Gary King’s article “Publication, Publication” offers great advice for starting researchers in his political-science courses as well as beyond. This Web page contains some commentary on King’s article for CS530 students.

Elements of the Paper

On data: Unfortunately there is no field-wide data repository in AI like ICPSR’s Publication Related Archive for political science, but repositories of often-used data do exist, such as the UCI Machine Learning Repository and the Linguistic Data Consortium.

On venues: Computer science is a peculiar field, in which some conferences “count more” than some journals (whatever that means). Use your judgment and ask for mine.

On work beyond replication: Only replication is required for your final project. If you have the time and energy, you can go beyond replication after you have completed replication.

(Update 11/20) King writes that you should not “allocate space in your paper in proportion to how much work you put in accomplishing each task”. That is still true, but if your work is purely to replicate, then you may not have anything to report in terms of “what you contribute to our collective knowledge about the world”. Instead, you should report what the work you replicate contributes, whether or not it has been reported by the original authors. Don’t forget that your paper cannot assume that the reader has read the original article or any background literature about your particular problem!

Ground Rules

On collaboration: King states several reasons he requires his students to coauthor their papers, which are also my reasons for preferring that you coauthor your papers.

The “formal way” described under item 5 in this section (and item 5c in the following section) does not apply in our course, but you should still treat your fellow students, your project proposal, and the course mailing list as informal ways to get feedback.

Style

Recall that your paper should have a point. Make that point as early as possible, such as in the title or the abstract, and stick to your point throughout the paper. See:

• A scrutiny of the abstract,
• How to write more clearly, think more clearly, and learn complex material more easily
• On organizing a paper,
• Good writing.

Item 10b makes a good point—that you should talk about what you are replicating rather than who did it—but then commits what in the opinion of this writer is a stylistic error. In a citation such as “Jones and Smith (2003)” (or “Jones and Smith [7]”), the parentheses indicate that what they enclose is parenthetical, so the text should remain grammatical and sensible without the parenthetical remark. It is ungrammatical to write “Jones and Smith (2003) is mistaken” for the same reason it is ungrammatical to write “Jones and Smith is mistaken”: the subject fails to agree with the verb in number. To state that Jones and Smith’s work is mistaken, write “Jones and Smith’s work (2003) is mistaken”. It is also ungrammatical to write “Our work replicates that of (Jones and Smith 2003)”, for the same reason it is ungrammatical to write “Our work replicates that of”.

When in doubt, be consistent. A good way to be consistent is to follow a style manual, such as the Chicago Manual of Style.