Office Hours

How many of you have been on the job hunt and saw, in the ‘required skills’ section, ‘strong Communication skills’? If not that specifically, there was sure to be something like ‘interpersonal skills’ or ‘presentation skills’. These are so commonly required in every industry that it may be better to ask, have you ever seen a job posting without one of the above?
The good news is that, as with any other skill, we can get better through practice and an understanding of the essentials. In this article, we are going to focus on applying some evidence based conclusions from the cognitive and education sciences in improving the way that we communicate our results as data scientists and analysts, though the points will be generalizable to any application in which you find yourself public speaking. If you’d like to check out the original sources, you can read through these academic papers.
Back to data scientists; let’s face it, those of us who spend their time buried in the numbers are not always the best at explaining them in an compelling and easy to understand way. But this is kind of silly, right? No matter how good our ability to discover patterns and make inferences, if we cannot convince our team members of those patterns, our results will have no impact. As someone who can relate to the challenge of communicating results and who needed some explicitation of the skills implicit in good presentation, I found the following techniques helpful. As you go through the list, I encourage you to reflect on how a recent presentation could have been improved with their application. So, without further ado:
1. Elaborative Investigation
Elaborative investigation "…prompts learners to generate an explanation [for] an explicitly stated fact." By asking questions, we engage our audience socially (more on this later) and encourage listeners to process the information being presented, increasing attention. As data scientists, we can employ elaborative investigation by asking team members to recall a relevant conclusion from a previous slide or meeting, or to draw on some domain knowledge that everyone is likely to have. An important distinction between elaborative investigation and other methods of engagement by questioning is that the listener is able to answer the question; with this technique, we are intentionally getting listeners to bring relevant information into their headspace.
2. Practice Testing
The header for this section is not entirely representative; the terminology we may prefer to use in our head going forward could be rhetorical or leading questioning. We do not want to be testing our listeners’ knowledge because of how that can make them feel, especially in the case of a question where we, the presenter, are going to be the only one that knows the answer, which is usually the case when presenting results. The header, however, may have had a similar effect on the reader, one which leads to this technique being beneficial:

Think back to test announcements in school, or even worse, the dreaded Pop Quiz. They probably resulted in you feeling a lot like Pikachu, surprised and wishing you knew more. Stress while learning markedly improves memory formation, however, the opposite is true for memory recollection. Thus, we aren’t interested in testing our colleagues; in fact, when asking questions during a presentation we often aren’t expecting them to answer at all. As data scientists, we are going to take advantage of this stress response and the resultant improvement in memory recall and brain engagement by asking rhetorical or leading questions. An effective implementation can be to present preliminary data and simply ask something along the lines of, "Now, what can we infer from what we’ve seen so far?" It is important to follow your question with a pause that is brief but long enough to allow listeners to consider interpretations and trends of their own, or at least long enough for them to think, "Well, I’m not sure." Then, move to the next slide and hit them with your hard-earned insight, which they will be more likely to remember and comprehend after their brain has been primed for its reception.
3. Mental Imagery
Continuing along with our methods of mental engagement, we can use imagery to improve the ability of our listeners to process new information. After all, our eyes are our primary source of sensory information. Besides the beautiful visualizations that you create for elucidation, for which there are many good resources on producing, it can be helpful to ask listeners to be imaginative in relaying a conclusion. Humans process images 60,000 times faster than text, and our brains are hardwired to process visual data. For example, pretend you’re sitting in a room with a group of realtors, and explaining that, with everything else about the house kept the same, our model suggests an additional bedroom makes the selling price of a house go down. This might feel counterintuitive (the realtors might also look at you like Pikachu), but if we ask our realtor friends to imagine themselves standing in the house they most recently sold, then suggest they further imagine that, without increases in total square footage, number of bathrooms, etc., to add another bedroom… Well, where does it fit? To make another livable bedroom we have to repurpose space from existing rooms making them more cramped, more people end up sharing the same number of bathrooms, and so on. By creating mental images we can convey results in a relatable, easy to understand way, even those that might not make sense at first.

4. Repetition and Highlighting
While educational studies suggest neither are actually very good learning techniques, regardless of their popularity, we can get value out of the reason why they suck. Researchers believe that highlighting is especially ineffective because you have to already know what information is valuable to highlight the correct bits. Luckily for us, we do know what information is important and can do the highlighting for our audience. This can be achieved through making important features of visualizations stand out and by limiting our slides’ text to a few bullet points that capture the main ideas succinctly, though, if you’re like me, the actual discussion will be much more long-winded. Finally, while we will get to effective summaries below, we can bring repetition in as a highlighting analogue by shortly collecting our thoughts at the end of each slide and conducting mini-summaries by stating, "Moving on, but the main takeaway from this slide is _____ and you’ll see more on why this is important later on."
5. Social and Agentic Inclusion
Listeners who come away feeling positively about their team environment are more likely to attend to our results because they are more likely to devote their attention to us and feel confident in us as co-workers. Social inclusion can include referring to individuals or teams by name, and by being present before and after the talk to make conversation or answer questions (We all know we need to show up early to make sure our Presentations and demos are prepared and working properly anyways!). We can also drop passing comments to coworkers if we happen to see them earlier on in the day, such as, "You’ll be excited to see some of the results in my presentation this afternoon!" Agentic inclusion, with ‘agentic’ referring to one’s ability to exert influence over their environment, inspires listeners to feel their presence in the results or feel part of directing the flow of the presentation. Be sure to acknowledge the efforts of other teams and individuals because hey, we’re all in it together; creating data insight is us just doing our part. Allow the audience to interrupt with questions while being careful to be respectful of time constraints and without being dismissive or allowing derailment from the story we’ve carefully crafted. It is perfectly acceptable to suggest further conversation later in order to make sure the group stays on track. It always feels great to evoke a question where the answer is, "Thanks Lisa, I’m glad you asked! Seeing this pattern, I was wondering about that too…" before clicking to the next slide with beautiful visualizations and talking points prepared for precisely that curiosity. Even if they don’t ask the question, it is this data-driven story with a logical flow that we’re all truly after.
6. Be Prepared for Different Cognitive Styles
We need to know our audience. Are we communicating to other members of a data team? Friends in the software department, with good coding backgrounds but not necessarily the data understanding? Or the people in the sales department, whose customer-facing expertise we rely on, or perhaps an interviewer to whom you are explaining a project, who may not have a technical background at all? Is it a mixed group? Make sure to prepare presentations with a target audience in mind. Additionally, within each group, recognize that different cognitive styles exist; some people learn better visually while others aurally. There are many reasons, even amongst similar individuals, why one may be challenged in their comprehension. Let’s do our best to be prepared for answering questions by keeping track of the difficulties we faced along the way, but also by considering how information we take for granted because of our unique background and experience can be confusing for others. Lastly, if presenting to the same group frequently, pay attention to the types of question they ask and the visualizations that prove useful. We can do a lot to improve our results communication by recognizing that repeated questions or multiple questions of the same type often mean that we are failing to include necessary information in our report in some way.
7. Make Your Results Memorable
To make the impact of presentations long-lasting and help increase our visibility, as well as cause coworkers to look forward to our results in the future, we should frame results within the context of our systemic work environment. We’ve just spent time detailing a specific result; let’s not forget to remind everyone why we were interested in this result in the first place. If the results create further questions, explain future paths of exploration and request support from other individuals and teams if necessary. If presenting a solution to a problem, impart the impact on day to day operation procedures. As data scientists, however, we may not be responsible for decisions regarding direction, but we can use this to our advantage in increasing our presentations pervasiveness. We can ask questions about how the results may affect our team and those adjacent, or create a discussion over what the next investigation should concern, in addition to any other questions that cause listeners to walk away wondering broadly about the next step through the lens of our conclusions. It can be good to end a meeting with, "Well, let’s all think about it over the next few days, and come back to it in our next meeting." Occupying the attention of colleagues outside of the presentation space and clarifying results’ impact makes our presentations better by increasing others perception of their value and giving them a reason to be curious about the next installment, increasing their desire to pay attention.
8. Summarization
After a thorough presentation, listeners have been presented with lots of new information. It is important for recollection and lasting understanding to identify the connections between ideas and which ideas and results are most important. Spend time on the summary, and do not read a list of bulleted points off a final slide. While it is good to have those points visually represented, our verbal delivery should connect the dots in a cohesive manner, summarizing the story from beginning to end.

Here, I’ll give writing an effective summary my best shot. While we have covered a lot of techniques, all of them work by giving cause for our audience to pay attention and by engaging listeners in processing the information being presented. We engage them through asking questions, encouraging reflection on the old and priming them for the new. We engage them through inclusion; socially, through creating a positive work environment, and agentically, by answering questions clearly with imagery and based on the individual and recognizing others contributions. Finally, we engage them by making sure to highlight important information through effective use of visualizations and the creation of ‘dots’ as mini-summaries in the form of repeated information, before ‘connecting the dots’ at the end in a final, cohesive picture. In this way our audience will walk away feeling the value of our results and its place in the context of our workplace as a whole.
If you’ve made it this far, thank you! Now that you’ve read the article and understand the techniques… which of them did I use in my writing? Did I leave any out, and how could my presentation have been better? Is there anything that I did that was effective but didn’t make the list? Do you have any favorite teachers and speakers, and do they use any techniques not represented here? Let me know what you think in the comments!