Graphical Multiple Correlation of Corrosion Data. By O. B. Ellis. CORROSION, Vol. 8, No. 6, 203-208 (1953) June.
Question by Hugh P. Godard, Aluminium Laboratories, Ltd., Kingston, Ont., Canada:
The simplicity of the graphical correlation method is extremely attractive and I am sure that it will find extensive use in treating corrosion data. I would like to ask what happens if you overlook an important variable in plotting the curves. How does this omission show itself?
Reply by O. B. Ellis:
If an important variable is overlooked, a relatively poor correlation will be obtained. This will be evident in the early work by failure of successive plots to reduce the deviations of the points from the estimated lines. If the omitted important variable is added later, an improvement in the deviations will be noticed almost immediately.
Discussion by F. C. Jelen, Solvay Process Div., Allied Chemical & Dye Corp.,...
