Wednesday, March 21, 2012

Presenting Data and Statistics

Statistics are the heart and soul of any data communication.
Who has not seen a slide like this one in a presentation? We know we hate them but we just keep on producing them. When the numbers are the primary thing you need to communicate to your audience that does not mean the numbers should overwhelm.
When communicating your data to others, there are a few important tips to remember:
• Don’t overload your audience with statistics. Tell them the story that the numbers reveal, don’t dwell on the numbers themselves. Give them just enough to get the point across. You can always give them additional data if they ask for it.
• Choose meaningful data. When narrowing down your data choices, try to choose the ones that show the magnitude of the problem, data that provide context or meaning, and data that are new or noteworthy.
• Avoid statistical terminology when there are simpler ways to explain results. Most people have low maths literacy, even the highly educated. Unless you’re communicating to scientists, the term “statistically significant” is rarely meaningful. Instead, say “more likely” or “less likely.”
• Avoid data terms like p-value, confidence limit, correlation, regression, and chi square. Instead, try to say things like number, count, percent, rate, and average. These are data terms almost anyone can understand.
• As much as possible, try to turn numbers into words. For example, instead of presenting an odds ratio as 2.0, say that smoking doubles the risk of having a heart attack. Even with relatively simple data, like percents, it can be useful to turn the data into words. Instead of saying "25% of children do not wear seatbelts," say "one out of every four children does not wear a seatbelt." This is a more clear and helpful way of saying the same thing.
Slides and Handouts
In a presentation you have three channels of communication available to you. One is you, the speaker and your words, the second is the slides and the third is the handout. Most statistically heavy presentations fail because the presenter has put too much data in the slides and forgotten that a handout document is a much better vehicle for carrying this sort of detailed information. Put a few pages on the seats before you start, if you really need to refer to some details. Keep the slides for simple, 'glanceable'images.
Good graphs
Graphs are a fantastic shortcut for presenting data. Edward Tufte is the guru of presenting statistical information visually. He coined the term 'Chartjunk' and his mantra is " Minimise the non data ink" . If you are using a pie chart on a slide keep it to 6 slices. A table should have only 4 lines. Colour scales need to be intuitive - not random or decorative, they carry a meaning, and don't use 3D effects and other decorative fillers and frills unless they serce a communicative purpose.

This post is adapted from

No comments:

Post a Comment