BookmarkSubscribeRSS Feed
smorkoc
Fluorite | Level 6

Hi,

I'm having a problem with the result of test statement. Actually here is my code, but in glm section, Idk what the problem is but i see bunch of dots.. 

 

proc glm data=ndata.butner_total_respiration_new ;
class block Spacing tno;
model log_CO2flux =block spacing tno spacing*tno block*spacing*tno ;
test h=block e=block*spacing*tno/ etype=3; test h=spacing e=block*spacing*tno/ etype=3; test h=tno e=block*spacing*tno/ etype=3; test h=spacing*tno e=block*spacing*tno/ etype=3; run; ods graphics off;
3 REPLIES 3
PGStats
Opal | Level 21

Your error sum of squares is zero. Meaning that all available degrees of freedom are taken by your model parameters. This is not a problem if your intent was to test the parameters with other error terms, as you did with your test statements.

PG
smorkoc
Fluorite | Level 6

So, should my class statement include all the variables seen in error terms?

PGStats
Opal | Level 21

All the terms mentioned in your test statements must appear in your model and all model categorical variables must be mentioned in the class statement.

PG