Don’t tell Tufte that I’ve used the radius, not the proportional area, of the circle as the value for each ellipse! A cardinal sin that I’m using in this case to improve proportion and clarify a point.
I don't understand why this is okay. I imagine he means it was hard to perceive the difference between correctly scaled circles, but that means the data is close, not that you should misrepresent it.
I can see he's written a book about this, so I should assume he has good reasons, but the ones he gave don't seem it.
The problem is that simply scaling the radius exaggerates the difference between numbers because:
* 26 is 30% more than 20, but
* a circle with radius 26 is 69% larger
than a circle with radius 20.
I just read an article criticizing this exact thing, but I can't find it again.
Actually both area and radius are completely meaningless as a representation of intelligence. Perhaps some figure that represents percentile score would work better -- something like an empty glass for 0%, a glass half-full for 50%, etc; or something like the circles that Consumer Reports uses for rating products.
I don't understand why this is okay. I imagine he means it was hard to perceive the difference between correctly scaled circles, but that means the data is close, not that you should misrepresent it.
I can see he's written a book about this, so I should assume he has good reasons, but the ones he gave don't seem it.
The problem is that simply scaling the radius exaggerates the difference between numbers because:
I just read an article criticizing this exact thing, but I can't find it again.