log log regression model

when to use a log log regression model?

the book says whenever the dependent variable bears a contstant propotional relationship with independent variable

can some1 explain it more clearly?


y = c(x^m)


log(y) = log[c(x^m)] = log(c) + m log(x)

If we let Y = log(y), b = log(c), and X = log(x), then we have:

Y = mX + b

which is a straight line.

ok so if the relationship is something which is explained by raising the dependent variable to a power than we use log log reg. model?