⚠ COVID-19 INFORMATION: Vaccine Information, Other Resources 

Data visualizations, part 2

Feature graphic from Ann K. Emery (https://depictdatastudio.com/).

So you have put your thinking cap on and figured out what you need to say and how you are going to say it.  And, for course, your data are cleaner than heaven itself.

Your first step in figuring out how to present data is to change your frame of reference.  You are no longer an analyst, you are an educator.  And your job is to teach your audience what you learned by doing the analysis, erasing any unnecessary confusion along the way.  You may still be confused yourself, which means either you are not ready to present or that building your presentation's first draft will really help you get clear.

Here is a diagram of basic table architecture that will help you in your quest to get rid of unnecessary confusion.  All the pieces of the architecture are required for a good, confusion-reducing table.  And each piece of the architecture needs its own attention --

From: United Nations, 2009. Making Data Meaningful (https://www.unece.org/stats/documents/writing/).


The title.  The table needs a simple and well-worded title.  It should differentiate the table from any other tables being presented and help the reader find the core message you are delivering.  Certain disciplines have 'required' elements for table titles – finance and statistics both have standards for table titles. Pay attention to these standards: they are there to help.

Data sources.  Include the source of the data.  This is a gift to yourself as well as to your audience.  A year from now, someone will ask you to update this table – pay yourself forward. Sometimes the source of the data is imbedded in the report elsewhere.  You can leave it out, but you might be sorry.

Notes. Footnotes clarify things that will make the reader pause.  Be proactive.  Try to anticipate their questions.  Add notes to explain, for example, why some data are missing or if you are using fiscal year instead of calendar year. 

Colors and designs. Reduce colors and extras as much as possible.  Some might think this makes their tables look boring, but only add colors when you need them to highlight something special or make the table more readable.  The column shading was added on the example table below because the confidence intervals make the columns bleed together.

Data (the star of the show).  Please pay close attention to the following, because this is where the going gets tough.

1.  A HUGE source of confusion for the reader is whether the main comparison is to be found reading the rows across or the columns down.  Do your readers the favor of adding a TOTAL row or column, even if it is populated with '100%,' '100%,' '100%'.  Your readers will send their eyes in the right direction without even thinking and be ever so grateful. If you have many tables, stick with the same orientation.

2.  It is customary that column headings reflect the 'dependent' variable (the thing you are trying to understand).  If you are trying to shed light on readmissions, the column headings would have the readmission information (readmitted or not, readmission within 7 or 30 days, etc.) and the row stubs would break the data into slices of 'independent' variables (the things that help you shed light on readmission patterns - age of patients, insurance type, etc.). 

However, sometimes doing this is confusing.  In the sample table above, the dependent variables are in the row stubs.  Their labels are too long to fit cleanly in column headings.  The point is: follow custom, but don't when it gets in the way of a clean message.

3. Sort the row stubs by frequency so that the most frequent item is first.  For example, a table that breaks down the population of each county in the Chicago region will always start with Cook County, which is the most populous.  Some exceptions:

  • Numeric and time categories are always listed in ascending or descending order, so age groups would typically start with the youngest even if it is the smallest group.
  • If repeated tables with the same row stubs would have a different sequence because the frequencies change in each table, chose a sequence that is logical and keep it the same throughout. Make it easy for your reader.

4. Align decimals. The easiest way to do this is to stick to a consistent number of decimal points and right justify, but that is not always possible. 

All this architecture applies to Powerpoint presentations, too. Using Powerpoint usually means you will have to summarize your data to a simpler presentation and will have to work harder to tell a clear story.  When in doubt, have someone unfamiliar with your work take a look.  Listen carefully to what they are confused about (their confusion is your teacher) and clarify those things before presenting the table to a larger group.


Subscribe now to have updates from The Why Axis delivered to your inbox.