This short introduction is intended to assist you in writing your own research paper. It begins with the idea of an ideal-typical construction of a research paper; Afterwards some practical tips will be discussed, ranging from the topic search to the creation of a bibliography. Some practical tips for creating your own final project can be found at the end of this short introduction.
- Preliminary remarks
The formal structure of research paper is often knit according to a fairly simple pattern; it is done to a certain extent after cooking recipe. As with cookbooks, scientific papers are as well: good cooks do not need cookery books, and ingenious cooks even make the most wonderful ingredients to conjure up wonderful things, but for novice and mediocre chefs, recipe templates can be quite helpful. In this sense, the following should be understood, not as a strict framework, but as a rough guide. Writing scientific texts is a craft best learned through practice.
- Ideal-typical construction of a research paper
Research paper often has the following structure:
- Introduction: Motivation and concise presentation of the problem, possibly outlook on the structure of the work.
- Literature Review: The ‘most important’ papers published in the immediate vicinity of this research question. Specifically, the specific question, the method used, the data basis and the results should be outlined.
- Economic (theoretical) model: representation of the presumed action channels, and consequent hypotheses. If possible, to what extent does this lead to consequences for econometric modeling (is it to be expected that one or more Gauss-Markov assumptions will be violated due to the correlations?)
- Data basis: Definition of variables, sources, descriptive statistics (outliers, distribution, scale level, etc.).
To what extent do data actually measure what should be measured by theory?
- Econometric model, estimation method and assumptions: which estimation method best suits the theoretical model and the data, and why?
To what extent are endogeneity problems (e.g. missing relevant variables, correlated with the considered variables, simultaneous causality, measurement errors in the explanatory variables) to be feared? How are these problems taken into account in the estimation?
- Results, Discussion & Conclusions: To what extent do the estimated parameters correspond to the (theoretical) expectations, significance of the individual parameters and possibly shared significance.
To what extent are assumptions about perturbations met (e.g., heteroscedasticity, autocorrelation); Results of specification tests.
Possibly: to what extent are the results ‘robust’ against errors in the assumptions, or what consequences would a suspected false specification probably have?
- Summary of the results, indications of a correct interpretation, outlook on future research, …
- Bibliography (References)
- Appendix (appendix): contains e.g. detailed evidence, long tables, program code, etc.
In summary, from the introduction of a paper:
“The remainder of this paper is organized as follows: Section 2 summarizes the literature, section 3″ explains the empirical methodology, section 4 describes the data, section 5 reports the empirical results, section 6 examines the robustness of our estimation assumptions, and section 7 Concludes. ”
A good piece of work should be compact in structure and well formulated. It is by no means necessary to treat all the above points as separate sections; individual points can be summarized or left out altogether. The general principle applies: KISS (“keep it short and simple”).
- Some practical tips
However, a lot of problems have to be solved before actually putting a paper on paper. The problems begin with the choice of the topic and extend to the formatting of the finished work. Good research paper writing help from aspiring researchers can be found at the good writing service. At the center of a scientific work are one or more research questions that are located within a more general research topic.
3.1. Find a topic
Start from your own strengths (prior knowledge). Think well before you get involved in a whole new field. Look for ideas in the world, not in literature (see Varian, 1997).
Do not try to solve all problems at the same time, look for detail problems; think in questions, not in answers!
3.2. First steps
In the first step, try to narrow the issue as closely as possible. Formulate assumptions (hypotheses) that – at least theoretically – can be wrong. Make yourself curious! If you are brave enough, do not try to grasp a topic in all its completeness and with all sorts of branches (do not start with the historical beginnings or with a pure textbook representation of all known theories!).
Do not try to find the one true and correct view, but develop different views and approaches. Try to get to the core of each view through abstraction.
If you have an idea of a suitable question
Try to summarize the basic concept and the research question (s) in the form of a short description (abstracts, approx. 200 words).
Brief description (abstract):
- Simple and accurate representation of the problem.
- What is the current view of the problem (possibly with a maximum of 1-2 central references).
- Brief description of the “research design”:
(a) Economic Problem (Model)
(b) Data basis
(c) Estimation method and test strategy
- (Expected) results
Only start with an extensive research if you have a concrete idea of your question.
Do not start with the literature that leads “your” keyword (random) in the title, but start more generally and orient yourself rather to the quality (reputation of the author and the publisher / journal). It is difficult later to escape the impression of the first article, so this should not be a random choice.
Start with (advanced) textbooks. Review articles (e.g., Journal of Economic Literature, Journal of Economic Perspectives). Try to identify “core people” (or “core institutions”) in your field! Who is the most frequently quoted?
Checklist for empirical work
- Is the question described in simple, clear words?
- Is the embedding of the question in the relevant literature clear?
- Are relevant data available and is the source of the data well documented and trustworthy (‘citable’)?
- Is the exact definition of the variables known, and to what extent does the definition correspond to the question?
- Have the scanned data been checked? (Descriptive statistics, graphics)
- Does the scale level correspond to the data of the selected method?
- Is one or more of the following problems to be expected?
– Simultaneous causality
– Missing relevant variables
– Measurement error in x variables
- Additional problems with time series:
– autocorrelation (especially with delayed endogenous variables)
– Structural breaks (graphic: Quandt-Andrews, CUSUM, …)
Stationarity: deterministic and stochastic trends (ADF…)
- Wrong function form (RESET test)
- Outliers (descriptive statistics, graphs, studentized residuals)
For the actual empirical work on the computer there should be some more punk attention:
- Create a separate subdirectory (Folder) for each project. Avoid Windows standard directories (confusing, contain spaces and special characters), consider a more suitable directory structure!
- Store data locally and format it so that it can be read by an appropriate program (for example, valid names as variable names, that is, no spaces or special characters in variable names, short descriptive names, etc.). Do not make any changes to the data at this stage (do not delete outliers, etc.)! Make changes to the data later in the program so that all changes are reproducible! Create backup copy!
- Sublime programs use a (period) as a decimal separator, comma (‘,’) is used as a symbol for number grouping. It is also often advisable to set the (dot) as decimal separator and comma (‘.’) For digit grouping in the operating system.