11.1 The basic structure

1. Introduction: the beginning of your report should be dedicated to showing your motivation; why you care about your research topic and how it is relevant to the public. Use unsolved problems, or conflicting results, as a base and proceed to outline the basic structure of the paper, where you describe what you plan to do, which theories you will employ and which methods you will apply. Formulate a clear and concise research question that is manageable within the scope of your paper. Questions like “The causes of poverty” are usually a bad idea for a 15-page paper, as they could fill entire libraries. Finding and pinpointing the research question is a crucial step that might take some time, but will greatly affect the quality of your work. Also, briefly present the current state of research, and demonstrate the contribution of your research to the field.
2. Theoretical framework: in this section you introduce the theories you will use to guide your research. Whether you use a grand-theory, or more empirically driven approaches, depends on your field, research question and style. Make sure that you dig into the relevant literature, so you know the important aspects that are used by other researchers. Formulate hypotheses that you want to test empirically. Make sure that these are compact and testable ones (usually from one up to three, for a seminar paper). They should have the form of: “The higher the value of variable X, the lower the value of variable Y” (H1).” See page 83 for an example.
3. Your hypotheses should follow logically from the theories you have outlined before. Do not explain your theoretical foundations in their entirety, as only about 20% of your paper should be dedicated to this section. The intellectual challenge is to extract the theoretical parts and mechanisms that are relevant to your research question, from the main theories which might fill entire books.
4. Descriptive statistics: when you start with the empirical part, it is important to introduce your dataset: who created it, how it was obtained, and which methods were used. Introduce the operationalization, which is about the variables you want to use, and how they are suited to your theories. It is relevant to explain how theoretical constructs, and actual variables, can be matched with each other and where deviations and problems lie. For example, when you need age of a pupil, as a control variable, because it is required by your theoretical framework, but only the class a pupil is in, is available, it is an important task to justify why it might be adequate to use this variable instead.
   Then, write about the central variables you want to use, and report some basic information like means, standard deviations or distributions (using histograms, boxplots or kernel-density plots). You can use tables to display a lot of information in compact form. Remember that each causal question requires a good description, as this often enables you to understand the data and certain relationships within it. Furthermore, include some information about how your dependent and independent variable are related (use correlation coefficients or simple crosstabs). Briefly explain the methods you want to use, and why they are able to answer your research question.
5. Statistical analysis: here you proceed with your method, and report results, and also any problems that may arise. In the first part, you should just display these numbers without commenting on them. The second part, the discussion, is used to explain how your theoretical framework, statistical computations and results of other researchers fit together. When there are large deviations from your hypotheses, it is important to look for problems or reasons that could explain these differences (for example, those that may arise, due to different datasets, operationalization or methods).
6. Summary: shortly summarize your results, and try to answer the research question you formulated in the introduction, in a few sentences. Highlight certain aspects, results or problems that seem especially important. Furthermore, give an outlook on how you, or other researchers, should proceed in the future, after considering your results.
7. References: always include all references and citations used. It does not matter whether they are for theoretical argumentation, empirical research or about your methods. Remember, missing references can be viewed as plagiarism, so always double check that you have reported the origin of direct and indirect citations! Other researchers want to know your sources.

Note that this is a very general framework which can vary depending on your field, the extent of your research and the opinion of your advisor. It is always a very good idea to talk to him or her before you start writing, and present your ideas and plans, as you will receive some feedback that can help you enormously. Also, talk to fellow students, maybe from higher semesters, to learn about certain aspects or idiosyncrasies you should pay attention to. For example, whether you put your graphics and tables inside the chapters, or in a separate chapter at the end of the paper, is not written in stone. For a more detailed introduction refer to the paper of Bhakar and Nathani (2015).

11.2 Master do-files

In chapter two you have learned how you can use do-files to save commands and structure your tasks. When you work on larger projects, like a seminar paper, you will see that it is extremely helpful to have your files and data organized. Another trick you can use, is splitting different parts of analysis over several do-files. For example, you start with data recoding and operationalization, to transform your variables, so you can use them later. Then you proceed with descriptive statistics, which gives you an overview of the data. After that, you switch to statistical methods and run your regressions (or whatever methods you want to use). You can use a different do-file for each part, and later run them sequentially with a Master-Do. The first step is to create each file, say operationalization.do, descriptives.do and analyses.do. Then you create a new file, master.do, which you put in the same folder as the other ones. In this do-file you write

***Master do-file***
do operationalization
do descriptives
do analyses

When you run the Master-Do, all the other do-files will be called in the sequence that you placed them. By doing this, you can organize and automatize your workflow even better. When you do not want to see the actual output of a do-file, for example when data is recoded, use

quietly do operationalization

When you later come back and see that you forgot to create an important variable, edit your operationalization.do and run the Master-Do again. You can also combine this design with logs, so you will receive a complete output in a file after the do-files are finished.