Hellos --
Writing in the hope someone might have a useful tip!
I get the "mi estimate: omitted terms vary" error message when I attempt to generate mi estimates using a logistic regression of a binary outcome on a vector of binary predictors.
This is the text of the full error message:
"mi estimate: omitted terms vary
The set of omitted variables or categories is not consistent between m=1
and m=8; this is not allowed. To identify varying sets, you can use mi
xeq to run the command on individual imputations or you can reissue the
command with mi estimate, noisily
no results will be saved"
When estimations are computed over dataset m=8 one of the variables, as well as 5 observations, are dropped and estimation stops.
I have tried overriding this problem by including, in turn, each of the following options:
- augment (in the hope the variable would not be dropped (as the reason why it is dropped is that it predicts success perfectly)
- errorok
- esampvaryok
However, none of these strategies overrides the error message. This despite the fact that, for example, errorok should discard results from computations that yielded an error.
I have actually also tried including both errorok and esampvaryok together, but that, too, does not work.
It seems that augment cannot be used in this context. Is there a different option I could use?
I find no guidance by simply googling the error message. Would anyone here have some advice?
Many thanks
G
Writing in the hope someone might have a useful tip!
I get the "mi estimate: omitted terms vary" error message when I attempt to generate mi estimates using a logistic regression of a binary outcome on a vector of binary predictors.
This is the text of the full error message:
"mi estimate: omitted terms vary
The set of omitted variables or categories is not consistent between m=1
and m=8; this is not allowed. To identify varying sets, you can use mi
xeq to run the command on individual imputations or you can reissue the
command with mi estimate, noisily
no results will be saved"
When estimations are computed over dataset m=8 one of the variables, as well as 5 observations, are dropped and estimation stops.
I have tried overriding this problem by including, in turn, each of the following options:
- augment (in the hope the variable would not be dropped (as the reason why it is dropped is that it predicts success perfectly)
- errorok
- esampvaryok
However, none of these strategies overrides the error message. This despite the fact that, for example, errorok should discard results from computations that yielded an error.
I have actually also tried including both errorok and esampvaryok together, but that, too, does not work.
It seems that augment cannot be used in this context. Is there a different option I could use?
I find no guidance by simply googling the error message. Would anyone here have some advice?
Many thanks
G
