topic Help with the %beta_regression macro in Statistical Procedures

Help with the %beta_regression macro

buhl2752 — Thu, 18 Aug 2016 21:20:43 GMT

I am hoping to use the %beta_regression macro for an analysis. But first I was trying to run a very simple data set with the macro to help me understand how it is working and to see how it differs from specifying a beta dstribution in GLIMMIX. I am interested in the macro instead of GLIMMIX becasue my data is zero-inflated proportion data. Unfortunately the macro will not run - it gets hung up on the nlmixed procedure and tells me that there was an execution error for observation 1. currently I am not sure what is going wrong and am looking for assistance from anyone familiar with the macro of nlmixed. My sample program is attached. Thanks.

Re: Help with the %beta_regression macro

ballardw — Thu, 18 Aug 2016 21:36:33 GMT

Run the code with Options Mprint;

Post the entire log.

Be prepared to describe the input data in some detail.

Often the "error" reported at the end of some of these macros is because something else earlier happened and you are only mentioning the final step. All of the intermediate steps need to be examined.

Re: Help with the %beta_regression macro

buhl2752 — Thu, 18 Aug 2016 21:54:15 GMT

I think I solved my own problem. Since I did not assign starting values for the parameters, it assigned a starting value of 1 for all parameters. If I provide starting values for my parameters with a parm statement, the procedure will run. Sorry for the premature post, but if anyone has any other insights, I would appreciate any help. Thanks.

Re: Help with the %beta_regression macro

ballardw — Thu, 18 Aug 2016 21:50:43 GMT

No problem. You aren't the first person to solve a problem after asking for help.

I might recommend that you edit your last post to emphasize the "provide the required parameters for the macro" and then mark it as the solution.