This paper commences by looking at data analysis strategies and techniques of analyzing qualitative and quantitative data. Qualitative and qualitative data must be clearly linked in order to allow for easier interpretation. This article contains a great deal of information presented in a coherent and logical manner with regard to data analysis and interpretation in business research. This is done by presenting statistical and non-statistical methods that pervade modern business research in analyzing and interpretation of data. Business research is essential for both managers and decision makers. Thus, business research should be done in the most professional way. As a fore mentioned, a myriad of tools are used for analysis and interpretation of business data-the focus of this paper is on the modern methods and analytical techniques available to decision makers operating in a supranational business environment. The main objective is to highlight data analysis and interpretation in a more comprehensive, with an emphasis on collaborative problem-solving through exploration of actual business problems and data (Schoenbach, 2004).
Thesis statement
This study aims at identifying and selecting appropriate statistical and non-statistical tools for practical data analysis and interpretation. In business analysis, data analysis and interpretation should be applicable to business decision making, it should make sense when presented to a professional audience.
Background
The main essence of data analysis and interpretation is to give meaning to what otherwise would be a mere collection of numbers and or values. However, the significance of data in business research depends on the clarity with which the researcher defines the research question or problem. This implies that, one will need to edit data properly even before analyzing it in order to detect errors as early as possible. This can be done in a number of ways such as by conducting consistency checks and range checks-and what such data can or cannot accomplish in the research. During the processes of data analysis and interpretation, the researcher needs to be well versed with basic techniques such as data coding, why and how it is used. In addition, the researcher may need to familiarize him or herself with the meaning of various basic business statistics terms used in the characterization of mathematical attributes of the different types of variables, i.e., categorical, ordinal scales, nominal scales, interval scales, ratio, and count, discrete and continuous among others. Most business research will require that the researcher should have clear understanding of the mentioned type of variables, in addition to these; the research may also find it necessary to have an understanding of the meaning of a "derived" variable and the various types of derived variables.
Furthermore, the researcher must be able to recognize the advantages and disadvantages of the different kinds of variables and how to treat them in different ways.
The Objectives of statistical hypothesis tests (also known as "significance" tests), the importance of the outcomes from these tests, and how to interpret them are also pertinent issues to business research. In general, business research has rather taken a more dogmatic approach to statistics but this should not be so since statistics is an integral part of research in general (Creswell, 2003).
Literature review
Data analysis and interpretation always come after data collection which is accommodated in the research design. Research design in most cases takes the form of qualitative, quantitative or even a mixed approach. This implies that data analysis can be qualitative, quantitative or both.
Qualitative data analysis and interpretation
Creswell (1998) defined qualitative research as,
"An inquiry process of understanding based on distinct and methodological traditions of inquiry that explore a social or a human problem. The researcher builds a complex, holistic picture, analyzes words, reports detailed views of informants and conducts the study in a natural setting" (p. 15).
In business research, qualitative methods provide answers to a number of questions. Ritchie and Spencer (1994) summarized these questions into four categories: evaluative, strategic, diagnostic and contextual. In this case, it is clear that qualitative analysis questions are answer using qualitative data. Proponents of qualitative data argue that, it is best used for in-depth comprehension of a given problem. This type of analysis answers questions like: what, why, and how (Silverman, 2000). Quantitative analysis on the other hand, utilizes statistical techniques in data analysis. There are majorly two types of statistics; descriptive statistics-as used in the analysis of non-randomly sampled data and; inferential statistics- used in the analysis of randomly sampled data. The methodology gives answers on percentages of distribution, rating, variability of data, relationship between two or more variables, and statistical significance of the results among others. Despite of the empirical advantages of quantitative data analysis, it can be noted that the methodology does not adequately provide answers to questions on what, why, and how a phenomenon occurs (Denzin & Lincoln, 2000; Silverman, 2000). In order to clearly understand the processes or the what, how and why of a given occurrence qualitative research methodology offers the necessary comprehensive and exploratory techniques to achieve a clear view of the process. Collis, Hussey and Hussey (2003) posited that qualitative research when used in the business environment it offers a stronger premise for analysis and interpretation since it is derived from the natural environment of the occurrence. However, it is imperative to note that, the type of methodology utilized by any research is dependent upon the main research question or problem (Denzin & Lincoln, 2000).
In qualitative research methodology, the researcher has to gather necessary information through observation, questionnaires, and surveys among other methods of data collection. During this process, data must be edited before presentation as information; this is done to guarantee accuracy. Editing of data can be done manually or electronically. Another of qualitative data analysis involves handling of blank responses, this is for questionnaires. If more than 25% of the questionnaire is blank, it is discarded. Coding is also an important part of qualitative data analysis. This is a final phase of quantifying qualitative data. Answers are coded in a manner that they can be understood by the computer. In coding, data sets are systematically condensed into smaller sets. Categorization is the process that follows after coding. In order to categorize data, it is imperative to divide it into classes or segments that are mutually exclusive. For example age, religion, gender etc. Nominal scales are utilized in this process. Note that categories are based on the research query. Whether one is talking about per capita income, spending habits, or risk, categories must be well and the items be listed in a reference table.
After categorization, entering data is now what the researcher needs to focus on. In contemporary times, technological advancement has made recording of data easier. Data can be collected on an answer sheet which can be scanned into a computer. This will enable the researcher to save such data directly into the computer. Alternatively, raw data can be manually fed into computer as a file. In such cases, software like SPSS data editor is very useful for entering, editing and viewing. This has made it easy to add, alter or even delete some values after the data has already been entered into the computer.
The final phase of qualitative research is data analysis and interpretation. Researchers have identified three major objectives of data analysis: to Get a feel of the data, to access the validity and reliability of the data and for purposes of hypotheses testing. Having a feel of data occurs when statements are well summarized. Tools of descriptive statistics are used for breaking down large data sets into smaller meaningful indicators presenting central tendencies and dispersions. There are three measures of central tendency that are used in statistical analysis, the mean, the medium and the mode. Each of the measures is designed to correspond to a particular score. The choice of the measure is dependent on the type of the distribution (normal or skewed) and also on the type of measurement scale, whether nominal, ordinal or interval.
The second purpose of analyzing data is to check for reliability and validity. In order for the data to be of use to the management, it should be both valid and reliable. Reliability is the measure of the dependability or the steadiness of the data. Validity represents the authenticity or genuineness of the data. The use of multi-methods helps in providing an in-depth data and also in validating the findings thus in turn increasing the research's reliability (Yin, 2003).
Finally the purpose of data analysis is to conduct a hypotheses testing. Once data has been cleared-has passed the test of reliability and validity, the next task is to test the hypotheses formulated for the report. A hypothesis can be null or alternative.
Quantitative data analysis and interpretation
Quantitative data analysis utilizes statistical tools such as probability distributions, Measures of central tendency-measures similarity of data, and Measures of dispersion- measures dissimilarity in the data. In addition there to the above, there is the Measures of Central Tendency-these include: mode-observation with highest frequency, median- simply the mid-point, and the mean- this is the expected value of the data. The statistical technique to use is dependent on the type of data set a researcher has. The main type of measurement scales are nominal, ordinal, interval, and ratio scales. Data is analyzed using the various measures such as; Measures of Dispersion-these include the range (the difference between the highest and lowest data
Values), and the Standard deviation -measures the variability from the mean. Standard deviation is superior to the range since it allows each particular case to affect its value. Range is advantageous due to its simplistic nature.
As mentioned above the type of variables as used in research should be well defined otherwise, the research would simply lose its meaning. There two main types of variables; Independent and dependent variables. Independent variables are the type of variable that explain changes in the dependent variable. For example, when valuing returns on of a stock, political risk is an independent variable since it affects the returns of a stock; Dependent variables (also known as explanatory variables) are the variables whose behavior the researcher wants to explain. As in the example above that would be the returns of a stock.
Statistical Significance
Statistical significance tests are tools used to estimate the likelihood of the results being wrong, how likely that the results of the analysis are statistically significant (they are not obtained by chance) within a certain margin say 95 percent. A test for significance is an estimate of the probability of obtaining the results by chance if there were no differences in the population. Most of the data analyzed does not go beyond 0.5.
Important statistical tests
The most commonly used statistical tests include Chi Square and t-Test. These tests are popularly used because they are easy to calculate and interpret. They are used to compare nominal data sets (such as marital status and age). These tests are also used to compare ordinal variables or even a combination of both nominal and ordinal variables. In addition, they are used to investigate whether one group of numerical results is statistically different from another group of results.
Hypothesis Testing
In business research, hypothesis is defined as the best guess of the relationship between variables. For example, 'there is a difference between GDPs of less developed countries and those of industrialized nations'. Normally, the researcher can state null and positive hypotheses (the above statement is a positive hypothesis) as the study demands. A null hypothesis is always a statement that negates the positive hypotheses. For example, 'there is no difference in the GDPs of less developed countries and their developed counterparts.
Linking Qualitative and Quantitative Data
In business research, the one of the questions researchers face is, Should quantitative and qualitative data analysis methods be linked when designing a study? The questions here are how and why? In practice, there are significant linkages between qualitative and quantitative data analysis methods. These linkages are done using techniques such as triangulation (corroboration). Triangulation is the use of more sources of information to give data more plausibility. This is done by using three or more methods of data collection such as; interviews, questionnaires, observations, historic data and expert panels among others. The importance of linking qualitative and quantitative data analysis method is to have a richer detail, to initiate new schools of thought and to expand the scope of the research. Defining the problem is critical to conducting a successful business research, data analysis and interpretation. This may sometimes demand a great deal of time but this is worth more the time and energy spend. Defining the research problem is tantamount to a successful data analysis and interpretation. Many researchers waste valuable time conducting good research on the wrong research problem.
Role of statistical computing in data analysis
The rapid and continued increases in computing technology dating back from early 20th century have had a significant impact on the practice of business research especially data analysis and interpretation. Early statistical analysis models were solely linear models. However, powerful computers, installed with data analysis software such as excel, STATA, E-views etc have stimulated analysts' interests in nonlinear models such as Logit models as well as the development of new models, such as multilevel models and generalized linear models (GLMs). In addition, the increased use computers has also led to the creation of powerful computational methods which are based on re-sampling techniques, such as the bootstrap, Monte Carlo simulation, and permutation tests. Other models such as the Bayesian model have been made much more feasible due to availability of techniques such as Gibbs sampling. It is apparent that the computer revolution will have even greater implications for the future of data analysis and interpretation, there is will be more emphasis on empirical and experimental data analysis.
Conclusion
In this article, data analysis in business research involves various steps. First, necessary data is collected through various data collection methods such as questionnaires, surveys, experiment etc. The data is then edited to ensure that only accurate data is used in the research. Next, data is coded, categorized, and entered in tables or computers. After this, basing on the research problem, hypotheses are formulated and tested using the most suitable and reliable measures. The results are then interpreted and an appropriate solution is made to answer the research problem. Researchers will often find data analysis as being the most enjoyable part of carrying out a business research, since after all the toiling and patience they now have a chance to find out the solutions to the research problem. In fact, if in any case the analyzed data do not answer the research question(s), this can be regarded as yet another opportunity to be creative! Thus, the analyzing and interpretation of business data followed by the presentation of results are indeed the "fruits" of the work of collecting, coding and recording of data. However, data does not have a voice of its own. Data analysis is simply used to reveal that which the analyst can detect; this is the whole essence of research. Hence, when a researcher, attempting to harvest these fruits of research, finds him or herself alone with no clue on how to analyze and interpret a dataset, there can be a feeling of anxiety rather than eager anticipation. In general, analysis and interpretation of business data should clearly relate to the research objectives and questions. Most analysts will approach would be to begin with descriptive analyses, this helps in exploring and gaining a "feel" for the data set. This helps the researcher to address specific research questions from the report or hypotheses, from research findings and study questions as reported in the literature review, and from trends and patterns presented in the descriptive analyses. This will ensure that data analysis and interpretation in business research is relevant and thus progressive.
