The Dataset: An Ordinal Variable and a Categorical Variable

Our dataset looked something like this. We listed the age ranges down the first column and then recorded the number of males and females in the other columns.

Even if you don’t work in public health, you probably have similar datasets—an ordinal variable (like age ranges) along with a categorical variable (like sex).

Before: A Stacked Histogram

The “before” version of the public health agency’s graph looked like this. They had designed a stacked column chart with one column per age bracket.

After: A Cleaned-Up Stacked Histogram

For starters, we removed the axis titles. I rarely remove axis titles from written products like reports and handouts, but in this case, the speaker would physically be present to explain which variables were on each axis. And, the title makes the axes obvious—the axes literally show new diagnoses by age and sex.

On to the next few edits. Let’s continue deleting unnecessary ink.

We re-sized the graph a bit to fill the slide.

We re-sized the vertical scale (from 0 to 50 instead of from 0 to 60).

We continued decluttering the vertical scale by only labeling the starting and ending points (just 0 and 50). In a moment, we’re going to add values to each of the columns, which means we won’t need an overly-labeled scale.

We decided to label each column with the specific number of people who had the disease.

Then, we widened the columns so that the numeric labels were legible. You can follow my tutorial to reduce the Gap Width between the columns.

We outlined each of the columns in white to ensure that the different colors can still be distinguished from one another if someone prints our slideshow in grayscale. In other words, the white outlines around the blue and magenta rectangles make it easier to distinguish the shades of gray from one another.

We removed the gray line along the horizontal x-axis. It’s barely visible, but every bit counts.

Next, we removed the legend…

… and directly labeled the categories. The Males and Females words are just text boxes that we dragged over to the right side of the graph. The words are intentionally color-coded to match the graph (blue letters to match the blue section of the graph). You certainly don’t have to use blue for boys and pink for girls. In fact, your color scheme won’t look anything like this. I advise using your viewer’s color palette. For example, if you’re designing visualizations for the higher-ups within your own company, then use the colors from your own company’s logo and branding. If you’re a consultant who’s designing visualizations for your client, then use the colors from your client’s logo and branding.

Finally, shrink the age labels until they’re no longer twisted diagonally.

But, now the font is waaaaaay too tiny for a slidedeck (size 14—when I aim for 18+ in slides).

I’m satisfied with every detail of this makeover except the tiny age labels along the bottom. Which brings us to option 2…

After: A Population Pyramid

A second option for this dataset is a population pyramid, in which the age ranges are listed down the center of the graph and the distribution of males and females are displayed on each side.

The graph almost reminds me of a Rorschach ink blot. Or a butterfly. Or two regular ol’ histograms that are tilted on their sides.

Side note for the #DataNerds: Okay, so I don’t think that this particular graph is a population pyramid in the strictest sense of the term.

A traditional population pyramid would compare the number of people in each age range—the population of your city, state, or country–and their sex.

This population pyramid-esque visualization compares the number of diagnoses in each age group and their sex.

It’s the same general concept though—providing both a categorical variable and an ordinal variable together in one graph to compare distributions. If you have a better name for this specific dataset’s graph, let me know.

Stacked Histogram vs. Population Pyramid: You Be the Judge

Which approach is better, the stacked histogram or the population pyramid?

It depends.

The stacked histogram allows us to see the total number of people affected by the disease.

The population pyramid allows us to see the age distribution of people affected by the disease.

As always, your choice of graph type will depend entirely on the pattern that you’re trying to get across to your audience.

How to Make a Stacked Histogram in Excel

I know at least a million people are going to email me and ask me how to make these charts, so to avoid jamming my own inbox and banging Future Ann’s head against the wall, I’m going to share the how-to steps in Microsoft Excel here. I’m using the latest version of Excel on a PC, so your screen may look a little different than mine.

The first graph, which I’m calling a stacked histogram, is the easy one.

Type your numbers into a table like this one. Highlight just this section of the table (everything except the total column, which is optional, and added to the table just out of habit anyway).

Go to the Insert tab and click on the Stacked Column icon.

Then, make the cosmetic edits we talked about earlier, like decluttering the graph and adding labels to the columns.

How to Make a Population Pyramid in Excel

The second option, which I’m calling a population pyramid, is much more advanced. Excel Vizardry Level 10,000! Seriously. Once you make it through these steps, comment on the blog post and let me know. You deserve a virtual high five!

You can’t simply go to the Insert tab and click on the population pyramid icon because it doesn’t exist. That’s right. There’s no button that instantly creates a population pyramid for you. But you can create a population pyramid in Excel with my secret strategy.

The secret strategy is going to look crazy at first. This is a counter intuitive (yet amazing) strategy to fool Excel into making the exact chart you want. You’ll have to trust me for the next few minutes. All the pieces will fall into place and the final product will look great. I promise.

First, you’ll need to transform your Regular Table into a Magic Table. The Magic Table has a few placeholder columns with seemingly random numbers.

How do you know which numbers to type into the placeholder columns? Later, when you’re trying this with your own dataset, just scan your real numbers—like the number of males in Column C—for the biggest number in the batch. In this dataset, the biggest number was a 31. There were 31 males in the 35-39 age range with the disease. Then, choose a number a little bit bigger than the biggest number in the batch. The biggest number in the batch was a 31, so I rounded up and chose a 35. Make your placeholder numbers in Column B and your real numbers in Column C add up to 35.

Do the same thing for the number of females. The placeholder numbers and the real numbers should add to 35.

Yes, the columns are a little out of order. The placeholder space for the males goes to the left of the real numbers, and the placeholder space for the females goes to the right of the real numbers. That’s on purpose. It’ll make sense later. You have to trust me.

Then, add a third placeholder space down the center. You’re going to have to guess on which numbers to type into that column. That’s right. A total guess. I went with 7s after a few iterations of guess-and-check. Again, you have to trust me. This will make sense later.

I often add a “Total” column to check my own math. In this example, each row should add to 77 ( because 35 + 7 + 35 = 77).

You’ve got the Magic Table! Now it’s time to insert your chart.

Highlight the middle section of the table (everything except the age ranges and the total column).

Go to the Insert tab and choose a 100% Stacked Bar Chart. Not a regular stacked bar chart. Not a clustered bar chart. Not a column chart. You want a 100% Stacked Bar Chart.

Your chart will look totally busted.

But your population pyramid is there! Can you see it??

Adjust the colors to make the population pyramid pop. I made the placeholder spaces white or transparent. Ah, that’s much better. Can you see the population pyramid now??

Does the placeholder space make sense? If not, sleep on it, and then come back to this blog post again tomorrow. It will click. I promise. We used a stacked bar chart with “invisible” white segments and placeholder data to fool Excel into making a population pyramid. The placeholder values nudge the real numbers into their correct positions.

The rest of the edits are just cosmetic.

Re-size the graph as needed. I made the graph much larger so that it would fill a slide.

Get rid of all the unnecessary clutter. Delete the built-in title (I suggest writing titles directly into your Word document or adding a brand new text box into your PowerPoint slide). Delete the vertical axis labels (the 1 through 15 along the left side). Delete the horizontal axis labels (the 0% to 100% along the bottom). Delete the legend. Remove the vertical grid lines. Remove the border.

Widen the bars so that your histograms have that smoothed-out shape. I suggest a 10% Gap Width for histograms.

Add numeric labels to the bars. Click on the blue bars to activate them. Then, right-click and select Add Data Labels. Don’t forget to match the text to the graph, i.e., blue font for blue bars. Repeat the same steps for the magenta bars.

Here comes the fancy part—adding the age ranges down the center without having to painstakingly add fifty million individual text boxes.

Click on the center “bars” to activate them. Remember, there are invisible white segments of a stacked bar chart with a made-up size of 7.

Right-click and Add Data Labels to those invisible bars. By default, you’ll see the made-up 7s.

Click on the 7s to activate them. You’ll see little squares surrounding the 7s.

Right-click and select Format Data Labels. The Format Data Labels editing window will appear along the right-side of your screen.

If you’re on a super old version of Excel, then your Format Data Labels editing window will appear in the center of your screen, not on the right side. Bad news—your computer is so decrepit that you’re missing some pretty cool features, like the one we’re about to use. Time to upgrade. I purchase Office 365 for around $75/year and get Excel, Word, PowerPoint, etc. across several different devices.

Check out the Label Options section at the top. By default, your computer will display the Value, which is why we see the placeholder 7s.

Click on the Value from Cells checkbox. This little window pops up:

Scroll all the way to the top of your spreadsheet and find the original data table. Highlight the age ranges, like I’ve done here:

Click Okay on the little window. Scroll down and find your population pyramid again. Now, you’ll see both the age ranges and the placeholder 7s.

We have to get rid of the placeholder 7s. On the Format Data Labels menu at the right, simply un-check the Value box.

See how the age ranges fit juuuuust right down the center of my graph? I had to experiment with a few different values until I got the width just right. I tried a 5, a 10, and a 20. Every placeholder value will re-size the graph a bit. Sometimes that center space will be too narrow. Sometimes it will be too wide. Experiment until the wording for your labels fits the space well.

FINALLY… Label your categories. I went to the Insert tab and created two new text boxes for the word Males and Females. The words are intentionally color-coded to match their charts (i.e., blue text for a blue section of the chart).

Congrats, you made it! Don’t forget to comment on this blog post and let me know once you’ve created your own stacked histograms or population pyramids.

Bonus: Save Time & Download the Templates

Want to explore how I made the stacked histogram and population pyramid? Purchase my Excel file.

Purchase the template ($5)

The post How to Visualize Age/Sex Patterns with Population Pyramids in Microsoft Excel appeared first on Depict Data Studio.

The Project: Need to Display Demographic Data about Survey Respondents

A little context: I recently spoke to graduate students about visualizing their survey data. A group of students had surveyed an art museum’s visitors to gather feedback about their experiences at the museum.

I’m pretending that I’m going to meet with the museum’s higher-ups and share data in a slideshow.

I’d begin with a single slide about the survey respondents. Same thing for a report: Limit your background information to just a page.

In the vast majority of cases, I suggest that you only present one graph per slide. Otherwise, the slide gets cramped and the audience is reading Graphs B, C, and D while you’re still droning on about Graph A. But, nobody really cares about demographic data. I mean, they want to know how many people responded. And they want to know whether the small sample was generally a good representation of everyone else as a whole. Reassure your audience that you gathered a nice big number of surveys and that you had a balanced mix of people, and then move on. Don’t drone on for four separate slides.

Your speaking points would be brief, too: “We surveyed 60 people. 50 percent were first-time visitors and 30% were Harvard affiliates. Most of the visitors were from Massachusetts, but we did have a bunch of other states and countries represented, like Canada and even China! And the people were surveyed tended to be on the younger side; 39 percent were in that 18 to 29 year old bracket.”

The content is fine as-is. We just need to tweak the visuals.

Step 1. Instead of a List, Experiment with a Two-Column or Three-Column Layout

First, break out of the regular ol’ one-column layout and experiment with two-column layouts. As you’re rearranging your key phrases on the slide, don’t forget to delete the bullet points altogether. Those little circles are chartjunk.

Step 2. Add One Visual Per Key Point

Then, add one visual per key point. I used two-slice donut charts to show that 50 percent of survey respondents were first-time visitors to a museum and 30 percent were connected with the nearby university. I used a map to emphasize how most of the survey respondents were from Massachusetts. I used a histogram to show the distribution of age ranges of survey respondents. All of the visuals have been decluttered. In other words, there aren’t any unnecessary lines, borders, tick marks, or legends on the graphs.

Step 3. Apply a Text Hierarchy

Size your text intentionally to guide viewers through the content. The most important information should be large, bold, and dark. The least important information should be small, but not too small. I suggest using at least size 18 for the less-important info on slides, like the “18-29” age bracket labels.

You’ll use your audience’s font to enhance their branding (not my font).

Step 4. Re-Write Your Slide’s Title

Do something, anything, I’m begging you, to translate that jargony slide title for non-technical audiences.

Step 5. Apply Intentional Color-Coding

Finally, color-code by category. Intentional color-coding breaks up your dense visuals and text into a few simple categories.

You’ll use your audience’s color palette to enhance their branding (not my colors).

Bonus! Watch a Video Tutorial

And better yet, watch a sample lesson from my online course, Great Graphs: Transform Spreadsheets into Stories with Better Data Visualization so that you can see the entire step-by-step transformation. Most lessons, like this one, are just a few minutes long. I know you’ve got other things going on. Learn a new technique while you eat lunch, and build your new skills steadily. Enjoy!

The post How to Visualize Demographic Data: From Boring Bullet Points into Great Graphs appeared first on Depict Data Studio.

Then, enter your color codes in Excel!

Step 1: Make Your Chart

Build your graph. Here’s a histogram that displays how many people fell into each age range. The default graph has Microsoft’s blue.

Step 2: Edit Your Chart as Needed

Declutter that cluttered graph. Delete the border, grid lines, title, and vertical axis. Label the columns directly. Nudge the columns closer together.

Step 3: Select Your Chart

Now it’s time to customize the color. Click on the bar, line, or pie wedge that you want to adjust. Can you see that the columns are selected?

Step 4: Click on More Fill Colors

Go to Chart Tools, then Format, and then Shape Outline (line graphs) or Shape Fill (all other types of graphs, like this one). Ignore the default colors! Go to the bottom where it says More Fill Colors.

Step 5: Choose a Custom Fill

On the pop-up window, click on the Custom tab.

Step 6: Type Your RGB Codes

Enter your RGB code. RGB stands for red, green, and blue. This is the recipe of reds, greens, and blues that get mixed together to produce your exact shade.

I used 120, 29, 125, which is the exact shade of purple from a client’s logo. Now I can quickly customize other pieces of the graph, like the numeric labels, because that purple is already stored under Recent Colors.

Bonus

With just a few clicks, custom colors reinforce branding and make you look professional.

Purchase the histogram template ($5)

The post How to Enter Your Custom Color Codes in Microsoft Excel appeared first on Depict Data Studio.

Let’s pretend you’re graphing age distributions for a given county.

You carefully formatted your histogram: you removed the border, lightened the grid lines, wrote a descriptive title and subtitle, selected custom RGB color codes, and called attention to a section of the graph with the saturated action color…

… but your chart still looks weird because the bars are so far apart. What’s with all that empty white space in between the vertical bars?!

By default, Microsoft Excel spaces the bars 150% apart from each other. If each bar is 1 centimeter wide, then the space between the bars will be 1.5 centimeters wide.

This huge space looks odd in a regular bar chart and horrible in a histogram. Histograms, in particular, are supposed to be smushed together. Our eyes are supposed to see the distribution as a seamless, unified shape rather than as a bunch of distinct bars.

How to Widen Your Chart’s Bars

Let’s reduce that spacing! There are only two steps. Follow along; it’s easy.

Step 1. Right-click on any of the colored bars. In the drop-down menu, select Format Data Series.

Step 2. Reduce the Gap Width.

Gap Width is a jargony name that simply refers to the size of the spacing or gap in between the columns. Excel’s default setting is typically around 150%.

Reduce the Gap Width from 150% to 30 to 50% for regular bar charts and from 150% to 5 to 15% for histograms.

Try various spacing options and see which one you (and your boss and viewers) like the best. There’s no absolute right answer on this; it’s aesthetic preferences.

Do aim for consistency within the same final product. For instance, you wouldn’t have a Gap Width of 5% for the histogram on the first page of your report and a Gap Width of 15% for the histogram on the second page of your report. Choose one width and make sure everyone on your team formats their graphs accordingly.

In this example, I reduced the Gap Width to 10%.

Bonus: Download the Materials for Free

Want to explore how I created this histogram? Download the spreadsheet and accompanying slides for free.

Download the Materials

The post How to Adjust Your Bar Chart’s Spacing in Microsoft Excel appeared first on Depict Data Studio.

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.

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.

Examples of Small Multiples

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

Your Turn

Have you seen excellent small multiples charts in the wild? Link to your favorite examples here!

The post Disaggregating with Small Multiples appeared first on Depict Data Studio.

But sometimes we don’t want to summarize, we want to highlight the variety in scores and remind readers that the chart is actually made up of individual people, not just the mean or median. Long live the messy data, the dispersion, the distribution, the spread!

I could tell you a few descriptive statistics: min = 26%, max = 100%, Q1 = 64%, Q3 = 83%, median = 74%, mean = 73%, standard deviation = 15%. Or, I could show you the spread in this unit-chart-turned-histogram.

Unit charts are not your new go-to chart. They do not replace bar charts. They are not appropriate for all datasets. They’re best for those few moments when you choose to emphasize individual units of data. A unit could be 1 person, or 10 people, or 1 school, and so on. Units can be represented in circles or squares or triangles. Units can be stacked on top of each other to form a histogram, or they can be plotted along a line.

The dataviz challenge: Re-create the chart in in Excel, R, or some other free software program. Then, tweet a screenshot to @annkemery.

Bonus: Make a unit chart for your own data. Or, do you emphasize individual differences with other chart types? Share your ideas with the community!

The prize for playing: A professional development opportunity and bragging rights. I’ll post the how-to guide in a couple weeks.

Want to learn more? I’m presenting about charting techniques at the American Evaluation Association’s annual conference on Thursday, October 17, 2013 at 11am in Washington, DC. Hope to see you there!

The post Dataviz Challenge #6: Can You Make a Unit Chart? appeared first on Depict Data Studio.