Blended Cube - easier to analyze but not as accurate.

Spyglass Dicer, the computer programon which this tutorial focuses on, reads in data, creates a 3D graph and allows the user to slice the graph on any axis. The color within the graph can be viewed in gradients which is much easier for the human eye to analyze but the accuracy suffers.

The following example is a data set from research conducted by a the Psychologist, Dr. Bill Lawless. Lawless tested thirteen subjects under three conditions - cooperative, ambiguous, and competitive - ten times under each condition, and characterized each sunject's behavior accordingly. The resulting behavior corrisponds to the following colors; cooperative in blue, a competitive behavior is red, and points for which there was not data is purple. Lawless's study on Air Force pilots was done to improve the method for picking which pilots to go on given missions.

The following graph is the view of the Nearest Neighbor Interpolation Method in Spyglass Dicer - which does not blend the colors together. The three testing conditions can been seen from left to write on the floor of the cube. And the ten repetitions can be seen on the left wall. The block view offers a more "honest" picture of the data.

Figure1 - Lawless data represented in color blocks.
(Nearest Neighbor view in Spyglass Dicer)

The following is the view of the same data set using color gradients. (To create this view in Spyglass Dicer, go to File Parameters, and select Linear under Interpolation Method.

Figure2 - Lawless data set viewed with color gradients.
(Linear view in Spyglass Dicer)

While the two figures above look very different, they were generated using the same data set. The nearest neighbor view allows you to see where the lines of the different test conditions begin and end, and gives you the idea of the different data regions. Given more data points the blocks would be smaller, and less stretched out. If zooming in was necessary, the amount of data streching will be seen more easily in the nearest neighbor view.

The linear view with color gradients, the linear view, shows trends thereby making it is easy to compare and analyze the data. If you look at the floor of the cube in the nearest neighbor view, you can see that more subjects show red in the middle (or ambiguous) condition in the first testing event than the other conditions. Still, telling which subjects have more red than others is difficult. The gradient view makes comparisons between the subjects easier, and shows, by the amount of red, which subject displayed the most competitive response in the ambiguous condition.

The following movie uses the color gradient viewing format and shows a plane moving through each axis of the Lawless data set. To test your interpretive ability, there are questions on this data set.

Last revised July 16, 1996