I am new to creating scatter plots and as I work my way through a Cambridge publication a guide to year 10 Australian mathematics, Advanced, I’m aware they are likely only touching upon the subject.
I have observed that “changes in scaling can change the apparent strength of the relationship” as Statistics by Jim puts it, still, I can’t find any reference to the idea I’m thinking of which is that shouldn’t there be some formal convention around the scaling of scatter plots so they may be compared without rendering the data afresh (if it’s even available).
I know that correlation coefficients give this objective measure of strength and I could induce that there is a practice of making the x and y axis not significantly bigger than they need be.
But I’m still wondering if I may not have missed something. Does such a convention exist?