Exercise 18 revisited the relationship between life expectancy and TVs per capita and saw that re-expression to the square root of TVs per capita made the plot more nearly straight. But was that the best choice of re-expression? Here is a scatterplot of life expectancy vs. the logarithm of TVs per person. How can you tell that this re-expression is too far along the Ladder of Powers?

Life Exp 75.0 67.5 60.0 52.5 + -2.25 -1.50 Log (TV/person) -0.75