has been posted as an exact duplicate on Stack Overflow
In case one sees a case of cross-posting, what’s the recommendation? Post a comment saying that cross-posting is against the rules? Or is there a “cross-posting” option in the “vote to close” menu?
I won’t vote to close this one though - I have to admit that cross-posting in this case was useful (I'm putting myself in a dangerous position here ;-P) Ben Bolker’s answer, using NLS, arrives at different results than me using log-transformations plus LS. This is to be expected, since the log weights the residuals for small values of
month more than the residual for larger
month values. However, I wouldn’t have expected the difference in the estimated curve to be so large.