Contact

A Guide to Scientific Presentations

I recently tweeted a doodle depicting vastly overgeneralized characteristics of slides from talks across different scientific areas of research.

The response was overwhelming relative to my normal twitter response (which is, to be fair, almost nonexistent). While this was fun to make, what I’ve actually been thinking about for a while is:

What makes a scientific talk good?

As a postdoc in a biostatistics department at a school of public health who did her graduate work in human genetics at a school of medicine but her undergraduate work at a liberal arts school in biology, I have been witness to many different types of presentations from lots of different types of scientists. Some have been incredible! Some have been less so. I’ve tried to keep track for a while now about what exactly makes those incredible presentations incredible.

To start, there is a difference between a good talk and an interesting talk. A talk can be good without being interesting, and it can also be interesting without being good. I’d argue that the goal, however, is to be both good AND interesting.

What makes a talk good?

  1. It tells a story
  2. Your audience can understand it
  3. The audience is able to ask questions

How to make a talk good

Making your talk a story (and in most cases one single story) is important. While you have spent months (or more likely years….ok, maybe decades?) thinking about the work, your audience likely has not. Taking the time to put a story together is important for the audience to be able to comprehend what’s going on. Including all the necessary details, but not every single detail is critical. Your audience should not have to work exceedingly hard to decipher the words coming out of your mouth or the information being conveyed by your slides. To make sure your presentation is understandable, it is generally safe to assume your audience knows less than you think they do and err on the side of simplicity.

As for presentation tips, here are a number I’ve received over the years:

Slide design
  1. Chose a compelling title. The shorter the better.
  2. Use headers that make a decisive statement. (i.e. “Sex prediction is 96% accurate” rather than just “Sex prediction”)
  3. Limit the amount of text on your slides.
  4. If you absolutely need a slide of all text, use bullet points and avoid full sentences.
  5. If you use bullet points, leave an empty line in between them. (This makes it easier to read AND limits how much you may try to jam onto a single slide.)
  6. Do your very best to limit the number of figures on a slide to one (or maybe two).
  7. Make sure all axes are labeled!
  8. Make sure font size is big enough. (16pt font is the absolute smallest font size to be used).
  9. Use colors that project well. (Be mindful of color blindness.)
  10. Avoid frivolous animations. Utilitarian animations only!
  11. Aesthetics matter: consistent font, colors, and alignment get extra points! (If you’re short on time, skip this. But, maybe, you know, go ahead and avoid comic sans…)
Verbally
  1. Speak clearly.
  2. Let your audience know where you’ll be going with your talk. Make sure this is clear and simple.
  3. Introduce your axes! (Remember: you’ve seen these plots before. Your audience likely has not.)
  4. Explain what your graph shows AND what your conclusion is from these data. (Simply stating your conclusion is a disservice to the audience. Walk them through the logic. It allows them to determine if they support your conclusion from the figure.)
  5. Limit what you’re saying AND what you’re showing.
  6. Practice what you’re going to say before you get up there.
  7. Keep within your allotted time.
  8. And for the love of God, pleae do not read directly off of your slides.
  9. Did I mention you should finish on time?

What makes a talk interesting?

Good talks are enjoyable to sit through. They can be comprehended with reasonable amounts of focus and do not exhaust the audience. Interesting talks, however, are the ones you remember. They cover topics of interest to the audience in the room and are often full of either novel information, a new and interesting technique, or an interesting story.

The most interesting talks I’ve heard have:

  1. Had a compelling speaker
  2. Discussed something helpful/interesting to the audience
  3. Made sense without a ton of work as a listener

How to make a talk interesting

From my perspective, making a talk interesting is harder than making a talk good. We’re not all naturally entertaining or superb public speakers. Regardless, we can work to be better presenters.

To accomplish this, we can draw our audience in with a good title. If you’re not great at coming up with such a title (like me!), thankfully someone near you is. Ask for help! Also, we must be mindful of our audience and do our best to cater our talk to suit those in the room. With a title and idea of the audience in mind, then it’s all about preparation. It’s critical to take the time and put in the effort to make both your slides and story good. Finally, it comes down to practice. Practicing before the actual gig (and maybe multiple times if you’re a novice) AND simply presenting more often with the goal of improving in mind are immensely helpful.

Why to spend the time

  1. Your audience won’t miss out on your awesome story!
  2. You’ll get great questions and suggestions!
  3. You won’t waste the time or underutilize brain power of those in the room!
  4. You are more likely to inspire or spark ideas in others!
  5. You won’t go way over your time limit, resulting in sincere appreciation from all involved!

Assessment

To assess whether or not you’re meeting your mark, it can be helpful to watch videos from presentations you’ve made. Recordings are often taken during talks and can be watched later on. Watching oneself present is a horrible and cringeworthy process, but it really is beneficial. That said, maybe remember that you don’t have to necessarily watch the whole thing. Sometimes, the pain of watching yourself present is only bearable for ten minutes…

Bonus: if you have a friend/partner/family member who maybe isn’t in your field (or even science at all!) but who is willing to have a watch, they can offer helpful feedback!

Conclusion

While I argue it is the responsibility of the presenter to work hard so that the audience doesn’t have to, life happens. Sometimes presentations just go badly. Sometimes you simply don’t have the time to prepare as you wanted. And, sometimes, you misjudge the audience. I’ve seen it happen and it will assuredly happen to me. But, I hope that I look back on this post years from now (…when I hopefully have my own successful career…) and say that I still prioritize presenting well and have improved over the years.

Limitations

I haven’t presented nearly as much as most people in the field, so others may be more qualified to pen this post. However, I made an explicit goal when I started grad school to stop hating public speaking, and it worked! I no longer shake and quiver when I’m in the front of the room. To reach that goal, I watched a lot of presentations and took careful note of what I like and what I don’t like. My preferences may differ from yours, and if that is the case, let me know! Maybe I’ll be able to get a feel of everyone’s preferences and write a better guide later (you know, by stealing combining…all of our ideas)! Also, I’m definitely interested to see how my thoughts on this topic change over the course of my career!

I’m not the first person to write about this

Others have put some great thoughts on the topic too! Here are a few of them…

Leek Group Colors

Let me start off by acknowledging that we are by no means the first people to ask which color palettes are best for graphing. Further, I fully admit that we entered into this quickly and did not read the literature throroughly nor think through all the possibilities that one should consider when embarking upon a proper research project. Which, to be fair, is why this is a blog post (and not a paper).

Jeff and I probably aren’t even the first mentor and mentee to have the following disagreement discussion:

Shannon: Well, and this is not a sexist comment, males don’t differentiate colors as well as females.
Jeff: What? Yes, that is sexist!
Shannon: No, it’s not. It’s a thing.
Jeff: Are you sure?
Shannon: I think so…
Jeff: Let’s test it.

With that, the desires to (1) develop a “Leek Group” color palette and plotting style and (2) determine whether or not males and females discriminate colors equally well were born. And, so, we asked the internet to take a quiz.

Quiz Participants

After removing duplicate responses, keeping only the most recent response from any individual (nice try, you guys!), we had 426 responses from 198 females (46.0%), 219 males (51.8%), 1 nonbinary individual (0.2%), and 7 people who preferred not to say (1.6%).

Color Palettes

To remind those who took the quiz and catch the rest of you who didn’t up to speed, these were the color palettes being considered:

Results

Color Palette Preference

Preferences

There was no overwhelmingly popular winner; however, the “bright” palette was preferred by the largest proportion of people for graphing generally (green above) and preferred by many when specifically applied to the boxplot (purple) and network (orange) figures presented. In group meeting this past week, it was suggested that the pink be “more pink,” and I thought that was a great suggestion, so, without further adieu, the Leek Group color palette will be “bright” (with slight pink adjustments):

As for the other plotting details for Leek Group graphs, you’ll just have to wait on the edge of your seats and see what Sean comes up with!

Color Preferences by Sex

Now…onto defending my honor.

First, when asked to select palettes in which all the colors included did not appear distinct at first glance, 25% of females and only 8.7% of males responded that all of the palettes appeared reasonably distinct (p<5e-05). Yes! Point Shannon. Blues However, when presented with three images of blue dots (above) and asked the correct number of distinct colors in each image, we got the following results: Blues_mf Note: the actual number of distinct colors is 5, 7, and 7 respectively.

As evident in the graphs above, when determining the number of distinct colors, there is no large difference between the sexes. But, I will break these results down anyway. In the middle plot, there is a tendency for females to get the correct answer (7) more often than men. However, this effect is attenuated in graph to the right in which more males see the correct number of distinct colors (7) and more females see more distinct colors than are actually there (8 and 9…¯\_(ツ)_/¯?). I guess I have to concede to Jeff. Here, there was no strong effect for color discrimination between males and females. Point Jeff.

Finally, while not about color discrimination, we did see sex preferences when broken down by color palette. Specifically, females showed the strongest preference (relative to males) for “bright”, “electric”, and “google” to males. Males, on the other hand had a preference for “spring” and “summer” relative to females.

Limitations

  1. Males are disproportionately affected by colorblindess, an effect not accounted for in this analysis.
  2. Color palettes could certainly have been chosen in a more scientific manner and were by no means unbiased. (I know. I know. I am sorry your favorite color palette was not included.)
  3. I received the following feedback from a friend: “i tried to do your google form about colors and i just ended up hating myself halfway through so i stopped.” Needless to say, I have no formal training in survey development, which could have biased who responded to the quiz. In the same vein, apologies to any and all whose days were negatively impacted by this quiz! Thank you for suffering in the name of our curiosity!

A few remaining points to be made

  1. Clearly people have different preferences in how data are presented and in which palettes are most pleasing to their eyes. This is not surprising, but is certainly a conclusion supported by these data.
  2. Interestingly, when I posted the quiz to facebook, 75% of the 150 responses I received were from females. When Jeff Leek and Jeramia Ory (two male academics) tweeted this quiz out ~12 hours later, the response rate was disproportionately male. Again, while not wholly surprising, this is a discussion for another day.
  3. Thanks to everyone who was interested enough in colors or bored enough scrolling through social media to take our quiz and a special thanks to Alyssa Frazee, from whom I absolutely stole code from RSkittleBrewer to generate the color palette images on the quiz.
  4. Data and code to reproduce these analyses are on github.