory research is usually carried out when problem is not well identified or it has not been visibly defined as yet, or its real scale is as yet unclear. It allows the research person to collect the information as much as possible relating to a specific problem. Exploratory research helps conclude the best research design, data collection technique and choice of subjects, and sometimes it even results that the problem does not be present. Exploratory research is quite casual, when it relying on secondary researches such as reviewing available literature, data, or qualitative approaches such as informal discussions with customers, employees, management or opponents, and more formal approaches through in depth interviews, focus groups, projective methods, case studies or pilot studies (Yin, 1994).
This is a research type in which the primary objective is to understand the temperament or mechanisms of the relationship between the independent and dependent variable. This approach is used when it is necessary to show that one variable causes or concludes the value of other variable. This research is good to use when there is no unambiguous apprehension about what model that should be used and what qualities and relations that is significant (Zikmund, 1994).
Descriptive research is used to get information concerning the present status of the phenomena to explain “what exists” with respect to variables or conditions in a state.
Descriptive research is used when the goal is to provide a systematic defination that is as factual and exact as possible or when the problem is well ordered and there is no objective to study cause/effect relation. It provides you the number of occurrences something happens, or frequency, leads itself to statistical computation such as calculating the average number of occurrences or central tendencies (Yin, 1994). One of its major restrictions is that it can not help conclude what causes a specific occurrence, behavior or motivation. We can say that, it cannot create a causal research relationship among variables.
My research purpose and research question reveal that this study is for the most part exploratory. It is exploratory because the data has been collected through unstructured interviews and questionnaires to explore the issues that influence Pakistani community intentions to adopt Internet banking services.
3.2 Research Approach
There are two basic types of research approaches, qualitative and quantitative. In the quantitative approach, outcomes are based on statistics and numbers that are presented in figures, whereas in the qualitative approach focus lies on describing an occurrence with the use of words.
Although this research on adoption of Internet banking services in Pakistan is not very extensive as compared to debate of the benefits, most of the concepts in this study have been rarely examined before, but generally in the western context. Only a slight research covers usually Singapore, Hong Kong or China, which are very developed economies and not envoys of all Asian countries. so to gain deeper understanding of the issues in the Pakistani context, this research is conducted as a qualitative study to explore the perception of internet banking in Pakistani community.
The Use of this approach provides richer and forensic particulars for exploring viewpoints in the early stage of research. Hence the intend is not to make any simplification, but instead set up a closer contact with the objectives of preceding research, which have it in mind to provide us a deeper understanding of the participants’ attitudes and perceptions. Finally my objective with this research is to describe, explore and find complete detailed information about the issues of Internet banking adoption in Pakistan, so quantitative approach is the most suitable method for my research.
3.3 Research Strategy
Research strategy is a broad plan which shows that in which way this research will go on, and how research person will answer the questions that has been set by the person conducting the research. It will have clear objectives, derived from research question that specify the source from which research person be going to to collect data and consider the limitations that research people will inevitably have such as access to data like time, location and money, ethical issues (saunders, 2000).
Qualitative research can be carried out using quite a few strategies including: case study, experiments, surveys, histories, and analysis of archival information (Yin, 1994). Following are the short explanation of above five research strategies:
3.3.1 Case Study
Case study refers to the collection and presentation of thorough information about a specific participant or small group of participants. A case study is a written explanation of a problem or situation and normally examines the interplay of all variables for providing as complete understanding of an event or situation as possible. Case study is ideal when the researcher has slight control over the events, and when there is a up to date focus within a real life context. The rationale of a case study is to place participants in the role of decision makers, asking them to differentiate relevant from unimportant facts, to identify central alternatives among numerous issues competing for attention, and to prepare strategies and policy recommendations (Yin, 1994).
The experimental method involves manipulating one variable to decide if changes in one variable ground changes in another variable. This method depends on controlled methods, random assignment, and the manipulation of variables to test a hypothesis. This strategy is used when the research person need to contrast two variables and examine their cause and effect relationships (Malhorta, 1996).
It’s a research technique in which information is collected by interviews with a huge number of respondents using a predesigned questionnaire (Zikmund, 1994). This research technique has three important characteristics:
I. Purpose: The purpose of survey research is to generate quantitative descriptions of some characteristics of the population in study. Survey analysis may be mainly related either with associations between variables or with projecting results descriptively to a pre-defined population (Yin, 1994). Basically Survey research is a quantitative approach, calling for standardized information about and/or from the subjects being studied. The subjects under study might be individuals, groups, organizations or communities; they also might be projects, applications, or systems.
II. Procedure: The most important way of collecting information is by raising people structured and predefined questions. Answers of questions given by people, which might refer to themselves or some other unit of analysis, comprises the data to be analyzed (Yin, 1994).
III. Analyses: Information is usually collected about only a portion of the population under study, but information is collected in such a way as to be able to take a broad view the whole population. Usually, the sample is huge enough to allow broad statistical analyses.
This method deals with the past, and is in use when none of the concerning person are alive to interview or report (Yin, 1994). This method is specially used to describe the content, structure and function of the data which collected for the research.
3.3.5 Analysis of Archival Information
The purpose of this method is to explain the incidence or pervasiveness of a phenomenon (Zikmund, 1994). The use of the archival information is difficult when this topic is coming research area.
The following table demonstrates the conditions that need to be addressed when shaping on a strategy.
Most considerable condition for selecting research strategy is to categorize the type of research question being asked. Based on the research question “what” that I set for this research, I have chosen to follow case study and survey research strategy, because this research is not reliant on a single critical, extreme, exclusive or revelatory case.
3.4 Sample Selection
Sampling is a survey based research where research persons needs to analyze the sample about a population in order to answer the research questions or fulfill the research objectives (Saunders, 2000). Once the problem has been vigilantly defined, the research person needs to set up the sample that will sketch out the investigation to be carried out. It is necessary for research person to clearly define the target population from whom the specific sample will be taken. Sampling is significant if cost and time constraints stops research from surveying the entire population. Sample gives higher level of accuracy and fast accurate result.
Occasionally, the whole population will be sufficiently small, and the research person can take account of the entire population in the study. This kind of research is named a census study since data is collected on each member of the population.
Generally, the population is quite large for the research person to attempt to survey the entire of its members. A small, but vigilantly chosen sample can be used to symbolize the population. The sample represents all of the characteristics of the population from which it is taken out.
Sampling technique can be classified into two types (Saunders, 2000):
3.4.1 Probability sampling
While using probability sampling, sample selection is done in such a way that every unit of the population has a known probability of getting selected. It is this perception of “known chance” that permits for the statistical projection of features based on the sample of the population (Saunders, 2000). The benefit of probability sampling is that sampling error can be intended. Sampling error is the amount to which a sample might be different from the population Probability method includes.
3.4.2 Non-Probability Sampling
While using non probability sampling, the selection of the sample is done in such a way that the possibility of being selected for each unit in the population is unknown. In fact, the process of choosing the subjects is random or subjective, since the research person relies on his/her experience, gut feeling and judgment. Consequently, there are no statistical procedures that permit for the measurement of sampling error, and the amount to which the sample differs from the population remains unknown and therefore it is not suitable to project the sample distinctiveness to the population (Saunders, 2000). Non-probability includes:
188.8.131.52 Convenience sampling
Convenience sampling is used in investigative research where the research person is concerned in getting a low-priced approximation of the reality. As the name shows, the sample is chosen because they are convenient. This non-probability technique is time and again used during preliminary research times to get a gross approximation of the results, without increasing the cost or time required to choose a random sample (Saunders, 2000).
184.108.40.206 Judgment sampling
Judgment sampling is a common non-probability technique. The research people choose the sample depending on judgmental approach. This is generally an addition of convenience sampling. For example, a research person may make a decision to draw the complete sample from one “representative” city, albeit the population comprises all cities. When using this method, the research person must be in no doubt that the selected sample is accurately representative of the whole population (Saunders, 2000).
220.127.116.11 Quota sampling
Quota sampling is the non-probability the same of stratified sampling. Like stratified sampling, the research people first recognizes the stratums and their magnitude as they are symbolized in the population. Then convenience or judgment sampling is used to choose the necessary quantity of subjects from every stratum. This diverges from stratified sampling, where all of the stratums are filled by random sampling (Saunders, 2000).
18.104.22.168 Snowball sampling
Snowball sampling is a special non-probability technique used when the preferred sample attribute is exceptional. It may be extremely difficult or unaffordable to find respondents in these situations. Snowball sampling depends on referrals from starting subjects to produce additional subjects. Whereas this technique can noticeably reduce the research costs, it comes at the cost of bringing in bias because this method by itself decreases the likelihood that the sample will symbolize a good cross-section from the population (Saunders, 2000).
Sampling in qualitative research involves two actions; (Miles and Huberman 1994):
Setting of boundaries: “To define aspects of cases that we can study and connecting it directly to the research question”.
Creation of frame: “to help us uncover, confirm, or qualify the basic process or constructs that strengthen our study”
Non-probability (convenience) sampling has been chosen for this research because we have targeted the Pakistani community which is dealing with the banks.
22.214.171.124 Sampling procedure
The process of sampling involves using large number of items or parts of subsets of population to make conclusion regarding the whole population. The purpose of sampling is to estimate some unknown characteristics of population.
Population is any complete set of group of object. Like people, stores, students, industries etc.
A sampling frame is the listing of the elements from which the actual sample will draw. Keeping the research in view we will draw the sampling frame as under.
Population is the people of Pakistan having bank accounts, we narrow down our study only to Punjab province
Sampling frame will be the major cities of Punjab like Lahore, Multan, Faisalabad, and Sahiwal.
Keeping in view the time and cost available for the research we have narrowed down the research only to four major cities of Punjab.
The sampling technique proposed is random sampling techniques. The calculation for the sample size is given below.
Population size = 100,000
Sample %age = 0.3%
n = 100,000?0.30/100
So our sample size will be 300
3.5 Data Collection Methods
As data collection method is highly influenced by the methodology, which is chosen (Saunders and Thornhill, 2000), questionnaire are used to collect the empirical data for this research in order to identify the issues that affect the adoption of Internet banking services in Pakistan.
As this research’s main concern is examining the issues that have influence on the adoption process of Internet banking in Pakistan, the questionnaire are designed based on the requirements for adopting such a service
The questionnaire consists of three pages and twenty questions (Appendix A). It has different type of questions including open end question, close ended questions and multiple questions.
It was designed to capture all the segments of community which uses the banking services.
3.7 Validity and Reliability
In order to reduce the possibility of getting incorrect answers, attention needs to be paid to validity and reliability (Saunders et al., 2003).
Validity is concerned with whether the findings are really about what they appear to be about (Saunders et al., 2003). Validity defined as the degree to which data compilation method or methods correctly measure what they were anticipated to measure (Saunders et al., 2003). Yin (1994) says in these words, no solo source has a absolute advantage over all of the others (P.85). The sources of different types are highly complementary and as many sources as possible should be used for a good case study. The usage of various sources of evidence can increases the validity of scientific study (Yin, 1994).
The following steps were taken to ensure the validity of this research:
The needed data was collected in the format of a structured questionnaire that had been designed based on the literature related to adoption of innovation.
The questionnaires were pre-tested. A pilot test was conducted with the questionnaire.
According to Saunders et al. (2003), reliability refers to the extent to which data collection method or methods will produce consistent results, analogous observations would be made or results reached by other research persons or there is clearness in how sense was made from the unprocessed data. Reliability can be assed by the following three questions:
Will the measure yield the same results on other occasions?
Will other observers reach similar observation?
Is there precision in the method how sense was made from the unprocessed data?
The role of reliability is to minimize the errors and biases in a study (Yin (1994). This means that reliability is to demonstrate that the operations of the study, such as the data collection procedures, can be repeated with the same result. Saunders et al. (2003) asserts that there may be four threats to reliability. The first of these is subject of participant error, which means that a questionnaire may generate a different result at different times of the week. The second threat to reliability is subject or participant bias, which is when interviewees may have been saying what they thought their bosses, wanted them to say.
Third, there may have been observer error that different interviewer may approach the questions in different ways. Finally, there may have been observer bias, which means that there may have been different approaches to interpreting the replies.
The work with this thesis started with a considerable literature study. The literature I came across (mainly articles) was from several authors and often had Internet banking and adoption of e-banking topics, which meant that I covered the area of Internet banking surroundings. This would suggest that bias, form reading only one author and reading only about one topic, be held at a minimum level. Widersheim-Paul and Eriksson (1997) describe some other erroneous belief that is to be avoided in order to attain high reliability. One of these is measuring error, which in turn consists of respondent errors, gauging errors and errors that are effect of interplay between the interviewer and the respondent. As I used a questionnaire, this latter error was avoided in advance.
The respondent errors are such errors that are due to the fact that respondents sometimes are unable or unwilling to provide truthful answers. In order to minimize effects of this kind of errors, I found it necessary to be careful about the language and the wording.
Furthermore, the use of wording in the questionnaire was of major concern to avoid ambiguous or emotional charged formulations. The chosen wording and language was simple, direct and as far as possible without technical terms.
The gauging errors arise when a questionnaire entails erroneously formulated question, wrong order of question etceteras (Widersheim-Paul and Eriksson, 1997). The order of the questions was also subject to analysis and it was found to be suitable to have a disposition where the initial questioning concerned facts that the respondents easily could give an answer to.
Numbers of different steps were taken to ensure the reliability of the study:
Case studies were used during the data collection.
The same type of questions were asked from company’s respondent in order to increase the reliability
Since the generalization is not the purpose of the study, multiple cases have been used to increase the degree to which the findings can be the same. It might be possible to get the same result on the findings to a larger number of similar cases.
The theories that have been selected for the study were clearly described and research questions have been formulated based on the previous theory. Data has been collected based on the research model that was drawn from the discussed theories. The objective is to make sure that if another investigator will follow the same procedures and used the same case study objects, the same conclusions would be made.
A pilot test of the questionnaire was carried out. All the test respondents filled in the questionnaire and their opinions how they felt about filling in the questionnaire. The test was followed by many revisions, before it was sent to respondents. After refining some questions and items within the questions, the second pilot study was run and asked respondents to check for the wording, coverage, relevancy of the items listed within the questions. Finally, at this stage little modifications were needed and finally, the well-improved questionnaire was developed.
By using these tools (reliability, validity and pilot study) I can further analyze the data that the respondents provided me in a more accurate way.
3.8 Data Analysis
Data analysis can be defined “as consistent of three concurrent streams of activity: data reduction, data display and conclusion drawing/verification” (p.10) by Miles & Huberman (1994). Data reduction should not be thought to be separate from analysis, but a portion of it. Data reduction stage of the analysis helps the researcher to make the data sorted, sharpen, focused, organized and discarded in order to be able to draw and verify conclusion (ibid.).
The data reduction stage of the analysis helps the researcher to make the data sharp, sorted, focused, discarded, and organized in order to be able to draw and verify conclusions. The data display is a way to organize and compress the reduced data so that it will make it easier to draw conclusions. This phase is useful when the researcher studies more than one case, a so-called multiple case. In the conclusion drawing and verification the researcher notes regularities, patterns, explanations, possible configurations, casual flows and propositions.
Data analysis involves examining, categorizing, tabulating or otherwise recombining the collected data (Yin, 1994). Every investigation should have a general analytical strategy in order to determine what to analyze and why. Two general strategies are suggested. The researcher can either follow the theoretical propositions that led to the case study or develop a descriptive framework to organize the case study. Within these strategies, there are four different techniques for analyzing the collected data. The first is pattern matching, which means to compare an empirical based pattern with a predictable one.
The second technique is explanation building, which refers to a kind of pattern matching where the goal is to analyze the case study data by building an explanation about the case. The third is time-series analysis that refers to repeated measures of the dependent variable/variables in order to look at changes over time. The last technique is to use program logic models, which is a combination of pattern-matching and time-series analysis where the analysis specifies a complex chain of patterns over time.
Data analysis of this thesis is based on the three steps defined by Miles & Huberman (1994) i.e., data reduction, data display and conclusion. After completing the data collection I have organized the data for every case study based on the issues that has been selected from research model according to the research question and literature review.
Within-case analysis I compared the findings of each case based on my research question and issues that selected from research model. Furthermore, I conducted a cross-case analysis to compare the different case study in order to find the resemblance and variation between the cases.