“There are three kinds of lies: lies, damned lies, and statistics.”
An interesting article I saw recently (well, actually I listened to podcast about it, then looked it up) about how people feel about different data visualizations.
There were a couple of outcomes that I found very interesting. First is that people are much more engaged with visualizations which speak to them in a direct in personal way (I know, not surprising, but still interesting). Whether it is through demographics, geography, or the story the visualization tells, people are drawn to visualizations that feel specific to them. One important implication of this is that the common practice of starting from an overview of the data and allowing the user to drill in to what is most relevant to them does not engage as well as a visualization which is personal to them (but then allows them to further explore the data if they want).
A second, more surprising (and a little terrifying) result was found when the sources of the visualizations were revealed (initially the sources were hidden to reduce bias). When the sources were revealed and the subjects were given the opportunity to change their rankings, the majority chose not to. Superficially, this seems like a great result – the value of a visualization is perceived as being independent of its origin.
Unfortunately, the reason the subjects gave for not changing their rankings was more surprising:
“We found that many people suggested that information has an objective quality that is immutable regardless of where that data may be showcased…”
So the majority of the subjects indicated essentially that “numbers don’t lie,” a disturbing conclusion to say the least. This obviously discounts the fact that a visualization is typically tailored to show exactly what its creator wants it to, and will often (intentionally or not) misrepresent the data in some way.
I recommend reading the full blog post, and/or listening to the podcast, in case I misrepresented anything (unintentionally, of course!)