Improving Data Visualizations from the Giving USA Report

Anybody who ever wanted to find out the specifics of philanthropy in the US knows to read Giving USA, the authoritative report on the charitable contributions of individuals and organizations as well as the recipients of these gifts.

I’ve used the report countless times to learn about the trends and to do benchmarking. I’ve also many times become confused and frustrated with its data visualizations.

In this post, I will review some of these data visualizations and also provide alternatives using the Giving USA 2016 report.

Let's get started.


Let’s start with the big infographic.

Can I say it looks cluttered? I feel that too many things are going on that graphic.


  • Unnecessary background
  • Unaligned content circles
  • Unuseful icons
  • Unhelpful chart
  • Extra digits (sorry for breaking the alliteration)
  • Incorrect text alignment

When I see graphics like these, I remember what my doctoral adviser, Dr. Menzies, told us about the world’s smallest mammal, a bumblebee bat. It is shorter than 1.5 inches, yet it’s a fully functioning mammal.

Small or simple doesn’t mean insufficient or ineffective.

Small or simple doesn’t mean insufficient or ineffective. #dataviz

Click to Tweet

People confuse complexity with wisdom. But the masters know that elegance rises from simplicity. Antoine de Saint Exupéry said, “perfection is attained not when there is nothing more to add, but when there is nothing more to take away.”

design perfection quote

Tufte said, “maximize information to pixel ratio”. I want to extend that and say, “maximize information to time ratio.”

Spare your readers. Provide useful information quickly. Refuse complexity.

@EdwardTufte said, “maximize information to pixel ratio”. I want to extend that and say, “maximize information to time ratio.” #dataviz

Click to Tweet

Let's see the problems in detail.

Unnecessary background

I can see the need of using the color palette of Giving USA’s brand, but I don’t know what to do with the coins in the background. They just add clutter.

Unaligned content circles

Although the text in the circles give more information on a few trends, their location makes me search for something that doesn’t exist. My eyes try to follow them, but there’s no prize at the end. 

As given in the book, White Space is Not Your Enemy, the design principles of Contrast, Repetition, Alignment, and Proximity (affectionately known as CRAP), let us see the important things in a graphic. Otherwise, we risk losing our readers.

Unuseful icons

What’s common among a hat, a globe, and a worshipper icon? They are useless, or more precisely, unuseful. They don’t serve any purpose as they aren’t used as legends.

Unhelpful chart

Yes, previously I’ve said some harsh things about pie charts, but this one isn’t actually that bad. The slices are big enough for a reader to quickly asses the proportions, but not necessarily compare them with each other. The big text legends actually saved the reader.

Why do then we need a chart that doesn’t actually help, yet takes a lot of space?

Extra digits

Perhaps, the designers followed some style guide that required to use two digits after the decimal point. I don’t see any added significance, but they further distort the clarity.

Incorrect text alignment

The problem of unuseful icons is made worse with the incorrect text alignment of the recipient categories. The text is aligned right, but the dollar figures are aligned left, making it even more harder on the reader. 

What happens when we try to improve on some of these problems?

giving usa infographic data visualization improved

What do you think?

Yes, I kept two icons, but you could remove them. I also used their brand colors (I'm nice like that).

I’d love to see your improvements. Please post them in the comments.

Let’s see some other charts.

Chart Number 1

giving usa 2016 report data visualization examples

This data visualization shows the total charitable contributions in the US from 1975 to 2015. As soon as you hear a timeline, you should think of a line chart. In this example, however, the designers wanted to show inflation adjusted dollars and the years of recession. They decided on a two-colored bar chart, plus a line with markers.

Bar charts are unsuitable for trend lines because:

  1. They take a lot of space
  2. They hide the trends

Bar charts are more suitable for discrete data.

The golden color to highlight the years of recession are distracting -- it stops the information flow. Also, you need to look for the legends to find what they are used for.

Here’s my attempt:

  • I used the “step” line graph to highlight the changes in giving.
  • I used black and gray colors for the dollar amounts.
  • I placed the labels for the inflation-adjusted and current dollars right above the line, so once you start on the left, you can keep going.
  • I used gray rectangles to highlight the years of recession.
  • I used the gridlines, in case the readers wanted to track individual years. Even without the gridlines, the chart would have given the same information.
giving by year dollars adjusted recession data visualization in R

What do you think?

Pro Tip:

Print your data visualization in black and white. Can you still tell your story?

Chart Number 2

giving usa 2016 report data visualization examples

This data visualization shows the total giving as a percentage of the GDP in the US. This data is grouped in five-year periods.

We see this information in a line-bar combo graph. This is actually a clever graph, but uses too much space to give the most important information: trends in giving over time.

Like chart number 1, bar chart is unsuitable for this purpose.

The jump from 1.6% to 2.2% looks dramatic as the height of the chart is stretched.

Cleveland, in his book The Elements of Graphic Data, described a better way of calculating the aspect ratio (banking at 45-degrees) to see the important trends.

Another bothersome thing is the missing y-axis and labels.

If we don’t have the axis labels and are showing all the data points on top of the chart, do we really need this chart?

Here’s my attempt:

  • I created a continuous trendline, but changed the x-axis labeling.
  • I squeezed the height of the chart to see a visible trend.
  • I added the starting and ending data points.
  • I added the gridlines to provide an enclosure to the chart. You can remove it and you still will get the same information.
giving by year gdp percentage data visualization r example

What do you think?

Chart number 3

giving usa 2016 report graphs improve data visualization

Yikes! This chart is unduly complex.

Worse, it doesn’t serve the intended purpose.

I believe the designer wanted to show the giving trend changes in each of the categories and still let the user compare each of the category with others. Except for the top category (religion), you really can’t compare the trends -- unless you read the numbers.

What’s the point of the data visualization then?

This chart has these problems:

  • The use of color to distinguish the categories.
  • Very small slices of rectangles.
  • The data labels.

You need to many times refer to the legend to figure out the category. The color palette makes it even harder to really compare.

Here are my attempts:

Attempt # 1

If we want the reader to see the trends in each of the category, we can use “panels” or “small multiples” for each of the category. The downward giving trend in religion is clearly visible and you can also see the small uptick in giving to education. In this attempt, I ordered the categories alphabetically.

giving by recipient percentage data visualization example panel improvement R

Attempt # 2

Same chart but ordered descending by the giving percentages. Now you can see that education is the second biggest area of giving.

giving by recipient percentage data visualization example panel improvement R

Attempt # 3

In this attempt, I want to compare the categories with each other. With this data visualization, I created the panels on the years rather than the categories.

You can now compare the giving to religion with giving to education in 1975 and in 2015. You can quickly conclude that the gap has closed down with an increase in giving to education and decrease in giving to religion.

giving by recipient percentage data visualization example panel improvement R

Chart number 4

giving usa 2016 report data visualizations

This chart shows the number of tax-exempt charities between 2005 and 2015. This chart has the same problems as the other charts:

  • Bar chart to show line trends.
  • Data point labels on each of the bar.
  • Missing y-axis and labels.

A good thing about this chart: the y-axis starts at 0. (but it hides the drop in 2011.)

Here’s my attempt:

Since I am not using a bar chart (i.e. the length of the bar) to encode the number of charities, I don’t need to start the y-axis at 0. I did try to use the Cleveland method of choosing the right aspect ratio, but finally chose the appropriate ratio and min-max values for the y-axis by hand. 

I created two versions: one with a step and another with a line. While the line chart accentuates the rise and the drop, the step chart shows the steady rise.

step chart data visualization example
line chart data visualization example

Chart number 5

This chart shows the giving priorities of for-profit companies. The data for this chart comes from CECP, and don’t even get me started on their data visualizations.

This chart suffers from various problems:

  • The color palette is distracting (it almost looks like we want to paint our house and our painter shows the color choices). Also, why is “Disaster relief” is of the darkest color?
  • You cannot easily compare the slices (giving areas). The slices could have been ordered descending on giving percentages with the first slice starting at 12’o clock.
  • It doesn’t add up to 100%. Most likely, this was a typo. The giving percentage to environment is actually 3.1% instead of 5%.

Purposes of a pie chart (or a stacked bar graph) are:

  • to show the proportion of a category to the whole,
  • and to compare each category with the other categories.

We fail at each of the purposes when we have too many slices. We can’t measure angles as well as we can measure lines.

We add colors and legends to solve that problem.

But if we are trying to solve a problem by adding more elements, the design is flawed.

We don’t add unnecessary parts to a machine, then why add extra elements to our graphics?

We don’t add unnecessary parts to a machine, then why add extra elements to our graphics? #dataviz

Click to Tweet

Let’s think about the purpose of a pie chart one more time. You are trying to quickly tell the reader that some proportions are bigger than others and that there is some ranking to those proportions.

Now, we compute these types of things very quickly.

Imagine you are standing on a subway platform during the rush hours.

A train comes up. You have a few seconds to measure which car is less full.

You scan the cars. You think, “pretty full. Full. less full.”

You try to squeeze in the car where you see some space.

Can we do the same to our part-to-whole data visualizations?

Here are my attempts to do so.

I broke the giving areas into three groups: Lot, Some, and Little. I also placed the greater than and equal to symbols to specify the rank order.

part to whole pie chart alternative

If we want to show the exact numbers, we can easily add those.

part to whole pie chart alternative

But let’s say we want to stick to something traditional. We can easily create ordered bar charts to show the giving percentages to each of the areas. You will notice that the x-axis gridlines are muted and show over the bars.

data visualization example improvement R

We can also use Cleveland's dotplots to show the same information using little space. Plus, highlighting the education sector is non-distracting and clear.

dot chart ggplot data visualization example improvement R

Chart number 6

I lied. This is actually not a chart, but a table. Multiple tables.

giving usa 2016 report data visualization example table

These tables suffer from similar problems we saw above:

  • Incorrect text alignment
  • Incorrect use of color
  • Overuse of enclosure lines

Apart from these problems, the structure of the data in Table 2 is very confusing and wrong. It took me a while to understand that the average, median and percentage grant amounts were not subtotal rows, but actual available data points. Because of the highlighting, I kept wondering why the individual data points weren’t listed.

Here’s my take on simplifying these tables.

  • First, I broke the tables apart as they are not showing the similar information.
  • I created a separate row to show the “Combined” aggregates.
  • I added minimal enclosure lines.
  • I removed all the background color.
  • I aligned all the text appropriately.

Here are the improved tables.

improving tables data visualizations
improving tables data visualizations
improving tables data visualizations

What do you think?

Here’s the summary of all the lessons:

  • Avoid unnecessary colors and shapes
  • Choose the right format
  • Remember the purpose of your chart
  • Practice good design principles
  • Align text properly
  • Make it easy for the reader
  • Maximize information to time ratio

Let me know your thoughts on how we can improve these charts further. Or, any other data visualizations that need improvement.

About the Author

A co-author of Data Science for Fundraising, an award winning keynote speaker, Ashutosh R. Nandeshwar is one of the few analytics professionals in the higher education industry who has developed analytical solutions for all stages of the student life cycle (from recruitment to giving). He enjoys speaking about the power of data, as well as ranting about data professionals who chase after “interesting” things. He earned his PhD/MS from West Virginia University and his BEng from Nagpur University, all in industrial engineering. Currently, he is leading the data science, reporting, and prospect development efforts at the University of Southern California.

  • […] charts become confusing or unclear when many proportions, hence slices, are present. They are especially challenging when the slices […]

  • >