Hello class. Happy Wednesday!

Tonight, we will review homework one.

We will briefly discuss the re-design we started last week.

Then, we will begin to shift our focus from visualization to narrative.

In upcoming lectures, we will begin to combine these ideas.

Sound good?

But first, let’s remind ourselves how tonight’s material fits into the overall course.

In our upcoming material, we’ll turn to narrative to understand what Kahneman, nobel prize winner, means by using story to enable decisions.

Then, later, we’ll talk about how to combine these components of our course.

And here’s our deliverables timeline. You’ve turned in homework 1 and have started homework 2.

You have all had some graphics practice, so you aren’t beginners with either the idea of data or visualization.

And in fact, you’ve been following along with our Class example data analysis project, Citi Bike. So you now have an idea of what types of data look like and how to create new variables from those you’ve found to visually explore and analyze.

Next up is homework two, due next week.

I’ve also made homework 3 available on our class website.

This individual homework serves multiple purposes. First, it provides each of you with practice investigating and concisely communicating about a project idea. Second, it serves as your first contribution to your group work, which we’ll discuss in a while.

So near the end of class, we’ll talk about homework three and its relationship to the group projects.

Ok, let’s begin to discuss the use of narrative in more detail.

Many people write in a way that writer Joan Didion describes here. Let’s read what she says together:

I write entirely to find out what I’m thinking, what I’m looking at, what I see, and what it means.

Now this is a very powerful way to help our own understandings. This is where many people who write stop, but this approach does not extend to communicating with others.

Let’s review a few points about communication we’ve already considered last week for explaining visuals.

Our goal in communicating with others, I showed you in the context of explaining data visuals last week. From our reading in Doumont, we want to

“get our audience to pay attention to, understand, and be able to act upon a maximum of messages, given constraints.”

So first, we need to know our audience.

And the most common failure in business communications is not explaining things for your audience.

Far more common is people write for their own understanding, but fail to restructure their communication for their particular audience. We’ll talk more about this.

Then, the first thing we need to do is get our audience’s attention. Until we have their attention, they will neither understand nor act.

How do we accomplish these three short steps? That’s the more challenging part! We will build on these seemingly simple ideas throughout the course.

We’ve talked a little about what Doumont means by a “message”, right? He gives us an example of the difference. Let’s take a look.

[FOR POLL ON MESSAGES, NOT JUST INFORMATION, SCROLL DOWN]

These two examples come from Doumont’s textbook. Let’s read the first one:

A concentration of 175 micrograms per cubic meter has been observed in urban areas.

Notice there is no opinion or information added by the writer; it’s just reporting a fact, just reporting information.

Now, let’s read Doumont’s second version.

A concentration in urban areas, 175 micrograms per cubic meter, is unacceptably high.

This time the writer adds their interpretation to the information; to provide something new for their audience. That’s what we want to aim for in most of our sentences.

Generalizing from this example of the difference is important. Does the difference make sense?

I also introduced you to Doumont’s three laws of communication, previously in the context of data visuals.

Here they are, again. Let’s read.

First, adapt to your audience. Second, maximize the signal-to-noise ratio. Third, use effective redundancy.

We will continue investigating each of these the rest of the semester in different contexts.

Tonight, we’ll focus on an analytics executive as our audience.

Now, the first law is to “adapt to your audience.” Let’s consider how we can do that.

This concept of adapting to your audience is the idea of starting your communication from common ground.

Neuroscientist Tali Sharot has studied and, in her book titled The Influential Mind, which she also discusses in the video I gave you to review, has written about the importance of finding common ground as a starting point for changing beliefs.

And, if we recall, our purpose is to change our audience’s mind in some way to inform decision making, right?

The idea I’ve written for you here on the slide is from her. Let’s read together: She says,

“When you provide someone with new data, they quickly accept evidence that confirms their preconceived notions (what are known as prior beliefs) and assess counter evidence with a critical eye. Focusing on what you and your audience have in common, rather than what you disagree about, enables change.”

Numerous studies confirms what she explains here.

Have any of you ever experienced this when trying to convince someone? And we’ve seen this so much in global society, too, right?

So Sharot explains that all of us have prior beliefs on any given topic, even if the prior belief is that we just have no information to make a decision. Now prior beliefs are our audience’s knowledge and opinions on your topic before you communicate. So in the context of our analytics project, at a minimum your analytics executive has not already decided to go forward with your project. Otherwise, our communication would be unnecessary.

With this in mind, how might we organize our communication?

Doumont suggests starting with language that makes your audience receptive to the topic of the communication. Motivating them.

Right afterwards, tell them your main message.

So the first thing you should confront in your communication is to answer the question, “why am I communicating?”

Only after you answer that question in your communication do you go into the various details that support your reasoning.

Keep this in mind as we consider a few other perspectives on narrative structure.

Now there are many, many texts that try to explain a story as fitting into templates for plots and such. These ideas in many ways, first became popular in Western cultures from William Shakespeare, whom I’ve shown here.

Now later, a German novelist and playwright, his name was Gutav Freytag, started studying the structure of stories, including those of Shakespeare.

And he sort of distilled patterns in those structures into what is known as a 5 act structure. That 5 act structure begins with some exposition, then includes rising action, a climax, falling action, and finally denouement or resolution.

Now many authors since him have thought about and tried to sort of label variations of this structure into a collection or taxonomy of stories.

I do not think this body of literature is very helpful on its own for us from an applied sense, even if we can generally map our communication onto a similar structure, like problem to solve, analytics solution, benefits of solution.

But a recent journalist and author, Will Storr, has explored these ideas and concepts using the tools of cognitive science. I’ve given you the first couple of chapters from his recent text on the subject.

And Mr. Storr, in his book — titled The Science of Storytelling — generalized all types of story into something that I believe helps us greatly in communicating.

He distills all stories into describing either unexpected change or the opening of an information gap. Information gap, you say? From our first class discussion, I thought we didn’t want those?! This use is different.

Unexpected change and information gaps are, Storr explains, universally interesting to us as humans.

… “Get our audience to become interested.” (Doumont)

Storr continues that the mechanism that moves a story forward is, you may guess this, continuing to introduce unexpected changes or opening information gaps.

We find something very similar to this concept from yet another successful author that has considered business presentations and consulted on presenting analytics projects, including helping former vice president Al Gore. Her name is Nancy Duarte.

Now this illustration I’ve adapted from Ms. Duarte’s work.

Similarly, with Mr. Storr, she describes creating a gap to create contrast between what is and what may be.

In the context of a data analytics project, we might think about what is, as a problem or opportunity, and what may be, as the new information our project can create that informs decisions.

Then we use something like our methodology and details to close the gap.

Now Ms. Duarte, again like Storr, applies the idea repeatedly to drive a narrative from beginning to end.

I know this may feel a little abstract at the moment, but writing in this way is a very, ahem, repeatable process.

But before we get into examples, let’s touch on a couple more concepts.

Another prominent author, long-time teacher, and journalist — the late, William Zinsser — has taught the importance of what writers, and especially journalists, call a lead and ending.

Zinsser explained that the beginning should provide something new and interesting to the audience, no surprisse there, now, right? And here’s the contribution: that the ending should echo the beginning in some way.

I’ll ask you to spot this in our examples tonight.

For our final concept in communication structure for today, let’s narrow down from document structure to sentence structure.

In a reference I provided you from Booth, the chapter on writing for clarity, the authors explain the importance of structuring sentences where we typically begin with “old” information and from that explain “new” information.

To understand or gain appreciation for this advice, let’s discuss one of their examples.

I pulled the two example we see here from Booth’s reference. Let’s compare the two.

Do you find either to be easier to understand. I’ll let you read for a minute. Then I’ll ask someone to explain.

[DISCUSS]

Part of adapting to our audience is cultural. To over-generalize, communication with others in the United States, and many Western cultures tend to be low context, meaning the communication is more explicit.

But societies differ, and other cultures expect more contextual understanding in the communications.

Ok, let’s take these concepts into a couple of specific examples. The first relates to our CitiBike case study. Both examples are memos, under 250 words, a constraint that mirrors your next homework.

Another component of your next homework is to communicate with a particular audience: the chief analytics officer. And in your homework, I’ve given you another short reference to help you understand what these people do, what they’re responsible for. You should review it.

Here, I’ve summarized their responsibilities, they

head up an organization’s data analytics operations, transforming data into organizational value, and drive data-related organizational change.

This oversimplifies their responsibilities, and again you should review the article I’ve cited as a resource in your homework by Minda Zetlin, titled what is a chief analytics officer?

Let’s consider four examples of chief analytics officers.

Here are four examples of analytics executives from a variety of organizations.

We have Scott Powers, who leads the analytics group at the Los Angeles Dodgers.

We have Kelly Jin, who is the Chief Analytics Officer at the City of New York.

We have Blair Borga, who until recently lead the data group at a tech marketing firm.

And we have Michael Frumin, who leads data science at Lyft, which is the parent company of Citi Bike, the subject of our class example case study.

You should stalk them on Linkedin to try to generalize the commonalities found in their experience as well as learn to appreciate their individual differences, extending that to expecting differences in any specific person you with whom you plan to communicate.

Let’s consider draft communications to two of them.

First, let’s consider Michael Frumin, who, again, runs analytics at Lyft, the operating company for Citibike. I’m listed very basic background information, including his education in computer science and operations research, and has 20 year of experience.

Before we start, review his full bio from the link on his name here. As we consider the next example, a memo to him, consider whether he would understand each word and phrase or whether it needs explanation in any way.

Let’s look at a draft memo to him, which is a continuation of our class example, Citibike.

On the right side of this visual, I’m showing you a draft memo written to Mr. Frumin. You’ve already reviewed this when reading from the section of your reader, it’s in section 1.3, data in wonderland, that I assigned to you in the syllabus.

As we review these memos in the context of the concepts of the elements of writing we discussed in class, I’ll pause with questions for you to contemplate, or think about, what I’m asking.

On the left, I’m reminding you of an organization structure for communications we considered from Doumont.

We should begin by motivating our audience and giving them our main point. Afterwards, we provide details in support.

Does the memo discuss our motivation for writing up front, before getting to other material? I’ll give you a moment to consider whether the memo does this and to identify exactly how, to identify the exact language that provides such motivation. Where do you see that?

[DISCUSS]

Does the title itself provide motivation to the chief analytics officer? If so, what kind?

What about the language, “To inform the public on rebalancing”. Would that have any motivational effect on Mr. Frumin?

To answer this, consider his role as the chief analytics officer for Lyft or Citibike. Would he be responsible in any way for rebalancing, or in how they explain those efforts?

Similarly, Doumont explained that we should uses messages, not just information. Does the memo provide messages first?

Would you consider the title including the second half of the title a message? If so, why? Try to explain. I’ll pause.

Would you even consider that message of the type of a call to action?

And where are the details in this memo; where do you find them relative to the motivation and message?

Ok, let’s consider this memo in the context of another concept.

We discussed the vital importance, from a passage out of Tali Sharot’s book, the influential mind, of establishing common ground.

Do you see any language or concepts in this memo that establish common ground? I’ll pause.

What about a common goal?

What about informing the public on rebalancing? Does the fact that Dani Simmons, the spokesperson for Citi Bike, told the public that Citi Bike was concerned with rebalancing, does that imply that Mr. Frumin would want to address the issue? So is that a form of common ground? What makes it so?

Let’s consider this example in the context of yet another concept we’ve discussed.

We also discussed the idea from Will Storr, from his text titled the Science of Storytelling, as beginning with unexpected change or creating an information gap. Do you find either here?

I’ll pause for you to try to answer that specifically.

What happens in the second paragraph of this memo, for example?

The memo identifies a past research project, right? And then explained that it did not do something, which would be helpful to do, then explains we can begin by doing that … in other words, by bridging that gap.

Might Mr. Frumin, upon reading that, become curious or interested in how we might “begin our analysis there”?

Then, in the next paragraph, does the memo then bridge that gap?

Ok, let’s keep analyzing this draft memo in the context of another tool or writing.

Our next concept is from the late Yale professor and Journalist, William Zinsser, from whom we considered the idea of closing the entire communication loop by having the end in some way echo the beginning. Do you see any evidence of this?

Take a look at the penultimate sentence of this memo. Why does it refer to Ms. Simmons? Do you find an earlier discussion in this memo about Ms. Simmons? See how the end is reflecting back to the beginning?

Ok, let’s continue analyzing the memo in the context of yet one more concept we’ve discussed.

This time we’ll pull out our microscopes and shift from document structure to sentence structure. From another excellent textbook, from Booth, which I’ve given you a reference to, he explains that we need to form the ideas of our sentences referring to old things before introducing new things, and we connect them through logic.

In this memo, are we beginning each sentence with information the author already shares with the audience before introducing something new? Try to identify examples of this? I’ll pause.

Let’s begin with the very first words, the title. Consider the phrase, “To inform the public on rebalancing”. Is this old information or new information. Another way to ask this is, does the document have an antecedent basis for those words? And this is a bit of a trick question, I suppose. I’ll suggest that it does, but you won’t find it explicitly here.

Mr. Frumin’s primary job includes informing the public on rebalancing, so the memo begins by stepping into his shoes and identifies part of his job, and so it provides context Mr. Frumin is familiar with where we can begin this memo. Even the next word of that title, “let’s”, is referring back to common information, he, the recipient, and me, the author. Once, we have common ground established, we can lead the audience somewhere new. In this case, the new is a proposal to do something, right?

Now, on your own, you should carefully review every sentence, every phrase, and ask yourself, does the that phrase begin by referring back to something old before leading the audience to somewhere new?

Now that we’ve thought about how this memo attempts to apply the concepts of the fundamentals of writing we’ve discussed, let’s now consider the substance of this memo, so that you may have substantive guidance in your own assigned memo.

How detailed is the information here? We can craft a test for adequate detail by answering a question:

Would it be enough for Mr. Frumin to have confidence that we have actually found specific data, specific, available measures, that we can analyze to be successful for our call to action? Have we identified specific variables with observed measurements? Have we let the audience in on how we plan to use the information on some level?

Let’s consider another example. The next example, you’ve also already seen in your reader, data in wonderland, same section.

Now more of us have experience with bicycles than baseball, so before we review the draft memo, let me give you some context. The audience for this memo is Scott Powers. Director of quantitative analytics. He received a PhD in Statistics from Stanford.

You should be thinking about your audience’s vocabulary. For example, someone with a phd in statistics would be comfortable with the following concepts:

mode, mean, expectations, maximum likelihood, models, inferences, probability distributions, joint distributions, decision theory, counterfactuals, simulations, R programming language and modeling packages.

And someone who’s job it is to help run an organization that plays baseball games would be expected to know the purpose of the organization — baseball — history.

So, here, Mr. Powers would be expected to know that Sandy Koufax was a pitcher for the Los Angeles Dodgers (a historical point), he would be expected to understand what a “perfect game” is, even if you don’t. He would know what data variables and measures are in the data collection called “Statcast”. And he would know what it means to steal a base. Finally, he would know what a “salary cap” is. This is all part of his vocabulary from experience and education.

To get the most out of analyzing this memo, you should do some basic googling so you understand the concepts identified in the memo. Wikipedia would probably give you almost all these ideas.

Ok, let’s see the draft memo, and question it in the same way we just did for the Citi Bike memo.

Does the memo mimic Doumont’s suggested structure? How? Explain? I’ll pause.

Did you identify the title as motivating this audience? How does it do that? Like in the last example, Mr. Power’s primary responsibilities includes informing game decisions through data science. What does the first sentence describe? “Our game decisions should optimize expectations.”

Does that title include a message? Isn’t it the author’s opinion, or added value, explaining what the Dodger’s should do about optimizing expectations?

Ok, let’s continue with the other writing concepts.

Do you see any language or concepts that establish common ground?

What about a common goal?

As I eluded to, this memo is a bit more technical, even in the title. Expectations are a statistical concept. An expectation, in statistics and probability, is roughly the average of possible outcomes. Does that make sense?

Now, again, Mr. Powers has a PhD in statistics from Stanford. He’s well-aware of this concept. He wants his team to win, right? And that means making optimal decisions. So the title begins by sharing the need to optimize decisions, just in a certain way.

Now let’s also look at the first sentence. This time, the memo draws on history from which Mr. Powers knows. He knows who Sandy Koufax is, he knows what a perfect game is, and he would agree that the Dodgers could not rely on this type of most likely sequence to make decisions. He would agree that the Dodgers would never expect or plan for a perfect game. In other words, he and the author (i.e., me), have common ground or agree with these ideas.

Let’s consider another writing concept. This time, let’s explore concepts from narrative and storytelling.

Back to Will Storr. Have we created unexpected change or an information gap? How? Explain?

Do you see one in the first sentence of the title? “Our game decisions should optimize expectations.” If I’m saying they should, does that imply that current game decisions do not optimize expectations? So immediately, we’ve created a gap, even it is implied. Right? So Mr. Powers, when he reads that, we might expect that the first thought would be “how would we optimize expectations”. Hopefully, he’s motivated and curious to learn more.

As you review the memo, look for more information gaps, and then bridging of that gap through the descriptions.

How about following the advice of William Zinsser, the late Yale writing professor and Journalist? Does the end of this memo echo the beginning? How? Explain? I’ll pause.

What does the first body paragraph sentence introduce? Now look at the last sentence of the memo.

Which is also, by the way, and explicit call to action. “Perfect games aside,” … see how the memo attempts to come full circle in the communication by echoing or reflecting on where the communication started?

What about Booth’s advice to start each sentence with something old or shared information, then leading the audience to new information. Has the memo consistently used this technique? How? Explain?

I’ll give you some hints. What does the use of the word “our” do? As with the last memo, what about the first words and how they relate to the audience’s background.

Continue reading through the sentences. The last phrase of the first body sentence uses “those” in “those, we do not expect or plan.” What does those refer to? Has it already been introduced?

Next sentence, “Since our decisions …” Has our decisions been identified, and is this referring back to that now old information? How does that sentence then lead the audience to new information? It discusses the consequences of the information; “we leave wins unclaimed.”

Next sentence begins, “To claim them,” What does this refer to? The prior sentence? As an exercise, continue walking through each sentence to see how each phrase builds on what came before: old before new.

Finally, as with the last memo, let’s also think about the substance of this memo. To know whether it’s sufficient for it’s call to action, which is what? “Testing the concept of optimizing expectations by modeling decisions to steal”, right?

What details does this memo provide to Mr. Powers such that he would be satisfied that we can carry out this call to action?

Does it identify specific, and available, data sources? Specific variables? What approaches to analysis does it identify for any identified variables? Does it identify any limitations that qualify how we might proceed?

Ok, so you’ve now considered two examples, both of which I wrote in under 250 words or exactly 250 words, so that you would get a better understanding of how much we might maximize our messages, given this constraint.

These are merely draft memos, so they could be improved. But that’s not the point here. I hope they’ve been helpful for you as a teaching tool.

Now let’s try a class exercise. We’ll start to revise a communication for a new audience. We have two forms of the original communication. We’ll get to those in a moment.

First, let me give you some background.

Traffic. How many of you think the traffic congestion is bad in New York City? To put that into perspective, on this slide, this image is of the traffic in Jakarta! Whoa!

So traffic is a problem there.

And there have been attempts to improve those conditions.

A data for good organization recently won an award for a paper they wrote describing their work on a data analytics project for the City of Jakarta. I’ve made the paper directly available to you here. The project used video analysis to identify traffic issues as a replacement to people watching videos.

Let’s review their audience, described on the last paragraph of the introduction. Let’s read that together.

“We want this project to provide a template for others who hope to successfully deploy machine learning and data driven systems in the developing world. . . . These lessons should be invaluable to the many researchers and data scientists who wish to partner with NGOs, governments, and other entities that are working to use machine learning in the developing world.”

So their audience for this paper is different than ours for our memo, right? Let’s consider someone more like our audience.

Juan Kanggrawan leads data and analytics for the City of Jakarta. Here’s a description of this executive.

In a moment, we will revise a document with him as the audience. Let’s look at that document now.

This is a blog post that was published before the project happened. It’s only 124 words. It’s audience mirrors that of the paper.

I want us to practice revising it for a new audience, Jakarta’s head of analytics. In groups, I’d like us to try to re-write the title and first two or thre sentences.

Then, I’ll ask groups to share what they wrote, and how they chose their approach.

We don’t want to take a lot of class time, because we have more material to cover. So let’s aim for about 5 minutes once you’re in groups, then another 5 minutes of sharing.

Sound good?

Ok, that’s plenty for us to think about for tonight. I hope our discussion and code review will give you plenty for your own practice upcoming homeworks.

As always, I’m providing you with references where you can dig deeper into the ideas we’re covering. I’ll stay for questions; otherwise have a great night!