%mkt merge

[PK247]

Hi,

Is there somebody who can explain how %mkt merge (choice modelling) handles missing data? It seems that missing data are replaced by zero, indicating that the option is "not choosen", however there is an important difference between "not chosen" and "missing". I would prefer to omit observations where the "choice" is missing, is this possible whithin %mkt merge?

Thank you in advance!

[RW9]

I have no idea what this macro is or where it comes from, are you sure you posted in the right forum?  If you want to be sure that missing data is not used in this macro, why not just drop it from the dataset *before* calling that macro e.g.:

data work.inter;
  set have (where=(your_variable ne .));
run;

%mktmerge (dataset=inter...);

[PK247]

Hi,

This macro is used in choice design and analysis. It is not possible to drop the missing observations before, the macro is used to merge the choice design and the data. In the data itself, single missing observations can not be removed as they are part of a choice set.

[Reeza]

Have you checked the documentation?

https://support.sas.com/techsup/technote/mr2010f.pdf