Dr. Asad Zaman
Based on experience with a lot of proposals, here are some very common problems which lead to rejection of proposals.
Despite repeated warnings and explanations, students continue to plagiarize, both in their proposals and in their submitted theses. I have prepared a separate note on plagiarism, which can be downloaded from the website: https://sites.google.com/site/iiieproposals/iiie-research-proposal
Here are two simple rules to avoid plagiarism:
1. Any sentence which is copied directly (cut & paste) MUST be in quotation marks and SOURCE must be cited.
2. Any ideas, theories, models, equations which are copied ( even if not cut and paste, and even with re-wording) must be referenced: this model was introduced by Rousseew (1990) [for example].
An annotated bibliography is a list of articles which describes the contents of each article. Most students think that this a literature review. This is NOT true. An annotated bibliography (AB) is the first step towards a literature review (LR). A literature review for a proposal must have the following characteristics:
1. It is organized by themes or ideas. For example a LR of Consumer Satisfaction Islamic Banking could be organized around the following ideas:
[1] Is Islam the major motivation for customer choice of Islamic Banks? On this issue XYZ says yes and produces following evidence. VW says no and argues as follows. ABC says that some are motivated Islamically and others are motivated by profits and stability.
[2] Do Customers Have Knowledge of Islamic Laws & Shariah Compliance Status of Islamic Banking Products? According to Survey by GHE, customers are very knowledgeable. According to PQR, customers don’t know anything.
[3] Are Customers Satisfied with Islamic Banks? Different authors have come to different conclusions.
[4] Etc. Any other themes, concepts or ideas about which literature exists.
2. Only Literature RELEVANT to research being carried out should be surveyed. The connection should be made clear. Some examples of what to do and what not to do are listed for clarification:
DON’T’s
[1] Theme: Difference between Muslim and Non-Muslim Clients of Islamic Banks. If we are studying Islamic Banking in Pakistan, then this issue SHOULD NOT be covered in literature survey, since this is not an issue here in Pakistan. Literature for Indonesia might discuss this, but we WILL NOT cover it, because we do not plan to do research on it.
[2] Author XYZ has estimated a Factor Analysis Model of Consumer Satisfaction and obtained the result that Distance to Bank and ATM’s are the most significant factors. This study can be cited ONLY if you plan to do factor analysis and use these factors in your study. If you are not doing factor analysis and/or you have no data on distance and ATM’s the DO NOT cite this study.
DO’s
[3] Related to Issue [2] in previous point: Are customers knowledgable about Islamic Law? As we have seen there is a controversy in the literature. Accordingly we have designed a questionnaire asking some simple questions about Islamic law. From these we will be able to decide whether XYZ is right or whether PQR is right – this type of sentence shows the relation between YOUR research and the previous literature.
FOR MORE DETAILED AND SPECIFIC INSTRUCTIONS on how to make an annotated bibliography, and how to convert it into a literature review, SEE LINK.
Author writes these are the research questions/hypotheses for the study: A, B, C.
I will find the answers to these questions. TWO issues need to be clarified to make sure that the proposal is worthwhile:
1. What is known about the answers to these questions uptil now? This is what literature review is planned to do. So from this, we learn what ADDITIONAL knowledge we will gain, which is not known uptil now.
2. What difference will this make? How would this knowledge help us in making decisions, or in understanding the world.
A more detailed discussion of common econometric errors is given here
A very common misunderstanding is that a thesis consists of estimating a regression model. We look at a data base. For example, the WDI data set has data on Income Inequality (II), Tax Rates (TR), Volume of Exports (VE), and Life Expectancy (LE). Now a student will say that I plan to explore the impact of three variables, II, TR, VE on Life Expectancy. This is new research because no one has done it before. I will run the regression and find out if Tax Rates have adverse effect on LE or not.
This is ridiculous. Most regression of variables U,V,W on some dependent variable Y DO NOT MAKE SENSE. We cannot arbitrarily pick a collection of variables and regress them on some others and then analyze results. THERE MUST BE A CAUSAL RELATIONSHIP, so that the exogenous variables determine the endogenous one. This relationship must be justified by some theory or some existing models in the literature. The set of variables being used must be complete. If someone says that I am going to examine impact of luxury imports and remittances on consumption, then he is omitting the most significant variable (income) and the regression is misspecified. Here are some key issues in setting up a good regression model, which can be estimated properly:
1. Make sure the regression is similar to something already done in the literature. This provides the model with some credibility, since hopefully previous researchers have done their homework in setting up the regression model. This is NOT a guarantee, since there are plenty of wrong models being estimated in the literature, but it does help to provide some validity to the regression model.
2. The regression model must have some theoretical justification. Again, this will usually come from the literature review. For example the Export Led Growth hypothesis says that exports drive growth. This theory, which is then explained by some reasoning, creates a justification for using exports to explain growth. However, it would be important to specify HOW exports affects growth – what are the intermediate variables in the causal path from exports to growth. On this issue econometricians are often very sloppy. This is clarified in the next point.
3. There should be a clear causal chain from the independent variable to dependent variable. For example, a regression trying to find effect of Migration from Kashmir (M) on Crop Yields (Y) is not correct. Crop Yield will be affected by Fertilizer, technology used, labor, land etc. These are the inputs to the production process. Migration is NOT an input to the production process. Migration may affect the supply of skilled labor on farms, which will then affect yield. Then the regression must be specified in this way. M è Skilled Labor è yield. Now each part of the causal path can be investigated separately.