Greetings all.
I have a fertilizer rate dataset in which yield (response) plateaus at a given fertilizer threshold (explanatory variable), which is common in rate trials.
I have fit the data with both a linear-plateau and a quadratic-plateau model but am not sure which is the best fit.
I know that there is no single agreed upon metric that measures goodness of fit for nonlinear models. I've seen prior posts discuss pseudo-R^2 for nonlinear models. Would it be best to calculate pseudo-R^2 for each model and compare? These models are nested, so I believe I can also conduct a chi-square test to determine which is a better fit. If so, is this simply something to be done by hand or do macros exist for these types of comparisons, similar to the %GOF macro used for nonlinear mixed models?
Thanks
