A recommended paper

[Glen]

I was going through some of my old nomography stuff, and came across this one again - I've been looking for it lately, and I'm glad I found it, because it was better than I remembered.

If you have access to it, I highly recommend Philip Lyle's paper (note only one L in the first name!):

Philip Lyle
The Construction of Nomograms for Use in Statistics: Part I. True and Empirical Nomograms
Applied Statistics, Vol. 3 No. 2 (Jun 1954), pp 116-125

(It's also available in Jstor if you have access to that)

It covers a lot of stuff related to my posts regarding curvy scales. Lyle's approach is better than what I described, in fact, because he is able to adjust (anamorph) two scales at once, to produce alignment nomograms with two curved scales in one go.

As you might guess from the title of the paper, there is a Part II. It covers the particular example of constructing an approximate nomogram from the results of a factorial experiment. It's quite good as well, particularly in terms of seeing some of the work that goes into empirical nomograms.

[There's some statistical jargon that may get in the way here and there if you're unfamiliar with it (it's a bit hard for me to judge, as a statistician, since I'm familiar with the jargon, how much of a barrier it might be), but most of the article is quite practical and chatty, not particularly theoretical, but delivers some solid, powerful ideas at the same time.]