I edit many papers after they’ve been rejected or sent back to the authors for major revisions. Often, I see the peer-review comments as well, and I see many mistakes come up again and again. Rejection is an inevitable part of academic publishing, but it is possible to minimize it somewhat. To give your paper the best possible chance of publication, I recommend the following.
To avoid being rejected by the editor during an initial screening:
- Choose the journal best matched in scope. Find some candidate journals from their “Aims and Scope” description. You can then use Google Scholar’s advanced search to look for recent papers published in these journals that also include keywords in your paper. This is an important step because the Aims and Scope descriptions of many journals are vague and out of date. Google Scholar may tell quite a different story.
- Choose a journal with a reasonable impact factor that is likely to publish the level of work you are doing. Again, look at typical published papers. If they are much better than yours (e.g., many experiments comparing the proposed method with a large number of state-of-the-art methods on massive datasets whereas you only test two performance metrics and compare your method with the baseline method on a tiny dataset), improve your evaluation or choose a lower-impact journal. Methods that only solve a small, very specific problem will also best fit in a specialized, lower-impact journal.
- Make sure your language is free of grammatical errors and typos. Editors and peer reviewers are often people that work hard on their English, especially if it is their second language. They have little patience with easily avoidable mistakes.
Peer reviewers are looking for two things: methods that work and methods that are novel. To avoid being rejected by peer reviewers, I recommend the following basic steps:
- Make sure your novelty is clearly outlined in the abstract. Outline it again in the Introduction. Then, make sure your “Related Work” section supports the novelty you claim by showing that no one else has done quite what you are doing. Summarize the novelty again in the Conclusion.
- After compiling your Related Work, go through recent papers, and see how they evaluated their method. For instance, in deep learning, the batch number used to train the neural network is reported. If you don’t report this, peer reviewers will want to know what it is. They will also want to know why you missed such a simple thing. In addition, metrics change over time. BLEU is often used to evaluate natural language processing systems, but there have been other recently proposed metrics that may be better. Also make sure you’re using a large, popular dataset for testing instead of a few ancient test images like Peppers.
- Again, make sure the language is clear and easy to read. If peer reviewers have to work to understand your work, they will get irritated and think you are not a careful researcher. Other mistakes that they find in your research (which they might politely ask you to correct in a revision of a well-written paper) will become mistakes that they use to reject your paper outright if your paper is also poorly written.
I hope these tips are helpful, and I wish you the best of luck with your research, whatever journal you choose for submission.
Header photo by Steve Johnson on Unsplash
Moravec Science Editing 