Why does ln(length) give a lower F-value but seem better for ANOVA?

[MaxManus]

Homework Statement

I am using Anova to check if length is dependent of age. The question is, "why is it better to use ln(length)?"

How should I answer this?
I think both seems linear on the two plots(attachment)
But ln(length) gives lower F-value and isn't that a bad thing so why is it better to use? Both has p-value = 2.16*10^-16 in R.

F-value ln(length) = 483
F-value length = 522

 

[D H]

The question is, "why is it better to use ln(age)?"

You looked at ln(length) v age.

If that is the question, why didn't you look at answering the question?

 

[MaxManus]

You looked at ln(length) v age.

If that is the question, why didn't you look at answering the question?

Sorry, the question is why is ln(length) vs age better than length vs age

 

[MaxManus]

Have tried to use a linear regression on the two models to see which is best by comparing R^2. On linear-linear I got 0.5678. To compute R^2 for log linear I first ran the regression ln(length) = b0 + b1*age
and then computed ^length^ = exp(^b0)exp(^b1^x) and estimated the R^2 between ^length^ and x. R^2 = 0.49, which is worse

where ^text^ means estimates for text