A few of the million+ decisions your software program can’t make for you:

10. Which chart is best for your data. Yes, I’m well aware of Excel’s Suggested Charts feature. No, don’t even think about it. There’s no substitute for your critical thinking skills.

9. How much ink is necessary. Will your viewers benefit from light gray grid lines, or should you remove them altogether? Ask 10 different people and you’ll get 10 different opinions. You’ll have to think yourself outta this one.

8. How many decimal places are needed. One argument is that decimal places add precision – if you’ve got decimal places, why not display them? Another argument is that our datasets are rarely complete enough, cleaned enough, or generally accurate enough to warrant giving the viewers lots of decimal places (and therefore, a false sense of precision).

7. Whether you’re going to stick with a default color scheme (please, NO) or customize your color palette to match your client’s brand identity.

(And who chose Excel 2013’s color palettes, anyway? Lime green, royal blue, and orange? But that’s an article for another day.)

6. The extent to which you’ll label your data. Do your viewers need to see every point on the line, or just the beginning and end points? Or just the highest line? Or just the lowest line?

5. Whether your chart needs to be rotated to avoid diagonal text. In this scenario the answer is yes! rotate that chart! but your software program won’t automatically do this for you. You’ll have to determine whether your data labels need some additional space, and if so, swap your vertical bar chart for a horizontal bar chart.

4. Whether your viewers need aggregated or disaggregated data. It doesn’t matter how much you paid for your software program. Your computer doesn’t know whether your viewers would benefit from a single visual that contains aggregated information or whether your viewers need to make comparisons across four different variables simultaneously through a small multiples layout. But yes! Your brain! It knows!

3. Whether your viewers will really understand that diverging stacked bar chart, or whether you should just stick to a regular old stacked bar chart.

2. Which patterns to emphasize and which patterns to hide, like whether viewers need to see all the peaks and valleys (line) or whether you need them to focus on just two points in time (slope).

1. And in case it isn’t obvious yet, your audience is pretty darn important. Your software program will never understand your audience’s numeracy level, data visualization level, interests, time limitations, or information needs as well as you.

Ready to dive in and start training your brain to think through these considerations and more? In 48 hours I’m co-leading a public data visualization workshop with THE Stephanie Evergreen in Washington, DC. Upcoming speaking engagements are listed on my calendar.

Want access to the files used in this article? Sign up for my newsletter and I’ll send ‘em your way.

My favorite part of data visualization workshops?

The airports. Kidding.

Watching jaws drop when folks are introduced to small multiples layouts for the first time. Definitely!

What are small multiples?

Small multiples layouts include 2 or 3 or 4 or more graphs – of the same size and scale – to allow for easy comparisons.

Why use a small multiples layout?

Old way: making a single aggregate chart that only displays means, medians, or frequencies from a bunch of different data series combined.

New and improved way: designing a small multiples layout that gives viewers the information they really care about – the disaggregated data for their particular company, city, organization, or school.

Additional benefits:

• To compare changes over time, like the social network map below.
• To compare changes across different groups.
• To add context. Is an organization doing well? Who can really know for sure unless you compare the organization to a few others using a small multiples layout?

What can be made into a small multiples?

Anything. 

Bar charts, histograms, geographic maps, line charts, social network maps, and more.

Just make a collage of several small charts. It’s really that easy.

You can make these in nearly any software program. If they aren’t part of your software program’s menu of default options, don’t be afraid to dive in and use a little elbow grease.

Here are a handful of the million+ ways to incorporate small multiples layouts into your charting repertoire:

Want to make small multiples on your own?

Join my mailing list and I’ll send you visualization resources like the charts and slides used in today’s post.

Or, join an upcoming webinar, like this how-to webinar on December 8th where I give live demos and let you keep a bunch of Excel templates. Templates!? You bet.

Ann K. Emery: Independent consultant!

I’m taking my data visualization and data analysis interests to the next level. Life is short; spend it well. My goal: To equip you to collect, analyze, and visualize your own data.

Is your team overdue to step up your data game? I’m leading full-day, half-day, and multi-day workshops about data visualization, data analysis, and spreadsheet strategies. I’ll come to your organization so your team has plenty of hands-on practice or provide your team with customized webinars.

Busy, and don’t have the time to learn all this yourself? I’m designing charts and dashboards from scratch through freelance projects of all sizes.

Already have reports, charts, and slides, but need a makeover? I’m redesigning existing visuals. 

Don’t have any data and not sure where to start? Have way too much data and not sure what to do with it? I’m partnering with nonprofits to conduct evaluations and improve internal performance management systems.

I’m kicking off the fall consulting season with a full lineup of data visualization workshops, like this 3-hour webinar through the American Evaluation Association (limited to the first 30 registrants).

Interested in my training or consulting services? Contact me. Want to partner on something fun? Let’s talk. Visiting DC ? Let’s grab coffee. Independent consultant? Share your specialties so I can refer people your way.

And now, back to your regularly-scheduled programming.

A few weeks ago I showed you how to create small multiples bar charts simply by creating a stacked bar chart in Excel with invisible white segments. In 5 minutes or less! Today’s strategy takes a few minutes longer, but the result is worth it.

The Original: A Data Table

The fake scenario: We’re prepping for an upcoming meeting with grantmakers. Like most grantmakers (and humans), they’re inundated with too much data. Our job is to make the data easier to “see” and understand. Rather than expecting them to read a 50+ page report prior to the meeting, we’re going to condense a few key facts into a single handout.

The fake data: Health indicators in select states where grantmakers have supported community health initiatives over a 20-year time span. Pretend that health indicator 1 is something like fertility rate (births per woman). I realize the numbers are too high to represent actual fertility rates. It’s fake data. But you get the gist of what this dataset might show you.

Option A: Exploring with Sparklines and Sparkbars

I began exploring the data table with sparklines and sparkbars.

You’ll need Excel 2010 or 2013 for sparklines, which is why I used my Mac for this post instead of my old-as-dirt-but-love-it-anyway PC with Excel 2007. Just highlight the row or column you want to visualize (e.g., C3 through M3), and find the icon that says Insert Sparklines.

Option B: Exploring with Conditional Formatting

Conditional formatting is the 8th World Wonder. Rather than color-coding cells by hand, you just highlight a row or column (like B3 through M3), click the Conditional Formatting icon (typically in your Home tab), and start clicking on whatever you feel like. You don’t need to know what you’re doing. You won’t break anything.

In this example, I selected Color Scales. (Data Bars are also a personal favorite–I teach you how to make them here). After creating Color Scales, I adjusted the print settings so that the heat table could serve as a one-page meeting handout in itself. I’ve been using heat tables a lot recently, so I’ll post more examples in future posts.

Option C: Visualizing Everything in One Chart

Raw numbers for indicator 1 + percentages for indicators 2 and 3 = nope. The computer instinctually knows this is a bad idea.

Option D: One State, One Chart

Again, the raw number plus percentage issue is like comparing apples to oranges.

The general idea here is that yes, you could narrow down the data table by focusing the reader’s attention on a single state. Maybe our group of fictional grant makers is especially interested in one state over another.

 Option E: Three States, Three Charts

Try to look past the horrid default formatting and the awkward raw number + percentage dilemma. The idea is that you compare each state side by side.

Option F: One Indicator, One Chart

You might decide most important comparison to share with your fictional grantmakers is an apples-to-apples emphasis on indicator 1.

Option G: Three Indicators, Three Charts

I’m optimistic that this layout, when formatted properly, would be extremely easy to understand.

Option H: Small Multiples

In my fictional example, I formed relationships with the fictional grantmakers and deduced that they would benefit from seeing at-a-glance patterns of each of the 9 data series separately, i.e. a small multiples chart which is essentially just a visual depiction of a traditional data table. Their fictional meeting allots 15 minutes to view and discuss the handout that I’m about to create, so I decided that displaying each variable in its own chart would allow them to make their own connections between patterns. This small multiples approach contrasts with Option F (where the layout assumes that comparisons across states are most important) and with Option G (where the layout assumes that comparisons across indicators are most important).

First, the unformatted small multiples:

Formatted Version H1: Isn’t it incredible what a little tweaking can do? First, I went through my Data Visualization Design Process article and made sure I covered all the big-picture aspects of the chart. Then, I went through the Data Visualization Checklist that Stephanie Evergreen and I developed to make sure I crossed my t’s and dotted my i’s.

This first version has states as rows and indicators as columns. The area charts are color-coded to match their indicators. State icons are gray.

Formatted Version H2: There are always multiple correct ways to display a dataset. This next version still has states as rows and indicators as columns, but the areas charts are color-coded to match their respective states. Now the state-shaped icons are also color-coded instead of being gray.

Formatted Version H3: I transposed the area charts (i.e., the states are columns and the indicators are rows). Now we can delete redundant vertical axes. In other words, since health indicator 1 ranges from 0 to 8, and all the indicator 1 charts are next to each other, we don’t need to repeat the 0 to 8 labeling in all three area charts. Our readers get more white space and less mental clutter. Without the repeated vertical axes, our small multiples chart is really more of a panel chart, but those nuances are for another day.

This version’s my personal favorite. How about you?

Here’s how the formatted versions compare:

How to Make a Small Multiples Area Chart in Excel

 
So how’d I create those final products anyway? I provide the click-by-click tutorial below. My Patreons can download the Excel file and PowerPoint deck used in this blog post and practice tweaking the templates for themselves.

Most of my design choices are based on the specific group of people I’m working with. Other design choices are based on my aesthetic preferences. Let me know how these strategies apply to your own work. Which option would be best for your audience, and why? Would you format the small multiples charts differently, and how?

In today’s post I’m teaching you:

• how to explore your data tables with Excel’s conditional formatting;
• how clustered bar charts are the worst charts ever;
• how a small multiples chart can save the day by quickly visualizing an entire table’s worth of details in the same amount of space as the original table; and
• how to create that small multiples chart in Excel (and it takes less than 5 minutes, I promise).

The Original: A Data Table

This example comes from a short and sweet Child Trends brief: www.childtrends.org/wp-content/uploads/2013/07/125_Flourishing_Measures.pdf. This is a pretty typical situation in research and evaluation projects in which we’ve got some interesting variables across the top and multiple demographic breakdowns along the side.

Option A: Conditional Formatting

I wanted to see the patterns and understand what was going on. For starters, I used conditional formatting to color-code any cells that were greater than the overall percentage. For example, in Column B, 93% of children “always” or “usually” were affectionate and tender with their parent. Girls were slightly more likely than boys to be affectionate and tender (95% of girls compared to 93% of children overall) so the 95% is highlighted.

Now what can I see? Something cool is going on with the kids above poverty and whose parents have higher levels of education.

This conditional formatting technique is a good start, but it didn’t seem to show me enough patterns, so I kept going.

Option B: Data Bars

Data bars–within-cell bar charts–are another underused conditional formatting technique for exploring preliminary patterns from the comfort of your spreadsheet. I teach you how to make them here.

What’s happening with column C? Why are those bars so short? Why are the kids less likely to bounce back? Now that I can see what’s going on, my brain’s full of questions.

Option C: Clustered Bar Chart

I made one of these bad boys just for fun, and the result was even more incomprehensible than I could have imagined. The not-so-fun part is that I still see charts like these in research and evaluation reports. If you’re still using these, please, contact me and I’ll help you out.

Option D: Small Multiples Bar Chart

I could’ve visualized this data through a million different individual bar charts: a bar chart showing the overall percentages for each of the four indicators (the option selected by the authors in their original report), a dot plot showing the gender differences for one or two or three or four of the indicators, a bar chart focusing on parental education, etc.

Instead of highlighting one or two patterns, I wanted to see the entire landscape of what was going on. Same real estate as the original data table. The chart could fit on just one page within the body of a report or in an appendix. But now, my brain receives the information through both images and text, rather than just text alone like in the original data table.

And since my primary goal was to see the big picture, I opted for a generic title, generic subtitle, and emphasis-free color scheme. In other words, my title just states the graph’s content rather than putting forth a “so what?” (it’s modeled after a data table from an appendix, after all). My subtitle describes the data source but doesn’t give my interpretation of what’s going on. My colors are matched to Child Trends’ logo rather than Excel’s default color scheme, but the colors don’t emphasize one category or series of data over another. I’ll describe these distinctions in more detail in a future post.

(You can also view the high resolution image here.)