You’ve gathered data to support the case you’re making. But how can you present that data so that the audience understands—and more—is grabbed by it? John Felton, Nebraska Library Commission Planning and Data Services Coordinator, offers some tips and techniques to make your data lift off the page and capture your audience.
NCompass Live: January 13, 2010.
4. Ask Yourself: What is my Message? “ The information that’s stored in our databases and spreadsheets cannot speak for itself. It has important stories to tell and only we can give them a voice.” -Stephen Few
11. Context & Meaning Attendance at Cornhusker Memorial Stadium 2008 = 595,490 Visits to Nebraska Public Libraries 2008 = 8,606,618 , which would fill the stadium 106 times.
12. Finally… “ It all boils down to Communication ” -Stephen Few
17. A Simple Table of Data Charts Chart Type Typical Applications Notes Line Charts - Find and compare trends, - Display a change in direction, - Compare two data series over time - Show correlation - Show rise & fall of values over time In a time series, the category spacing on the x-axis should be proportional. Only use line charts when the x-axis variable is continuous (time, distance, etc.). Area Charts - Display data change over time - Compare two or more quantities Based on line charts, but shows magnitude better. Column Graphs - Show frequency distribution (histogram) - Show comparison of data sets - Show relationship between data series Multiple columns can be used to present data for several variables. Avoid stacked bar charts. Bar Charts - Good for ranking data sets - Show comparison of data sets To highlight high values, sort in descending order, to emphasize low values, sort in ascending order Pie Charts - Compare data sets as percentages of a whole Popular, but has limitations. Use no more than five “slices.” Label the slices themselves instead of using a legend. If the values are close, distinctions will be difficult to decipher. Our eyes are great at comparing line lengths, but can’t judge angles very well.
18. A Primer: Chart Basics The x Axis – for Categorical Information – Providing Context for the Data The y Axis Used for Quantitative Information – Your Data
19.
20. A Line Chart Used as a Time Series Stressing Direction
21. A Bar Chart Used as a Time Series Stressing Magnitude
22. A Line Chart – Used as a Comparison of 2 Data Sets
I spend a lot of time collecting and analyzing data, but today I’m going to talk about the next step – presenting data to your target audience. What I’m not going to do is provide you with a step-by-step tutorial on how to create charts and graphs from spreadsheets. There are lots of free tutorials with your applications and on the web. In fact while I will show you how charts can be a good method for presenting data, I plan to go beyond those techniques and demonstrate some other means of making data come alive for your audience.
So, here is our agenda for this program. So settle in and get ready by…
Getting your virtual treats before we get started. This is probably what I would be providing if we were all together, but at least I can give you an idea of what might get you through my presentation. I know I’ve had my coffee.
Before you can present anything, you need to become familiar with your subject and decide which message to convey. Once you know which story to tell, shape your presentation around it to help your audience make sense of the data being shown. For instance, your message might be to show that more residents per capita in your community make regular use of the library than other similar communities. That’s your message – now you can plan how best to convey it. My quote is from Stephen Few, a data presentation expert who has published several books & articles on the subject and publishes a blog called Perceptual Edge.
Next, it’s important to understand your audience. Are they going to be familiar with complex data visualization techniques, or will it be more effective to stick with more common styles, like line charts or bar graphs? What will hold their interest and help them focus on your data? Who are they? Your library board? The city council or village board? The county Commissioners? Or a group of your patrons?
These simple rules apply regardless of the type of presentation you’re making, but are particularly apt when making data understandable.
The quote on this slide is from another expert, Edward Tufte, who is everyone’s guru in the data presentation field. Make sure that the core message is the focus of what you are showing your audience. Be direct and clear about it and don’t lose the numbers in an overly complicated chart or graph.
You’ve heard this before: Keep it Simple. I’m sure you’ve seen some fancy, colorful charts using lovely background images in publications, on the web, on television. Next time you notice one, see how easy it is to figure out the message for that chart and which parts are really just distractions. Sometimes, being “cool” doesn’t translate into being effective.
What do I need to say about this? If you want your audience to pay attention and remember what you presented, give them just what is needed and no more. One of my former English professors gave me the best advice I have ever received about writing clearly. She simply said, “Pack meaning into your verbs.” It expresses the point two ways – She didn’t mince words and the meaning of the short sentence came through and stuck with me.
Oh my – how the media have re-defined the terms accuracy and reliability. This graphic has been all over the Internet since it was actually used in a live broadcast. I’m obviously showing you what to avoid in my slide this time. Here’s a tip: pie charts must add up to only 100% (not 193%) since they express a parts of a whole relationship. So, double-check your table of data first and then do the same for any chart you produce or you will immediately lose all credibility.
You have probably heard someone say that “Content is King” for web sites. Well, for presentation of data, I would say that Context is King. Statistics have no meaning without meaningful context. You can say, “Our circulation last year was 60,000, but so what. Unless you provide a context like what the total was the previous year or how it compares with another library of similar size, it means nothing. For context and meaning on this slide we have a graphic that everyone in Nebraska has seen – memorial Stadium full of more than 80,000 Husker fans. Relating this to library usage gives the audience something familiar with which to compare the numbers
The zebra on the right might be telling his companion something really important, like” there’s a pride of lions creeping up on us” So you want the message to get delivered clearly and you want the audience to easily understand it. This is the crux of what I’m talking about today. So, what is the correct method of presenting your data. Well, it depends…
Somethimes just presenting the data itself is the most effective method. Here’s a simple table of financial statistics about a mutual fund that I found in a newsletter. This same data was presented in a chart on the following page, but I had already gotten the message quite clearly from this table. When you have a small table of easily understood statistics, just emphasizing the pertinent numbers as they have here can tell your story better than converting it into a visual that will have to be de-coded by the audience. Tables are best when you need to look up or compare individual values and you need precise values.
If you do decide that a chart is the best way to explain your data, how do you pick the right one? Well, there are some tools on the web that can help with this task. JuiceAnalytics is one of my favorite sites and they developed this handy application for choosing which chart to use. Are you going to show a comparison, click here. Is it a trend you want to demonstrate? Click here or which harts work best for that. Do not, however, substitute someone else’s decision about chart type for your own intuition and understanding of your audience.
Here’s another interesting tool that can be downloaded from the web. It’s a similar method, using the message type to pick a chart.
Here’s a application that I thought was quite clever, using the periodic chart of elements as a model for picking the proper visualization, of data, concepts, information, and other things. I’m not wild about the examples they show on the mouse rollovers, but the idea is sound.
And, finally, here is my own simple chart for deciding what to use. Next I’ll be showing you some examples of these charts
Before we get into the actual charts, here’s a primer on the parts of a typical data chart. I still get confused when I am building a chart and the menu wants to know which information to put into the x-axis. So, here is a cheat sheet we can all use. Remember what I said about context – you’ll find it in the title and the x-axis of this chart.
And, as we go along blithely creating charts and graphs, be aware that not everyone will appreciate what you are doing. Here’s a funny chart that says it all. Who says data geeks don’t have a sense of humor?
Finally, a chart. Line charts are quite common, probably because they are so easy to produce. They just use a line to connect data points you plot. They are really good at showing fluctuations in values over time. As you can see here, the emphasis is not so much on the numbers as it is the direction of the trend. Remember that these are the only charts that display data contiguously – in fact, you can only use a line chart when the variable on the x-axis is continuous, like time, distance, or temperature.
This is a column or vertical bar chart displaying the same information as the previous line chart. You might want to use this type of chart when you are more interested in showing the size of the values than in showing a trend.
Here’s a simple line chart again, this time used to compare and correlate two data sets over time. Again, this chart stresses the direction of the companion values. It’s easy to see that children’s circulation has been leveling off, while adult circ continues to rise.
An area chart is basically just a line chart with the values filled in. These are better at showing magnitude of the data and adding interest.
A column chart can also be used to display frequency distribution – also known as a histogram. In this example, it shows the distribution of test scores on the x-axis and how frequently those scores were achieved by measuring the number of students scoring in each range.
Bar charts are great for demonstrating how data are ranked. If this were a chart of survey result, for instance, you would show the possible responses to a question on the y-axis and the number or percentage of participants who selected each response.
Ah, the wildly popular, but ,often misused and misunderstood pie chart. They are really appropriate only show a parts-of-a-whole relationship. When I first started using charts and graphs, I thought pie charts were just so cool and fun to make. But the more I examined the results the less enamored I became. The problem is that if you want your data to speak, you may find that pie charts look good, but mumble when they attempt to tell your story. They are only useful for a small set of percentage values. If you have much more than five slices, it gets too crowded to discern the values, especially if the values are very close in size. For the human eye, angles are more difficult to estimate than distances like line lengths.
As shown here, sometimes you can use a bar chart instead of a pie to display parts of a whole more clearly.
Or, here is another alternative to a pie chart that has begun to show up in data visualizations. It’s easier to decode this chart and to see small values such as this example , where the single square really shows how small savings is as part of household income.
It is so easy to misuse a pie chart! Remember the caveat about not using too many slices, well here’s a exasperating violation of that rule. Does this chart mean anything to you?
Breaking the rules of clarity, simplicity, and brevity all at once is this abomination. It’s a glammed up version of the pie chart’s foodie cousin, the donut chart, which here is trying to show something about consumer expenditures. It isn’t necessary for a chart to be read at a glance, but come on! It would take considerable study to make sense of this thing. I’d prefer to just see the numbers. This, my friends, is what we call ChartJunk.
To sum up the problems with pie charts, we have this brief, simple illustration.
Switching gears quickly, let’s start talking about different types of data visualization. This is an example of a data dashboard that’s becoming popular in situations where managers need continuous, real-time updates of what are know as “Key Performance Indicators” (KPI). These applications are tied to computers that track and automatically update databases of information. This is a pretty useful example, although sometimes people go a bit overboard on the dashboard metaphor and try to show create “fuel gauges” and speedometers showing changes in indicators.
Here we have my own hybrid of a data dashboard and a summary of spreadsheet data. The information is static because it’s not critical for real-time decision-making, but it does present on one page some significant statistics. It’s a snapshot of data and also a combination of both tables and charts.
Geographic context is often important and we now have map mashups that make it relatively easy to show data by location. This map mashup might be familiar to you as one I created to show annual library statistics in a new way. It tells the viewer at a glance how public libraries are distributed across the state, with the colors indicating the population size of the communities they serve. You can easy say that 58% of libraries in Nebraska serve communities with populations below 1000, but with this map you can demonstrate how that looks.
Here’s a similar map that demonstrates the level of Internet access among libraries in the state.
Even though it’s not interactive, I love this map by David Drozd from the State data Center at UNO. It is so easy with this map to see quickly how the population has been changing in the state.
Another approach to presenting statistics in an interesting manner is shown in this “By the Numbers” publication from ALA. It draws the reader’s attention to the oversized numbers and them compels them to look closer to find the context in the accompanying text. It pulls the audience into the page and then provides an opportunity for some self-discovery. This means that the information will tend to stay with the reader.
Using familiar comparisons also helps, as I showed earlier in the slide about context. The graphics and the invitation to relate library data with local landmarks helps bring home the message.
Pockets guides such as this one from the New Jersey State Library are like print versions of the “elevator speech.” You can pack your most significant or surprising statistics in to a small, easily distributed handout for quick distribution of information.
Now, back to those charts…This is part of the choices from which you are asked to choose when charting in Microsoft Excel 2007. Think this is an improvement? Well, only a few of these choices are good choices for an effective presentation. The rest are just eye candy that encourage the user to produce chartjunk. Microsoft and other software publishers would do better if they provided a few really good chart templates and give the user a useful lesson in selecting the right chart for presenting the data in your database or spreadsheet.
That’s my presentation. I hope you don’t all look like the audience in this picture. If you want more information about the topic, here’s my contact information.
One more thing. Here’s a pitch for visiting the Library Commission’s web site. We work really hard to provide up-to-date, useful information on this site and I want to encourage everyone to take a look at it frequently to keep up with what’s going on with our services. On the home page at the top, you will see important announcements. The NCompass blog in the second URL contains some interesting and topical articles. If you want to know more about library statistics, the annual Public Library Survey (Bibliostat), or national library surveys, check out the data services page that’s located thru the third address. And to keep up on NCompass Live presentations like this, check out the last URL, where you can also find out how to view past programs.