Visual Revelations

We discussed Howard Wainer's book Visual Revelations in class today. Here are some of the ideas that I came away from the discussion with:

Charts and graphs are to communicate, and if they don't do so well, just don't use them.

The methods that Wainer discusses to represent data should be tailored to the specific data set of a project.

Double y-axis graphs are the devil and bring bad karma. They can manipulate any data set to portray any conclusion the creator chooses. Don't trust them. Don't use them. Don't let your friends use them.

Wainer discusses the various methods from an almost 'pure data' perspective. I was surprised that visual and hierarchical representation methods were not discussed in the pages that we read. After all, the book is called Visual Revelations. There is a great opportunity to enhance the 'single-glance comprehension' of any data set by using visual elements.

There is a big issue with data presentation standards, like in many design areas. People are familiar with pie charts, so they are often used, although they are inaccurate and ineffective for most data.

The trilinear plot method of representing data is interesting, although I think its presentation could be greatly enhanced by opening it up into a linear format. Because the data progresses from one area to another, and cannot complete a fully-circular pattern, the two most extreme areas should not physically touch - the layout is very misleading and requires the viewer to memorize the rules of the chart before interpreting the data.

If the medium is the message, there is a great potential for using implicit graphs on PCs to enhance their clarity. That way, the reader would be able to choose the data on each side of the chart, and then quickly run through many different scenarios without cluttering up the chart with lines.

Additionally, I had an interesting conversation last night about the book. I asserted that bar graphs are easier to understand than pie graphs because the bar more closely resembles the linear human experience of time (which is common to all), and a pie graph requires a cognitive leap from that format. Riaz and a few others that were milling about replied that time is experienced differently in different parts of the world. As the discussion progressed, I began to see that my experience of NOW can change my future, but can also change my understanding of the past. Therefore, time is not limited to forward movement. Ashton expressed that her experience of time is cyclical, like a clock, to which I replied that that was a learned way of interpreting the world. She disagreed, and seemed to indicate that each day is the beginning of another circular path in time.

There seems to also be a difference between perceived time (time flies when you're having fun) and 'pure' time (as kept by a clock). Wouldn't it be annoying if someone followed you around all day with a metronome?

Wainer says in his book that the standard error of any statistic is 1 over the square root of the sample size, and that this formula is attributed to God - what in hell does that mean?

2/09/2005