By Anne M. Burton, Barton Willage, and Preeti Pushpalata Zanwar
Discussing a paper at a conference is something that many of us have seen and done but have often wondered (especially when starting out in the profession) how to do well. In the hopes of making it a little easier for those who don’t have many (or any) conference discussions under their belt, we have compiled some advice on what we think makes, and doesn’t make, a good discussion.
How long should a good discussion be for a typical conference presentation of ~ 15 minutes?
BW: Luckily, I have never had to make a decision on this. In my experience, every conference or session chair has set a time limit for both presenters and discussants. One thing that I think is critical here is to honor your time allotment. Often, discussants get something like 7 minutes per paper discussed (or 15 minutes if discussing 2 papers). It is better to go a bit under than over, in my opinion, because every minute you go over your own time limit is a minute stolen from someone else. Like all types of presenting, practicing out loud in front of a peer will help you keep on time.
PZ: It depends on how much time you have and how many papers are being discussed. I have heard some good discussions that were 30 minutes long for an hourlong webinar focusing on only one paper.
AB: Most often the session chair or conference program will explicitly say how much time you have as a discussant, and it’s usually 5-7 minutes for a 15-minute presentation. You don’t need more time than that and often you can do a good discussion in even less time.
What are crucial topics or themes to include in a discussion?
PZ: Providing the significance of the author’s work is an important theme that a discussant can cover and introduce the relevance of the work of the authors from their perspective and what’s hot in the field or for funding. In this regard, I have also made use of figures and statistics to provide background and significance. Quantifying author’s work for who, what, why also matters. Providing the historical context and or perspective outside the scope of paper but relevant to the paper. A non-linearly trained discussant or someone with interdisciplinary or multidisciplinary training would be a bonus to have and receive feedback from as they would bring some novel perspectives to the work being discussed.
AB: Really good discussions I’ve seen identify contributions of the paper that the author(s) hadn’t thought of and place the paper in the context of the existing literature. They also identify ways to make the paper better, whether that’s through reframing it, additional robustness checks, or suggestions for addressing potential issues of endogeneity. They also identify things the paper/author(s) did well or that the discussant liked, whether that’s the economics underpinning the paper, ways in which the author addresses endogeneity issues, a new or painstakingly collected dataset, etc.
BW: I always include a few things I enjoyed about the paper, and a few things I think could be strengthened. The main goals of a discussant are to help the author with feedback, and to give more context to the audience; I think including strengths and areas of improvement help achieve that goal. Within those broad categories, the focus could be on anything. Often, I have a few suggestions about empirics, such as placebo tests that could further support the identification strategy. My favorite is when I think the authors are underselling some contribution; this is rare, but I think pointing out when a paper is contributing to a literature and the authors do not mention that contribution can be very helpful. Also very helpful, is if you have a solution to a weakness in the paper. Just saying “I don’t believe this” or “the data are not good” is not helpful; but suggesting how to make the project more credible or a good source of supplementary data is helpful.
How much time should you spend summarizing the paper that was just presented?
AB: None. I find it helpful for a discussant to summarize or restate/reframe the contribution but it’s the author’s job to provide an overview of the paper, not the discussant’s. You don’t need to fill the time allotted; it’s always fine to give a shorter discussion, leaving more time for audience questions.
BW: Assuming this is a standard conference, with limited time for discussants, I’d say 0 minutes. In every instance when I’ve been a discussant, the author of the paper had just presented for approximately 15 minutes. I do not see any reason for the discussant to spend time giving an abbreviated re-cap of an already short presentation. One of my least favorite types of discussion is when the discussant spends their entire time allotment re-presenting the paper. There are some exceptions, such as the very rare situation where the discussant presents and discusses the paper, and the author only responds.
Any advice about how to approach reading the paper and preparing a discussion?
BW: My approach to reading the paper is the same as when I referee. I do a thorough read, making notes throughout. I tend to underline or circle more important thoughts, and I make smaller notes about little things like awkward phrasing. However, I do not write a full referee report. I do make slides, both to help me when I’m talking and so the authors have a mini-report on what I thought were the main issues.
PZ: Focus on research questions and robustness of methods used to answer the question and future steps or directions.
AB: I will pull up the paper in one window and have a word document open in another window that I use to write down thoughts, questions, or comments I have as I am reading the paper. Sometimes I think of a suggestion early on (e.g., a particular robustness check as I’m reading the introduction) that the author(s) include later in the paper. If that happens, I either cross it out or change the suggestion to “mention this robustness check earlier in the paper”. Many of the things I write down will not go in my discussion, but I will send to the author(s) separately, for example, comments about the organization of the paper. Once I’ve read the paper I will go through my list and reorganize it to prioritize the most important comments once I’ve seen the universe of comments I have. I’ll make a separate section for comments that are worth emailing the author(s) but don’t belong in a discussion.
What are things that don’t belong in a discussion?
AB: A summary of the paper (assuming an author is presenting the paper). Pointing out spelling/grammatical errors or typos in equations, comments about the organization or layout of the paper (fine to email that to the author(s) but if it’s a comment specific to the paper and not the presentation, it adds no value to tell the audience). Also, being mean (especially punching down). Every paper has flaws, and works in progress may have big flaws, but you can always point them out nicely and in a constructive way.
BW: Nitty gritty minutiae, suggestions about writing, and anything that would embarrass the author (especially if they are a graduate student or very junior). However, I think a lot of this type of feedback is valuable to the author, so I generally send them more detailed notes or a marked-up version of the paper. I just don’t think giving this kind of feedback in front of a lot of other people is the optimal format.
What is the value-added of a good discussant (what can a discussant provide that an audience Q&A can’t)?
PZ: Discussants are generally knowledgeable about the topic and current with the literature. Sometimes they speak about a sequence of papers in a session and how a particular paper fits within a session.
BW: I think a good discussant can be invaluable. My first conference presentation was at ASHEcon in 2017, and I had an amazing discussant. He pointed out a lot of ways I could make my paper better, but he was also very excited and kind in how he gave that feedback. This is such a high value added, especially early in one’s career. Additionally, the audience is only seeing a very short presentation, and junior people need feedback from people who actually read the paper to get projects in good enough shape to submit to journals, which the audience generally cannot provide. However, I’m on the fence about whether the “discussant” part of this process is necessary. Do 5-50 other people need to listen to the conversation? And does the audience care much about what the discussant is saying? I think we might be able to achieve providing feedback to junior authors more efficiently if each presenter in a session read someone else’s paper and gave comments privately to the author.
AB: As (typically) the only person in the room who has read the paper besides the author(s), the discussant is uniquely positioned to pick up on nuances of the paper that the audience often can’t see with a 15-minute presentation. The discussant has also had more time to think about the paper and can, for example, identify a new contribution of the paper the author(s) hadn’t thought about or identify solutions to limitations or weaknesses of the paper.
Does a discussion need slides?
BW: A discussion doesn’t need slides, and I’ve seen successful discussions without them. But, I would recommend having slides at least the first few times you discuss. You can have them and not use them; however, if you don’t have them and need them, you’re out of luck.
AB: No, provided you know what you are going to say and don’t need slides to remember or keep on track. If slides help you organize your thoughts or give you more confidence when discussing, by all means go for it, but don’t use them as a crutch and just read off the slides.
PZ: Not necessarily. I have seen some insightful and deep discussions and comments without any slides. I have shared my discussant slides with the author and session chair. For the author, it is good to receive written notes/points even if discussant provides comments orally. This helps with reflection and incorporating critical comments during paper revision given revision process can span at times months and/or years.
Any other thoughts about being a good discussant?
BW: Be kind. This is good advice for all realms of professional and personal life, but this is one situation where you are giving feedback to someone publicly. Think about how you would like to receive feedback and be 10% kinder. It’s very unlikely people will remember if you were an awesome or a horrible discussant, unless you were also rude to the author. And generally, people will not be impressed if you “gotcha’ed” someone doing their best.
PZ: Since discussants are reviewing a paper on a volunteer basis it is a good idea to thank and appreciate their time and their insights. Discussing is also an art and science: how to deliver critical yet supportive feedback.
AB: At Southerns a few years ago a session organizer for a session I was in had a clever innovation for discussions that I liked. Rather than have each discussant go up to the podium and present with slides, she told each discussant to come up with 2 or 3 big-picture comments about the paper. Each discussant spoke for 1 to 2 minutes, leaving more time for both the authors to present their papers and the audience to ask questions. It also helped keep the discussions focused on higher-level comments that were useful for the authors and audience to hear, as opposed to nitty-gritty details about the paper. Also, echoing what Barton said: be kind. Being mean doesn’t make you look smart and it reflects poorly on you. The marginal cost of being kind is almost 0.