The adage that a picture is worth a thousand words is especially true in science. Scientists usually benefit by "seeing" their data early as it can allow them to detect patterns and relationships that aren’t immediately apparent. But historically the process of turning data into a dynamic graph or visual model has been expensive, technically tricky and regarded as more of an experiment's end product.
"It used to be, that to do beyond what you can do in one Excel spreadsheet, you needed some [computer] programming background," said Jim Hendler, a professor of computer and cognitive science at Rensselaer Polytechnic Institute and co-author of a study appearing today (Feb. 10) in Science.
Fortunately, cheaper and easier visualization technologies are on the rise. "Now on the Web outside of the scientific community, visualization tools have become easier to use and more viable," Hendler said.
Seeing is knowing
Humans are visual creatures, after all — we have evolved remarkable pattern-detecting abilities — and by bringing numbers to life, scientists can "get insights into data they wouldn’t have had otherwise," Hendler said.
A simple example: color-coded weather maps that show data such as temperature and precipitation. These maps offer a clearly easier method of conceptualizing the depth and breadth of heat waves or storms, say, than by squinting at lists of cities, temperatures and rain totals.
In this way, visualizations are far more than a pretty picture. They can allow for exploration of complexly interacting variables in real-time, as well as reveal when instruments are not performing as intended.
"There are certain kinds of errors that you, with your visual perception, find are easy to recognize and that a computer can't," Hendler said.
By way of example, Hendler mentioned a government database on wildfires in the United States recorded over 40 years. Once the data was converted into a graph, a puzzling dip for one year "jumped right out at you," Hendler said. Sure enough, this apparently fire-free year was no statistical outlier; instead, a previously unnoticed data collection error had occurred at the government agency.
In many cases, the sheer size of data sets had prevented scientists from translating them into visualizations. But with companies like Google and Facebook routinely poring over mammoth databases, corralling numbers has become easier than ever.
"The amount of information in Google is not small compared to a scientific database," Hendler said, "and [Google does] visualizations and abstractions all the time."
Importantly, many free or affordable programs exist today that can visually present data. Hendler pointed to programs like Google Analytics — a free, online, website traffic-tracking service — that can generate handy pie charts and graphs. [Check out: INFOGRAPHIC: Increasingly Digital World ]
Along with an ongoing trend toward making data more web-friendly and exportable into visualization software, scientists will increasingly have the ability to bring their research into view, as it were.
"With visualizations, the goal is not only to communicate science but to perform science," Hendler said.