Twitter Me This

Time for a confession: I’ve been a Twitter skeptic from day one.

Even though I understand how it works (140-character electronic updates – “Tweets” – that people post for their followers – friends, family, political junkies – and that fill the gaps between other types of communications, such as e-mail and blog postings), I’ve still wondered, “Why would I want to do that?”

It’s only after experiencing Twitter over time that I’ve come to understand its value. And these real-world experiences have made me care about Twitter in a way that neutral facts or statistics never could. 140 characters cleverly arranged are much more than friendly updates. In some cases, they have enormous influence – good, bad, and occasionally ugly (you know it’s true). Tweets can be powerful.

In reflecting on my skepticism about Twitter, I also realized that I had been a bit of a hypocrite (a Twittercrite?): almost daily, I use display devices such as Sparklines (to name only one) that condense lots of data into one concise display – a sort of “Twitter for data visualizations.”

And as happens with Twitter, once I began using them regularly, it became clear that, deployed in a clever and correct way, this “condensing and concentrating” type of display tool could empower me to deliver far more information on my dashboards and reports than could other methods.

Edward Tufte coined the term “Sparkline” in his book Beautiful Evidence: “These little data lines, because of their active quality over time, are named sparklines – small, high-resolution graphics usually embedded in a full context of words, numbers, images. Sparklines are datawords: data-intense, design-simple, word-sized graphics” (47).

Typically displayed without axes or coordinates, Sparklines present trends and variations associated with some measurement of frequent “sparks” of data in a simple, compact way. They can be small enough to insert into a line of text, or several Sparklines may be grouped as elements of a Small-Multiple chart. Here are a few examples.

Example 1: Patient Vital Signs

Here, 24-hour Patient Vital Signs (blood pressure, heart-rate, etc.) are displayed in the blue Sparkline, along with the normal range of values, displayed in the shaded bar behind them. To the right of the Sparkline is a simple table that shows the median, minimum, and maximum values recorded in the same 24-hour time-frame.

Click to expand

Click to expand

This basic display delivers a lot of valuable information to care-givers monitoring patients, making it clear that during the same period around the middle of each day, all of the patients’ vital signs fall outside normal ranges.

Example 2: Deviation from Clinic Budget

In this second example, we used a deviation Sparkline to show whether use of available surgical-center hours at three different locations is above or below budget. We added two colors to the Sparkline to make clear the difference between the two values (blue for “above”; orange for “below”) within a rolling 12-month time-frame.

Click to expand

Click to expand

Example 3: Deviation From Hospital Budget

Here we created a deviation Sparkline to show the departure from hospital budget numbers across several metrics (“Average Daily Census” and “Outpatient Visits,” for two examples), but instead of using brighter colors to indicate where the performance falls, as we did in Example # 2, we have chosen a pale gray shade to indicate when daily real performance drops below projected targets.

Click to expand

Click to expand

Please note as well that in each of these three examples, we have embedded the Sparklines into the display and provided context through the use of words, numbers, and icons. We do this because most of the time Sparklines cannot stand on their own; rather, they require some additional framework to convey information and signal value to the viewer.

Finally: although I have been a Twitter skeptic, HDV does have a Twitter account at @vizhealth and Tweets occasionally about things that interest us, or what the company is up to. Take a look!

This entry was posted in Communicating Data to the Public, Data Visualization, Newsletters. Bookmark the permalink.

One Response to Twitter Me This

  1. Happy to see on the last example that +/- 1% doesn’t produce an over/under flag. Decimal dust differences should not require any (wasteful) reaction!

Leave a Reply

Your email address will not be published. Required fields are marked *

one + = 6