How to structure empirical term papers
Table of contents
If necessary, list of tables, list of figures (only add these lists if necessary; do not use decimal numbers for the structure; can be placed at the end of the paper)
Situatedness of the topic in the author’s field of work
Informs the reader why the author chose this specific topic or what current occasion caused the author to choose the topic.
Explains the relevance of the topic.
Describes the subject as precisely as possible without the detailed description the conceptual apparatus of chapter 2 offers. (Even nuances can be important if they distinguish the work from other papers.)
Asks questions that the paper seeks to answer or formulates hypotheses that the paper is ought to support.
Sets the limits – what is the paper about and what is it not
Specifies the issue and focusses on some of its aspects.
It often makes sense to explicitly tell, what is not being examined, especially if some aspects are apparent but not part of the examination (But do not rule out anything that no one would have considered anyway).
Describes the procedure.
The introduction needs to inform the reader about what to expect from the paper and prepare him for the following line of argumentation. Do not introduce new concepts, terms, etc. during the following course of the paper.
Avoid misleading the reader with extensive passages that have no relation to the topic. Make clear what is background- or additional knowledge that helps to locate the paper and what is the actual subject of your work. It is legitimate to mention additional information for the reader but it must not qualitatively or argumentatively exceed the actual subject.
In empirical papers, long treatises on theoretical and historical traditions or public discourse are to be avoided. Usually, a short hint at central positions and references to additional literature are sufficient.
- State of research and postulation of hypotheses
What is known about the topic?
What can the research contribute to answering my questions?
Report on the results. Only include details if they are important for the outcome.
Cluster your work thematically. Do not jump from one study to another. In general, do not print detailed lists of results from other studies. Instead, put insights into your own words and make their relevance perceptible for your own research.
Do not enumerate on what is irrelevant in the works you cited. Criticize, if necessary, the theoretical framework or methods of other studies.
Which researcher uses which terms? How do they differ?
Are they useful for your paper?
Do they make causal statements or do they simply publish descriptive findings? Look for content-related explanations of empirical findings.
Are there any gaps in this kind of research? Watch out for the congruency between the depicted state of research and your own work. Do not just repeat what some branch of psychology has accomplished so far. The depictions you make have to be entrusted for a specific purpose and pave the way for your analyses. Check all terminological determinations, differentiations, statements about coherence, and findings for their relevance for the topic.
Which own assumptions (hypotheses) do you as an author have yourself?
How can they be justified with regard to the current level of knowledge? Show what relation your work and findings have to the research that you have cited.
A common mistake is the justification hypotheses through empirical observations or references to former studies that have come up with similar results. Justification requires explanation, meaning mediating mechanisms. So do not state that a connection has often been observed, but rather explain why there should be a connection.
Which terminological determination or shift do you perform?
Where do you follow other researchers and where do you go your own way?
Only formulate hypotheses that you can actually test.
3. Data and methods
Selecting the form of the examination
Why is it perfectly suited to answer the research question?
What are the alternatives?
Would a secondary analysis of existing data sets have been possible? What other selection criteria needed to be taken into account (research economical reasons, etc.)?
Survey method: How was the data collected?
Basic population: About which social entities do you want to make assertions?
Random sample: Which social entities were observed? Extent, selection criteria, inquiry period, regional aspects, special characteristics. Comparison with the basic population or similar random samples (socio-demographic basic values) and structural peculiarities (deviating marginal distribution).
Instrument of elicitation: precursors, discussion and its appropriateness for the given purpose, development (pretests), outline of the final version (extent, structure; no details of the sections that are not being used in the following analysis)
Explicitly name the origin of items and scales that you have adopted from other studies. Details of the course of the pretests are usually neglectable. An output-driven description including the number of pretests and their scope is sufficient.
For the reader, it is, however, relevant if certain constructs are, in principle, difficult to operationalize, whether scales form research reports have not worked in your study, whether doubts occurred about the validity of what went wrong in the field, whether items had to be deleted from scales, etc. It is not necessary to be embarrassed by such problems, as they are part of everyday life in empirical social research. Reputable reports do no try to cover up these problems as they can serve as a learning opportunity for the readers.
Operationalization of the theoretical concepts.
Do cite literally, completely and precisely.
Do not describe items that do not occur in your analysis.
References to item numbers are only useful when they are available for the reader (for example as part of the appendix).
Discuss, if necessary, the validity of your operationalization (especially important for secondary analyses).
Present introductory questions for item batteries.
Describe generated variables and indices. Confine yourself to the logic of variable generation (aggregate index generation, OR-relations of conditions, dichotomization, etc.) or the used algorithm. Details about the programming in particular software products are rarely useful. They can, however, be placed in the appendix. Please take into account that the “generation of variables” is no section on its own, but an addition to operationalization that does not apply variables as directly, as they have been collected. Also, state how you plan to test the alleged connections statistically and under which circumstances you see your hypotheses as confirmed. This can happen in one sentence. For example: “I will test the correlation of X and Y and accept the hypothesis if the result is a statistically significant correlation.”
If necessary, explain uncommon analytical methods or particular statistical techniques.
Descriptive, univariate depiction.
- Whether a univariate depiction makes sense cannot be generally stated, as it depends, among other things, on whether or not the categories possess informative content on their own. With natural scales or categories, this is more likely than with artificially constructed scales that feature arbitrary values, that can hardly be interpreted by the reader. It can, however, makes sense to point out lopsided distributions, as they can have an impact on the quality of the scale. Apart from that, it can make sense to point out surprising distributions. This, however, requires explaining why the distribution in question is deemed as surprising. In the end, it is decisive what significance the particular distribution has for the overall statement of the paper.
If you want to point out textual findings, a tabular or graphical representation is recommended, as long as you verbally explain it. It can be neat to connect a univariate descriptive statement with a correlational statement by, for example, tabling the average value of two subgroups (e.g. migrants and natives, men and women, etc.) in and note the overall median average in the same table. Try to summarize similar details in tables, as you will save a lot of space with this method. But remember to always have a textual reference to the tables. A solution for a number of small results, that might be of importance for interested readers, but not for a broader audience should be placed in a table appendix.
What is true for univariate results in tabular form is especially true for illustrations (bar- and pie charts). If they are used too often without showing any special abnormities they appear redundant and might bore the reader.
As hypothesis tests will represent the core of your argumentation, there is hardly any clear alternative to the tabulation of your decisive results, especially, when you apply multivariate analyses. Only a table provides a clear depiction of the degrees of impact (standardized and non-standardized) and at the same time allows to read the significances. Don’t forget to mention the sample size in tables.
5. Conclusion and discussion
- Conclusion of the findings (that hold no new insights!) Interpretation of the results with regard to the questions that have been asked in the introduction. Has a definite answer been found? Comparison with earlier studies that treated the same topic. Discussion of inconsistent findings. Limits of comparability. Debate the meaning of the results for politics and practice.
Which questions remain unanswered?
Which new questions have surfaced that need further research?
Describe problems that arose during the research process.
Bibliography (complete and consistent)
Is not counted as a single chapter in the structure.
Appendix (if necessary)
Contains text passages and detailed results in the form of tables and illustrations that are expendable for the argumentation but should be taken into account for those, interested in further investigation of the topic
In extensive papers please subdivide chapters into smaller subchapters. In this regard, there are no standards defined, as systematics vary depending on the extent and the theoretical and empirical focus.
University of Bielefeld – Kurt Salentin
copyright by PIB-Zentrum.de