The art to finding the right graph paper
S. A. Rudin
As any fool can plainly see, a straight line is the shortest distance between two points. If, as is frequently the case, point A is where you are and point B is research money, it is most important to see to it that the line is as straight as possible. Besides, it looks more scientific. That is why graph paper was invented.
The first invention was simple graph paper, which popularized the straight line (figure 1)
Figure 1
But people who had been working the constantly accelerating or decelerating paper had to switch to log paper (figure 2).
Figure 2
If both coordinates were logarithmic, log-log paper was necessary (figure 3).
Figure 3
Or, if you had a really galloping variable on your hands, double log-log paper was the thing. And so on for all combinations and permutations of the above (figure 4).
Figure 4
For the statistician, there is always probability paper, which will turn a normal ogive into a straight line or a normal curve into a tent. It is especially popular with statisticians, since it makes their work look more precise (figure 5).
Figure 5
Sometimes correlation coefficient scattergrams come out at 0.00 with a distribution shaped like a matzo ball (figure 6A). But using "correlation paper" Pearson r's of any desirable degree of magnitude can be obtained (figure 6B). Naturally, negative correlation paper is available; it simply points the diagram the other way.
Figure 6
When you get a cycle where you should be getting a straight line, you use the following method. First, the peaks and troughs of the original plot are marked (figure 7A). Then, an overlay of transparent plastic sheet is put over it, and the dots alone are copied. Now, it is obvious that these points are simply departures from a straight line, which is presented in dashed form (figure 7B). Finally, the straight line alone is recopied on another graph paper (figure 7C).
Figure 7
There is nothing so graphic as a graph to make a point graphically.