Reference
Gary King. 2006. “Publication, Publication.” Political Science and Politics, 39, Pp. 119–125.
Gary King is the Weatherhead University Professor at the Harvard University. He also serves as the Director of the Institute for Quantitative Social Science. His web page is available here.
Advantages of the method
In his paper, King told students how to write a publishable paper by beginning with the replication of a published article. Many faculty in political science and most other scientific fields followed such procedure to start with a published work. The advantages are:
By replicating an existing article, the student puts himself at the edge of the field. If he can improve 1 aspect of the research, he will make substantive difference and have a publishable paper.
By improving 1 aspect of the existing work, the student only needs to defend the one area that he has improved.
Summary of advice to students
He advised students:
One point with several supporting point is better than a lot of unrelated points.
In order to make a contribution, the point should unambiguously answer the question: Whose mind are you going to change about what?
Locate an article published in a peer-reviewed scholarly journal, preferably within the last 4–5 years.
Clarify with precision the extent of replication of the authors’ results. If replication cannot be done, it means it is not feasible to build on the work for further progress.
If 1 part of the original article is incorrect, then show the reason together with what has led the authors of the original article to think otherwise.
If the article contains no the problem, it may mean that it is not feasible to have a publishable paper. It is necessary to start with a different article.
Allocate space in the new paper in proportion to how much of a contribution it makes to change the minds of someone in the literature about something important.
After replicating the article, try to improve the presentation of the original results. Find useful, additional or even contradictory information that has not been discussed in the article.
Run some controlled methodological experiments designed to make 1 improvement possible to produce 1 interesting substantive result that is different from the original article.
Provide evidence that the model fits the data in and out of sample.
The paper should be rigorously structured and organized into sections and subsections.
Keep revising the list of sections while writing, until they convey the key points on their own without the paper.
Make reading the new paper as easy as possible.
If there is a problem that could not be solved in the new paper, clearly delineate the problem.
The title of the new paper should convey key points by summarising clearly the arguments or angles.
Include a 1-paragraph abstract not longer than 150 words containing all relevant information about the importance of the work.
In the text, identify immediately the specific empirical question discussed by the paper and get to it. “If the first bite of an apple tastes bad, you don’t keep taking bites to see whether some other part of it might be better.”
Any literature review should be short and only directed to the key points made in the new paper.
Use active voice rather than passive voice.
Spend time carefully to interpret the results in substantive terms such that a layman would understand.
Provide sufficient information about the analysis so that it is possible for others who read the new paper to replicate the analysis.
Epilogue
According to an article entitled Is there a Reproducibility Crisis? in the Nature magazine in May 2016, more than 70% of researchers have tried and failed to reproduce another scientist’s experiments, and more than half have failed to reproduce their own experiments.
Another article entitled Repeating Experiments is Not Enough in the same magazine in January 2018 argued that routine replication might actually make matters worse. Verifying results requires disparate lines of evidence — a technique called triangulation.