Analysis of Research Data Using EViews

Introduction

The book, “Analysis of Research Data Using EViews”, is published by Jovila Educational Services, Lagos, Nigeria. In this website, you are presented with the online version to learn data analysis. The contents of the online version and the print version are the same but pagination differs. Therefore, any references, in terms of paraphrasing and direct quotation, should indicate page numbers or chapters as appropriate, somewhat as under:

Direct Quotation

“If the data distribution of a variable is not symmetrical, we calculate the median and its semi-interquartile range” (Avwokeni, 2020, ch.8 p.77).

Paraphrasing

The median and the semi-interquartile range are the appropriate descriptive statistics when the data distribution is not symmetrical (Avwokeni, 2020, ch.8).

References

Avwokeni, A. J. (2020). Analysis of Research Data Using EViews. Jovila Academy Press. Available at https://jovilaacademy.com

Organisation of the Book

The book is organised into seven parts, as under:

Part I: The EViews Software

This part of the book teaches how to operate the EViews software. In this Part, we take you round the EViews environment, teach you how to create an undated workfile and enter the research variables. We also teach you how to enter data for each of the research variables by typing data directly, by copying and pasting from Microsoft Word or Excel, or by importing from Microsoft Excel to EViews. We introduce you to how to calculate descriptive statistics and change the sample size to re-calculate descriptive statistics.

Part II: Analysis Associated with the ANOVA Framework

In this part of the book, we teach you how to conduct those analyses one does in the ANOVA framework using the regression analysis framework. Specifically, you will learn how to

1. conduct the Pearson’s chi-square as a regression analysis for one and more explanatory variables.

2. conduct the independent t-test as a regression analysis.

3. conduct One-Way-ANOVA, Two-Way-ANOVA as a regression analysis.

4. conduct analysis of variance as a regression analysis.

As a handsome bonus, we teach you how to conduct equality test and normal distribution test.

Part III: Survey Models

In this part of the book, we teach you analyses associated with survey research. Specifically, you will learn analysis to answer the following research questions.

1. Explaining frequency. To explain, supposed we observed that a friend of ours is being hospitalized monthly, we may ask, “why the frequent hospitalization?” In other words, what factors explain the frequent hospitalization of our friend? We teach you the analysis required to answer this kind of research question.

2. Explaining Opinion or Choice. Opinion should be explained because it is not immune to outside influence. There are factors influencing opinion or choice and these should be detected for a complete explanation. As an example, we may conduct an opinion survey to determine whether companies collapse after a clean audit certificate has eroded public confidence in the audit profession. Supposed the respondents were stockbrokers (representing investors). 60 percent of the stockbrokers asserted that there is no more trust of audit certificate while 40 percent asserted that professional audit certificate is required for someone to take responsibility. The fundamental question that remains to be answered is, “what factors explain the opinion of each group of stockbrokers?” You will learn how to do analysis to answer this kind of research question that involves explaining opinion or choice. As a handsome bonus, we teach you how to calculate marginal effect.

3. Sample Selection Bias. When you administer a questionnaire, some copies may not be useful. The reason may be that those copies were not completed as requested (e.g. missing data, incorrect filling, etc.), or that the copies were even empty (i.e. returned blank). By discarding “bad copies” of the returned questionnaire, you have excluded some respondents from the sample of the study. In calibrated pedagogy, you have “truncated the sample”. If this happened, the remaining sample may be biased. Therefore, we need to test whether the truncated sample is biased or not. We teach you how to test whether a truncated sample is biased, and what to do if the sample is biased.

Part IV: Regression Framework for Quantitative Variables

There are two broad brush objectives of a researcher who embarked upon a regression analysis. The first is to quantify the relationship between each of the independent variable and the dependent variable. This means that when we do a regression analysis, we can tell whether the relationship between the independent variable and the dependent variable is zero, negative, or positive. The second is to quantify the effect of each independent variable on the dependent variable. Coincidentally, if the effect is not significant, then the relationship between the independent variable and the dependent variable (if any) is also not significant. In that case, we have zero effect and zero relationship. In this part of the book, you will learn to

1. Conduct a regression analysis when the data distribution of each research variable in the research setting consists of figures, not category of a research variable. To explain, the data distribution of Age consists of real figures such as 25 years, 30 years, etc. But the data distribution of Gender consists of male and female. So, you will learn to do a regression analysis when the data distribution of each research variable consists of figures, not categories. Specifically, you will learn:

• Joint Hypothesis Test. To explain, in Accounting and Finance, the liquidity status of a company can be measured by current ratio, cash ratio, cash flow yield, cash cycle, etc. Now, supposed your regression equation is that the market price of a firm’s stock depends on liquidity, earnings, and equity. A germane question is, “which observed variable will you use to surrogate liquidity, given that we have several observed variables?” It is unethical to pick one and leave the other without an explanation. One approach is to substitute all the observed variables to replace liquidity in the regression equation. When you do this, you cannot interpret the relationship and effect of current ratio, cash ratio, etc. on the market price of stock. This is because they all measure one variable—liquidity status. It is like developing questionnaire items to measure a concept such as attitude towards miniskirts. Then, you choose one of the questionnaire items to represent “attitude towards miniskirts”. You cannot do this because other questionnaire items also measure attitude towards miniskirts. All items contribute to the measurement battery; therefore, you must include all the items that measure attitude towards miniskirts.

To interpret the relationship and effect of liquidity on market price of stock, you must conduct a joint hypothesis test. So, we teach you how to do so.

• Interaction effect. To explain, supposed the regression equation is that the market price of a firm’s stock depends on liquidity, earnings, and equity. We observed that equity depends on earnings. When this is the state of affair, we cannot just estimate a regression equation that states that market price of stock is a function of liquidity, earnings, and equity and conclude. We must also estimate a second regression equation that states that market price of stock is a function of liquidity, earnings, equity, and the interaction of earnings and equity. We teach you how to do the analysis, interpret and conclude.

• Multicollinearity. The direction of relationship and effect size from a regression result is affected by multicollinearity and inappropriate model. We teach you how to assess whether the linear model you specified to analyse your data is appropriate and whether it is plagued with multicollinearity.

2. Extend a regression analysis to path analysis. When you run a regression analysis for your hypothesis, do not just take the results for what they are. If one of the variables fails to make a significant impact, then probe further. You may begin with an interaction probe if there is a theoretical reason to justify it. If interaction is not significant, then move to path analysis. Try reasoning out what other variables could have supervened to suppress the effect, and if you are able to theorise such other variables, then extend the regression analysis to path analysis. This will bring out the real effect. We teach you how to extend a regression analysis to path analysis.

3. Analyse simultaneous Structural Regression Equations. There are research questions that lend themselves to simultaneous structural equations. In the Finance discipline (as an example), questions such as: “Does the use of debts in financing affect liquidity?” “Does the use of debt affects free cash flow?” abound. We might develop the following hypotheses from finance theory:

(a) Liquidity is affected by the cash cycle of a firm, the amount of debts to service, the available stream of free cash, and dividends policy. (b) Free cash flow is affected by liquidity, debts, and cash flow yield.

If you examine our hypotheses (please, do), you would find that liquidity which is a dependent variable in model (a) is an independent variable in model (b). Similarly, free cash flow which is an independent variable in model (a) is the dependent variable in model (b). This state of affairs poses a simultaneous bias, and hence should be handled specially. We teach you how to handle a regression analysis when the simultaneous bias is present.

Part V: Methodology of Factor Analysis

Factor analysis is a great technique to learn. It can be used to do a lot of tasks.

1. We can do factor analysis to measure a latent variable in a regression equation. To explain, consider a hypothesis which states that the market price of stock of a firm depends on liquidity status, earnings, and equity. Of these variables, “liquidity status” is a latent variable because it can be surrogated by several observed variables. Examples include cash cycle, cash ratio, current ratio, free cash flow, and cash flow yield. We can do a factor analysis to select the most representative surrogate.

2. We can do factor analysis to develop a measurement scale for a variable, e.g. we can develop a scale to measure “corporate governance” and publish the scale. Then, in any research in which corporate governance features as a variable, scale would be invoked to measure it.

3. We can do factor analysis to prove that items in a questionnaire are valid or reliable measure of a latent variable.

4. We can do factor analysis to confirm that opinion or perception is the cause of an event.

The list of what we can do with factor analysis is endless. In this part of the book, we teach you how to do factor analysis to accomplish the above task and many more.

Part VI: Pool and Panel Analysis

A pool analysis is done to test the differences between regression coefficients. If your study involves comparing two groups in terms of a hypothesis, what we should do is to model the hypothesis as a regression equation and estimate it separately for each group. Then, we conduct a test to check whether the coefficients differ. This same strategy should be used to determine whether the two groups could be pooled into a single sample so that a single regression equation can be estimated. If this were the objective, then the two grouped could be pooled into a single sample if the result of the test shows that the regression coefficients do not differ. You will learn this great technique in this book.

A panel analysis is appropriate if one wants to detect whether the behaviour of a variable has changed over time or whether the behaviour of a variable changes owing to an experiment. We may even suspend “time” in the air and do the analysis to detect whether behaviour changes irrespective of time. You will learn how to do all of these in this great book.

Part VII: Time Series Analysis

When our data are time series distributions, we can do analysis to accomplish a lot of tasks.

1. We can study the structure of a time series variable to know whether if we model the series and use it for prediction or forecast, time will supervene to distort or suppress the prediction. As an example, we may study the underlying structure in the distribution of tax revenue to learn whether budgeted figures are reliable or not.

2. We can do analysis to detect the direction of movement of a variable in short-run or long-run analysis. This is important for planning purpose. As an example, we may do analysis to detect whether forex rate of three countries move in the same direction or opposite direction. This information can be useful for hedging purpose.

3. We can do analysis to detect whether there is a structural break in a distribution as a result of some event. As an example, we may examine the distribution of tax revenue before and after COVID-19 for any break due to the pandemic.

4. We can do analysis to forecast some value.

5. We can do analysis to detect volatility in the distribution of a variable. The result is useful for planning purpose.

You can do these and many more after reading this great book.

Is the Author Simple on Readers?

Practice is illustrated by examples. So, readers can follow the explanations and examples to practice analysis. However, we would like to announce that this is not an econometrics text, but a text to illustrate data analysis in research projects. Please, bear this in mind before you choose to subscribe. Our video lectures further demonstrate the fundamental elements of the analysis described above.

Pricing Policy

You can subscribe to read the book for a period of 3 months, 6 months, and 12 months. Your subscription places you in a membership category:

• 3-Month Membership
• 6-Month Membership
• 12-Month Membership

If you have no knowledge of EViews and would like to learn how to operate EViews and use it to conduct analysis, we advise you sign up for 12-Month membership. If, however, you already know how to operate the EViews software, then you may sign up for 3-Month or 6-Month Membership, depending on the project you have at hand. If you are a PhD student, 12-Month Membership is advised. Masters and senior undergraduate may sign up for 3-Month Membership. Lecturers and professors may sign up for 6-Month or 12 Month Membership, depending on the frequency of research project they undertake.

Payments are made secured through Flutterwave Incorporation. Flutterwave is a U. S. company that handles intermediary cash payment between any parties.

Payments are made in U. S. Dollar but you may choose to pay in Pounds (£) or Naira (N) but must mail director@jovilaacademy.com for terms and conditions.

Flutterwave accepts MasterCards and VisaCards. Once you have paid, Flutterwave will issue a receipt document and send us a receipt duplicate. Your log in information will automatically be activated.