A recent article by Jac Depczyk called “Show me. New ways of visualizing data” in Economist inspired this post. Actually it’s something I’ve been thinking about for a while now. The article just gave me the motivation to post something.
Basic summary of the article: Too much data, a picture is worth 1000 spreadsheets, new cool ways to visualize data. Here are some of the visualisation programs that are mentioned:
Here is a snip from the article:
“Looking through a numerical table takes a lot of mental effort, but information presented visually can be grasped in a few seconds. The brain identifies patterns, proportions and relationships to make instant subliminal comparisons. Businesses care about such things. Farecast, the online price-prediction service, hired applied psychologists to design the site’s charts and colour schemes.”
A completely agree that looking at a spreadsheet is tough and that visualisations can be very useful. I am guitly of using Gapminder in a earlier post. And I agree that the “brain identifies patterns, proportions and relationships” but as one of my first stats profs said, “it’s cheating to make pairwise comparisons with your eyes.” Beyond the cool data visualization we need data analysts who are wiling to share raw data, coding, justify decisions and debate analysis strategies. Like in a simple table a complex data visualization can play with scales to make results seem more important than they may be. This may help convince uniformed users of a certain course of action or conclusion, but in my opinion does not do the research world much of a service. We may also spend a lot of time and effort working on visualizations that show non-significant or spurious relationships. I’ll end with a TED by Peter Donnelly about how stats fool us.