How to Make Your JSM Talk Great

Dick-DeVeaux
Richard (Dick) De Veaux is C. Carlisle and Margaret Tippet Professor of Statistics at Williams College. He holds degrees in civil engineering (BSE, Princeton), mathematics (AB, Princeton), dance education (MA, Stanford), and statistics (PhD, Stanford). While at Stanford, he studied statistics with Persi Diaconis and dance with Inga Weiss.

At many conferences run by companies, there are several keynote talks, some by technical experts and some by motivational speakers. In 2006, I was invited to give a keynote at a conference in Barcelona. My topic was data mining, and my slot was right after David Brashears. Who? Well, David Brashears is the guy who hauled an IMAX camera up to the top of Mt. Everest and filmed one of the most amazing and breathtakingly beautiful films ever made. Now, I know data mining is fascinating stuff, but Mt. Everest? I knew I was in trouble. To make it worse, David was a certified hero. He spent two days going up and down the mountain saving lives (read Into Thin Air). Okay, so I knew I was in trouble.

But as David started talking, I felt a little relieved. I mean, not that I wanted him to go down in flames (well, okay, maybe a little), but I was expecting him to be as dynamic as his movie is. But, for a keynote talk, he really started out low key (no pun intended). His voice was quiet, almost monotone. There was no sense of passion or intensity that is often the hallmark of good talks. If you’ve ever seen Hans Rosling talk about statistics, you know what I mean. (If you haven’t, go to YouTube right now). But, as I learned that day, there are many styles that can lead to a great talk. David had a story to tell. And tell it he did. He calmly told us the challenges of climbing the highest mountain on the planet with enough 70mm film to record hours of IMAX quality footage. It was a riveting, fascinating talk. David had us in the palm of his hand. As W.C. Fields once said, never share the stage with children or animals. He should have added David Brashears.

The point of this anecdote is that there is no single way to give a great talk. You don’t need to have a lot of flash, jokes, or anything else. What you need is the desire to communicate your story to an audience. Of course, saving people’s lives at 30,000 feet is pretty good material. And even your best statistics achievement might not measure up to that, but there are plenty of ways to communicate your passion and story effectively. Everyone, and I mean everyone, can give a great talk.

And every talk can be made better.

One of most important points to remember when giving a talk is that, in fact, you are talking to people. This sounds obvious, but I see far too many talks where it seems the speaker has forgotten that point. Instead, he avoids all eye contact, talking to his shoes or maybe the person at the very right of Row 1. He also avoids any voice or facial expressions and tries to sound like a journal article.

The audience is there because they want to hear what you have to say. Otherwise, they wouldn’t have bothered to come.

To make your talk a success, first ask yourself who your audience is, what they might know, and—just as important—what they might not know. Most talks at JSM lose people too fast. Most of your talk at a general meeting like JSM should be accessible to anyone with a background in statistics, with only a small part for those who specialize in your particular field. Of course, there are always exceptions, but making the majority of the talk interesting and accessible to more people always seems like a good idea.

Yes, we all want to show how much work we put into our research, but if it took you two years to understand your research area and another three to develop your results, how quickly do you think the audience will be able to understand the details? Fifteen minutes? Is everyone else in the world really that much faster than you are? Respect for the audience is key, and an important aspect of that respect is to have them understand what your story is about. So, don’t put all the technical details (as important as they are) into the talk. What you can hope is that they become interested enough in your subject to want to read the paper, so point them to where they can do that. But don’t make the mistake of overestimating the amount of new material people can absorb in a short time.

Scott Berry, William Li, Chris Natchsheim, and I presented a workshop on giving presentations before JSM, and much of this material comes from that. I also should mention a colleague of mine (not a statistician) from Grand Valley State University, George Nezlek, who has taught me much about the art of presentations.

Five Myths of Giving a Great Talk

There is only one way to give a great talk. As I hope my opening story pointed out, this just isn’t true. There are many styles. The one style that won’t work for you is someone else’s. Be yourself. When you tell a story to your friends, I doubt you try to do it in anyone else’s style. The same should be true of your presentation. Being yourself and being honest is a great way to start the communication between you and the audience.

However, there are some elements that are essential to a good talk. The talk should be structured so it makes clear what you are going to talk about, proceeds to tell your story in a logical order, and then wraps up by reminding the audience where you’ve taken them and what’s left to do. Using examples and specifics is a great way to get your point across. Staying completely abstract will lose most of the audience. Speaking of the audience, don’t forget about them. Look at them! Make eye contact. The more you can connect with the audience, the more they will want to hear whatever it is you have to say. And it’s okay to be nervous. Be honest. Be yourself. There are people out there who want to hear you tell a great story.

There is no place for humor in a serious scholarly talk. Humor is a device, not only to make your audience (and you) relax, but also to help your audience learn. Research has shown the effectiveness of humor in acquiring new knowledge. How to do it? Unless you are very good at telling jokes, don’t do it. Jokes, per se, are dangerous territory. And jokes don’t ‘travel’ well. Be especially careful of humor that depends on a certain culture or knowledge.

Instead, humor should be natural. One sure-fire type of humor is to be self-deprecating. No one minds if you make fun of yourself! And we all have plenty of material to use (trust me).

I’ve been working on this research so long, I’m ready to give the talk. You’ve spent months (years?) working on your research. How much time did you spend preparing the talk? To honor your effort and the audience, there should be a balance between the two. It takes time to structure a great talk.

Following are some technical points about slides to keep in mind:

  • Use a large font and keep to only a few major points per slide.
  • Don’t use complete sentences on your slides! If you have them, your audience will be reading them instead of listening to you. Just have key points that remind you of the story structure and some key equations. Never, ever read your slides to the audience!
  • Remember that equations are a compact form of expression, so go easy on them. They are packed with information. If possible, explain the concepts around them instead of relying on them. Too many talks pack slides with equations thinking that people can read them in real time. Sadly, there are only a very few of us who can do that.
  • We’ve all seen slides in 6-point font, either with 30 bullet items on it or—worse—more Greek than you’d see in an entire Annals article. Avoid both to the extent you can.
  • Use graphics instead of tables. We all teach our students the importance of using graphical displays to communicate, but how many talks have you seen that give tables of numbers instead of well-constructed graphics? Be as visual as you can, and don’t rely on someone in the back row having eagle eyesight.

Good preparation is essential. You don’t necessarily have to practice giving the talk aloud (although it never hurts), but you should know the flow of the talk inside and out and practice to make sure you can comfortably cover the material in the time allowed. How many talks have you seen where, two minutes before the allotted time is up, the presenter hasn’t finished more than a third of the talk? Racing through the next 60 slides doesn’t do anyone any good! Make sure you can finish the talk comfortably in the time allowed. Having a few minutes for questions at the end is a wonderful outcome.

I need to show everything I’ve done to impress the audience. This is probably the mistake I see most often at conferences, especially at JSM. Fifteen minutes (or five for a speed session) is a short amount of time. You can’t tell the ‘whole’ story. So, think hard about the important points you want to convey. Those probably aren’t the technical details. Give an outline and the main points of how you used Slutsky’s Theorem to show your main theoretical result, but don’t try to wade through all the details unless you have several hours (unlikely at JSM unless you’re giving a series of Medallion lectures). If we want to follow up after seeing your talk, we can read the paper or contact you.

I can’t do this. You want to give a great talk and everyone wants to see the speaker do well. So breathe. Be yourself, look at the audience, and tell your story. You can give a fantastic talk.

We all get nervous in some situations. Tandy Beal, a masterful performer with whom I had the privilege of touring when I danced professionally, taught me two important things about the audience. First, she said, imagine that someone who loves you very much is in the audience (she always visualized Row 8) and wants you to give a great talk. And, in any case, “Remember, 11 o’clock p.m. always comes.”

Have a great talk. See you in Boston!