I’ve chaired my share of panels, but there’s one thing I haven’t seen a consensus on: should questions and comments be taken after each paper, or should they be held until the end?
For hypothetical sake, say we’re talking about a 75-minute session with three 20-minute papers. They’re in somewhat similar areas, but it’s not a panel organized to address a specific topic.
I find it’s better to hold questions for the end, which sometimes promotes discussions among the panelists as well as with the audience. Also the panelists prefer it, since they’re eager to deliver their papers, and their papers sometimes run over so this way you know exactly how much time you have for q and a.
I’m not an academic, but I’ve chaired tons of sessions and real panels,. and have been program chair for several conferences.
What you are calling a panel we call a paper session - 3 25 minute talks in a 90 minute slot, with 5 minutes each for introductions and questions. What we call panels have a moderator and 4 - 5 people, with short position statements from each panelist, and then a long session of questions from the moderator and audience, and discussion among the panelists.
Questions after papers is definitely the way to go, not holding them for the end. First, people move between sessions (we have 4 - 5 parallel sessions) so some people would not have a chance to ask their question. Second, people might forget their question, or a question for the first speaker might get buried by questions for the 3rd. Third, sometimes a question can be answered by going back in the Powerpoint presentation, much harder to do at the very end.
There was one workshop where each session had 3 speakers and two discussants, whose job it was ask the initial questions. But attendance was limited to 50 people, foils, not Powerpoint was used, and there was lots of interaction during the talk, which doesn’t work for larger conferences.
Since we have parallel sessions, session chairs are strictly instructed to not let papers or q and a run over. The program committee member in charge of the session is there in the back of the room to help. After 40 years, we are very good at it. The bigger problem are papers that are too short and which don’t inspire discussion. Starting the next paper early is as bad as starting it late, so there could be dead time, and we don’t recruit session chairs for their tap dancing ability.
I am in no way an academic but my future MIL is and this is a topic she speaks frequently on, so I wanted to relay her opinion.
She feels brief questions after each paper with a bigger discussion/summary/ at the end is ideal. Like Voyager said, several people move between papers so it’s ideal to get the questions in before the next paper is read. If left to the end, forgetting questions and intermixing questions from speaker 1-3 makes it hard for the group to process the answers, in addition to the several people who have shifted out of the room.
In my opinion, questions after each paper are better if and only ifthe chair is strong enough to hold people to their times, presenters and audience. I don’t know why, but people seem to have a really hard time doing this, and if you can’t keep it moving it’s really unfair to the second panelist and following. I think it also depends on how cohesive the panel is.
(Edited to remove first sentence, which wasn’t useful.)
I also think after each paper is best, with a strict eye being given to the allotted time for q & a.
I prefer questions after each speaker, personally. I’ve been a situation with both, and in the case of all the questions at the end, I think it unfairly benefits the last speaker.
Given that everyone in the room is (supposedly) a knowledgable expert in your generalized field, but is still unfamiliar with your work, I think it’s unfair to expect them to remember precise details and questions while simultaneously trying to absorb new material.
The way we control this is to have microphones set up in the aisles, and have one or two lines of people who wish to ask questions after the paper. This lets the session chair know how many people wish to ask, and the questioners also. They guy who has five questions is a lot more willing to ask only one if five people are behind him - on the other hand, if no one is, five questions might be okay. It is much easier to manage than people popping up in the audience with questions. I ran one internal panel where I used a traveling mike and went to audience members with questions. That was very easy to control.
The other thing the chair has to control is nastiness. We have a rule that it is out of bounds to beat up on a grad student - beating up on a professor, though, is just fine. 
I say hold them to the end so the last person to speak doesn’t get shafted on time, but as far as I know that’s always standard in my field; it probably depends a bit on what the audience and panelists are used to.
Actually, this is a reason to schedule them after papers. If someone runs over, it is fairly simple to not have q &a for that paper at all, which gives a buffer. People who stick to their time get feedback, people who don’t don’t.
Really? Wish you’d’ve chaired my panels. At a large humanities conference that will remain nameless, this crusty old Russian lady tore me, a then third year undergrad student, to absolute ribbons, although my supervisor lauded me for responding to her assault brilliantly. 
Thanks for all your responses. In my experience, I’ve seen questions held to the end about 2/3 of the time and that is what I usually do. However, I do feel this may lead to less flamboyant papers getting the shaft, and this is no fate for solid work.
Still thinking about it…
Was your supervisor there? Maybe you were doing so well that he didn’t think he had to intervene. I have seen supervisors spring to the defense when the chair isn’t enforcing the rule, though.
My first paper was presented at a workshop. It was just after my advisor died, so I had no supervision and was pretty clueless. The paper was really bad, but the attendees were nicer than I deserved. It made a big difference. I got to know everyone in that field, and for the most part they were pretty nice.