Jeromy Anglim's Blog: Psychology and Statistics


Thursday, July 7, 2016

Workflow for Completing a Revise and Resubmit of a Journal Article in Psychology

This post discusses my workflow for completing a revise and resubmit.
I have a template document for representing revise and resubmit responses.
See my templates page on github and specifically see the file "response-to-reviewers.dotx".

Setting up the Response Document

The document has the following core styles:
  • Heading 1: Divides up major sections of the review (e.g., Editor, Reviewer 1, Reviewer 2
  • Heading 2: Summary statement for each reviewer actions
  • Reviewer Comment: Exact quote of a particular reviewer comment
  • Body text: For recording my response
  • Quote: For formatting quotes of specifically modified sections of the text
Step 1 is to paste the full text of the editor and reviewers into a new Word response document. Apply reviewer comment style

Step 2 is to set up the response document. Level 1 headings are added that divide up the the reviewer sections.

Reviewer comments are divided into discrete points. The division of revision points may or may not be clear. Some reviewers provide numbered points. Others provide a more narrative review where each paragraph includes multiple points. Some points are interconnected but involve distinct actions.
For each point that is identified, I add a level 2 heading. The level 2 heading includes an identifier and a brief summary statement of the requirement. Identifiers are for example, "R1.2", which would refer to reviewer 1's second point. In some cases, where there are connected points, you get "R1.2.1", "R1.2.2" and so on.

There are several benefits to using identifiers. In some cases, multiple reviewers make the same point. Thus, you can quickly refer the reviewer to another review point. E.g., "This point was addressed in reviewer point R1.2". It can also be an efficient way of keeping track of reviewer points when you are working through a large number of them.

The summary statements are important. I aim to keep them short. Ideally they'll fit on one line so that they are easy to quickly understand (i.e., around 50 characters). I try to make them commands. For example:
  • Clarify unique contribution
  • Improve study motivation in introduction
  • Describe x more clearly
  • Add references to ...
  • Justify statistical method
  • Consider using ... method
  • Include ... Table 1
In some cases, the required action is not explicitly stated by the reviewer. For example, if a reviewer critiques a methodological decision, there are various possible actions including, justifying your choice, adding a limitation, and so on.

Benefits of the above approach
  • Using formal Headings in MS Word allows you to view a document map on the side that can quickly allow you to navigate between reviewer points.
  • Another benefit of the above process is that reviewer comments start to appear more manageable. When you first receive a few pages of reviewer comments, it can feel overwhelming. The above process begins to divide up each point into a more manageable task. The act of providing a summary statement also forces you to read and understand what action is required to respond to the reviewer comment.

Record initial reflections

Above, I show how the first reading is used to parse reviewer comments into discrete points and give descriptive titles. In the second reading, I add comments to each reviewer point using the comment feature in the Word Processor. This is an opportunity to have some initial reflections on (a) how easy it will be to satisfy the revision, (b) whether a change to the manuscript is required, and (c) what should be done. After I've added these, I often circulate the response document to collaborators, to allow them to add comments.

Sequencing the Revisions

The next task is to determine a sequence for working through the revisions. This involves keeping track of which points still need to be addressed and deciding on an order to work through the points.
At a basic level, I place an asterisk at the start of each heading that has not yet been addressed. This is removed once the point has been adequately addressed.

A more challenging issue is deciding on how to work through the changes. Some changes are interdependent. However, major revisions often have to be worked through first as they can have broader structural implications for the manuscript.
A few useful steps for thinking about sequencing include:
  • Organise the points into categories
  • Read through each point, and make some tentative notes about what to do (e.g., using comments in Word).
  • Decide on an explicit sequence to work on the points. This often requires you to brainstorm the pros and cons of working on one point versus another first.
In some cases, sequencing will raise some more meta-issues about the paper that transcend any given review point.  I mostly find it easiest to work through points in the following order: analyses, results, method, introduction, discussion. The rationale is that any new analyses that you run and incorporate into your paper will change your results. And these may further require changes to the method, which in turn influence the framing and discussion. Likewise, if the introduction is changed, this may have implications for how the discussion integrates topics raised in the introduction.

Logistically, I generate a table of contents in MS Word. This lists all the reviewer point titles (i.e., the IDs and the titles such as "R1.1 Update method to include ..."). This works because all the reviewer points are formatted using heading styles. I then copy and paste this as plain text into a work document. These points are then organised thematically under headings and into an appropriate sequential order.

Addressing Revision Points

If sequencing issues have been resolved, it is a matter of working through each revision point. I have a few guiding principles:
  • Write in a manner which focuses on the scientific issue.
  • Treat the reviewer with respect.
  • If a reviewer has misunderstood something in the manuscript, take responsibility for making the manuscript clearer.
Another point is that the response document should be self-contained. Ideally, the reviewer should not need to look at the actual manuscript to judge whether you have effectively responded to their requested changes. This makes the experience of the reviewer much more pleasant. From a strategic perspective, they may also be less inclined to read through the entire manuscript again and come up with all new concerns.
  • If a table or figure is updated, then paste a screenshot of the updated table or figure.
  • If a new paragraph has been added, include a copy of that paragraph.
  • If a sentence or two has been added to a paragraph, include a copy of the whole paragraph and bold the section that has been added.
  • Only if the point is very basic is it sufficient to say, "this change was made". Examples of this might be adding a reference, fixing up typos, and so on.
Another useful strategy is to indicate new text in the manuscript with a different colour font (e.g., purple).

Collaborations and Revisions

It is often easiest if one person leads revisions. The lead person can also allocate specific revision tasks to co-authors. There is the issue of how to synchronise the revisions in the manuscript with the response document. If the changes are particularly complex or the collaborators are likely to make substantial additional changes to the manuscript, then it may be worth waiting a little bit before completing the response document. Or alternatively just see the response document as an initial draft to be returned once the manuscript has been finalised.

Track Changes

Some journals require that you include a version of the manuscript with track changes. In other cases, it can just be a useful addition to the submission. If you are using MS Word, then the compare documents feature is ideal for generating this document. This feature allows you to anonymise the change because you can label the change with "author" rather than your actual name.