Chapter 10 Data Journalism

10.1 Introduction

Presenting your work to other people can take many different forms.

  1. Written documents are a communication medium that provides the opportunity for the most thorough discussion. The downside is that many people won’t read the document if it is too overwhelming. Having a good document structure where people can jump around the document and read just what they are interested it can mitigate that. Because the author has no direct interaction with the reader, it is important to anticipate the questions that the readers might have and answer them. As a result, creating a written document takes quite a lot of time.

  2. Oral presentations allow you to interact with the audience and that allows you to explain confusing points. The downside is that the audience can’t skip around and re-visit points that they didn’t understand. Most people won’t interrupt to ask questions, so the presenter needs to read facial expressions to see how much confusion there is among the audience members.

  3. Integration into other people’s work. Well-created graphics that clearly explain a concept or idea will commonly be used by other folks. They should clearly attribute the graph to who ever created the graph, but it is useful to make sure that your graphics are in a state that they are stand-alone state that is interpretable.

10.2 Audience

Understanding the perspective of your audience is really important. I often think about presenting information to:

  1. A high school educated individual without knowledge of the subject area I’m working with. Most popular news articles are written are written at this level and most of the details of the subject area are overlooked.

  2. A college educated person with some knowledge of the subject. Articles and presentations at targeted at this level includes some large details and serve to keep individuals educated about fundamental ideas in a field.

  3. An expert in the subject area. These articles and presentations are usually highly technical and very difficult to understand but are very important to the progression of the science.

As we write and create graphics, we want to be mindful of the different audiences and ideally a graph that was an quick introductory slide for a presentation given to an expert might be the final take-home message for the simplest audience level.

A really nice example of detail shifting is given by a YouTube series by Wired Magazine called 5 Levels.

10.3 Outline

  1. Background information - Whatever background information your audience needs. When the audience is not highly educated about the topic, you need to include enough background to understand the context. A common mistake, however, is to include too much background and too much detail and overwhelm the audience. Try to be thoughtful giving just the details that absolutely necessary but it is probably better to give too much background information than too little.

  2. Data Source and Methods - To make a convincing discussion, it is necessary to describe the data that you are about to show and why it is appropriate to inform our discussion. It is also important to give enough information so that the reader/listener could access the data if it is publicly available. Any non-obvious transformations should be explained in sufficient detail as well. As your presentations include more sophisticated methods, some discussion of the analysis justification is appropriate here. The more sophisticated your audience is, the more important this section becomes.

  3. Results - This section is the presents the data gathered and the graphs that support our insights. There might be many several graphs were we see the overall results and then follow it with how those results vary by groups.

  4. Discussion - This is the take home message that you want to leave the audience with. It is important to realize that the audience won’t remember or understand everything written/said so making sure the take-home message is clear is important.

Scholarly articles tend to follow this outline extremely closely and generally have exactly these section headings. Articles intended for wide-spread consumption typically have these same sections but might re-arrange the sections so that the extreme details of the data source and methodology are moved to the end of the article (or in scientific journals, an appendix).

Scientific articles have such a defined structure that most readers tend to read in a non-linear fashion. I typically do the following:

  • Read the abstract, which is a one paragraph summary of the paper.
  • Read the Introduction/Background section.
  • Skim the graphs in the Results section.
  • Read the Discussion section.
  • If I’m still interested, I’ll go back to the Results section and read that in more detail.
  • Finally, if I’m really really interested, I’ll read the Methods section in detail.

10.4 Figures

When creating figures for exploratory analysis, I typically don’t spend too much time polishing the graph because the audience is just myself. However once I’ve picked the general form of the graphs I want to share, I begin the process of polishing the graph for wide-spread presentation.

  1. Make sure each figure has a title that describes what the reader what relationship is being explored.
  2. Make sure all axis labels are understandable without further explanation. If appropriate, include what measurement units are used.
  3. Highlight important feature that you want readers to see with annotations or figure captions.
  4. With multiple graphs, try to add as much consistency in axes, color/shape encoding, etc. This makes it easier for readers to connect the information across graphs.
  5. Consider color scales and typography for how well figures could be shared.

10.5 Presentations

Giving oral presentations are a bit different than creating a written document. In particular talks must follow the order the presenter chooses, but the audience can express confusion and allow the presenter to clarify a point. As a result, it is critical for the audience to provide feedback through questions or at least facial expression.

There are a few issues that are often quite surprising for people who are first giving a “live” presentation.

  1. Projectors have much lower resolution and color contrast.
    1. Color palettes in need to be reduced because we’ll only be able to distinguish between 4 to 5 separate colors.
    2. Dark themes don’t work well. Stick with dark text on a light background.
  2. Slides
    1. Minimalist Content Style- Just headers that serve as a speaker prompt about what to talk about. These give the audience a visual cue when you switch topics.
    2. Detailed Content Style - If you include details, it is important to keep them organized within the slide so that the audience isn’t overwhelmed and not read anything.
    3. No Wall of Text!
    4. It is fine to mix the content styles within a presentation.
    5. Regardless of the style, you should consider each slide should take approximately 1 to 2 minutes. Switching slides every 10 - 15 seconds just leaves the audience feel like you are rushing.
    6. Slides given in person don’t have to use full sentences. However, if people will be reading the slides afterwards, then they should make sense when being read. In that case, the detail level has to go way up.

10.6 Exercises

  1. (20%) Identify a project question and data source(s).Be careful in the scope and ambition of your project. If you just download a data set from Kaggle and thus have very little data wrangling, the visualization and insightfulness needs to be amazing. Similarly, if you get one (or more) messy data sets and have to spend substantial time and effort data wrangling, then my expectation for the visualization is less. Don’t get lazy on a project that is worth 25% of your total score in the class.

  2. Produce an article/essay that has a target audience of your fellow classmates. This article should include the following:

    1. (10%) Introduces your sufficient background information to understand why your question is interesting.
    2. (15% data wrangling) Give sufficient detail into how the data was collected and prepared and why particular variables were chosen. Be sure to describe the data source so that the audience trusts the data source and could conceivable recreate the analysis. A formal “literature cited” section is fine, but a section that clearly links to the necessary information is also sufficient.
    3. (15%) Shows both a general trend and and addresses how the trend changes (if at all) across appropriate sub-groups. Make sure to consider interesting follow-up questions. The primary grading aspect here is the insightfulness of the graphics.
    4. (15%) The article should have at least three different graphics prepared by you. These graphs should be ready to be shared broadly and enough context within the graph to stand on by themselves. The primary grading aspect here is the professionalism of your graphics.
    5. (10%) The article shall be clearly written using correct grammar and professional tone. The article should flow well and be clearly structured. The text will provide a narrative explanation to be read in conjunction with the graphics. For an example of an extremely well written article, see this student example.
  3. (15%) Produce an approximately 8-10 minute presentation of your article. You will present this to the class. My grading criteria will be on how well you translate your article to the presentation,