Research Statistics Central

Objective :  We are fitting nonlinear regression lines to data, but have multiple groups (treatments), each with its own line.  Since groups are a factor of interest, particularly in how they change the lines (parameters of the model), we want to compare parameter estimates among the groups. First approach is to fit the nonlinear model to each group separately, then compare the parameter estimates using t-tests.  The code below generates a random example dataset, with 8 replicates for each of 5 treatments, all measured over 12 days.  Then Proc Nlmixed is used to fit the model explaining change in prate with water changes over the days, "by treat", and parameter estimates are output to data ppp.  This ppp dataset is processed to collect the estimates and standard errors, and t-tests are calculated for all 5*(5-1)/2 comparisons.  Code will need to be customized for new data, including number and values of treatments, degrees of freedom, and parameter names.  Proc NLIN could