9 Comments

    Leave a Reply

    Your email address will not be published. Required fields are marked *

    Pie Chart Makeover: Quarterly Breakouts by Gender

    Mar 30th, 2017 / Data Visualization / , , , , , , , , , ,

    A few weeks ago I was in Indianapolis with the juvenile detention alternatives initiative. This group wanted to graph:

    • how many youth went to juvenile detention centers,
    • by gender, and
    • by quarter.

    Brainstorming

    There are a dozen ways to graph each dataset. For this data, we could utilize:

    1. table
    2. heat table
    3. quarter-by-quarter pies (well, not the best choice)
    4. one year-end pie
    5. line graph with numbers
    6. small multiples line graphs with numbers
    7. clustered bar
    8. small multiples bar
    9. stacked bars with numbers
    10. stacked bars with percentages
    11. stacked area with numbers
    12. stacked area with percents
    13. and probably a few more that I haven’t thought of yet

    Sketching a dozen options for displaying quarterly breakouts by gender.

    Before: Quarter-by-Quarter Pies

    Naturally, the quarter-by-quarter-no-way-these-don’t-work-pies were currently being used by a few people in the room. I adjusted the wording from youth going to detention centers to youth being served so that this example is relevant to even more of you.

    The quarter-by-quarter pies are off the table. According to my pie chart guidelines, gender breakdowns are one of the only times it’s okay to use a pie chart (nominal variable, just a few slices, etc.).

    But it’s really hard to make comparisons across multiple pies. Your eyes would have to jump from the 1st quarter pie to the 2nd quarter pie to the 3rd quarter pie to the 4th quarter pie and back again. I’m tired just thinking about it.

    Pie chart makeover: Quarterly breakdowns by gender. Here's the before version.

     

    After

    After we sketched ideas and discussed the pros and cons of each approach, we decided that stacked bars with numbers were the most promising. Stacked bars can display the number of males, females, and the totals. They’re also easy to label (compared to, say, a stacked area chart). We used vertical bars because time is ordinal.

    My rule of thumb is to display numbers for anything less than 100 and percentages for anything over 100. Fewer than 100 youth went to these detention centers each quarter, so we displayed numbers.

    For a Slideshow: Stacked Columns with a Short Title

    Here’s the after version for a slideshow:

    Pie chart makeover: Quarterly breakouts by gender. Here's the after version for a slideshow.
    For a Report: Stacked Columns with a Subtitle

    Here’s the after version for a handout or report, which intentionally contains a bit more explanatory text.

    Pie chart makeover: Quarterly breakouts by gender. Here's the after version for a slideshow.

    Bonus: Make This Yourself

    No, I didn’t add the totals to each quarter’s bar chart with text boxes. Yes, that’s possible. It’s also a ton of work to crunch the numbers, create text boxes, and center them perfectly above each bar. This is a stacked bar chart with three segments: males, females, and totals. The total segment has a transparent fill so it looks invisible.

    If you want to explore the strategy in more detail, watch the video or download the template.

    Bonus: Download the Templates


    Purchase the templates ($5)

    9 Comments

      Leave a Reply

      Your email address will not be published. Required fields are marked *

      Related Courses

      Most “professional” reports are too long, dense, and jargony. Transform your reports with these practical tips. You’ll never look at reports the same way again.

      Learn More

      Data visualization best practices, practical how-tos, tutorials in multiple software platforms, and guest experts. Designed with busy number-crunchers in mind.

      Learn More

      SPONSORED

      #f-post-el-35{display:none !important}

      #f-post-el-35{display:none !important}

      Learn More