Dear Statlist users,
I am currently struggling with an unexpected error in Stata/MP 18.0. I use longitudinal data from the SHARE-project (https://share-eric.eu/).
To explore my data analysis strategy I am looking into logistic regression models using the logit command and multilevel logistic regression models using the melogit command.
Everything runs well with the logit-command. Switching to melogit, a null model (outcome variable, constant and random intercept variable) and models including continuous covariates run. However, when specifying a factor (categorical) covariate using a prefix (e.g. "i." or "b1.") Stata returns r(119):
This is odd as I didn't use the mi set command. If I check with mi describe or try mi unset, Stata returns errors because it -logically- does not find the required _mi_ variables.
It is also unclear to me why it errors only for melogit models with factor variables (and not logit).
I went through all steps of my data management again and found out that the error already occurs after merging all different original SHARE datasets (9 waves of +- 10 separate questionnaire modules). Is it possible that the error results of one of the original merged datasets being mi set by accident?
And is there a silver bullet to undo this in my final dataset or do I need to check all 90 original datasets one-by-one? As far as I am aware the orginial data sources do not include any _mi variables neither.
I am currently struggling with an unexpected error in Stata/MP 18.0. I use longitudinal data from the SHARE-project (https://share-eric.eu/).
To explore my data analysis strategy I am looking into logistic regression models using the logit command and multilevel logistic regression models using the melogit command.
Everything runs well with the logit-command. Switching to melogit, a null model (outcome variable, constant and random intercept variable) and models including continuous covariates run. However, when specifying a factor (categorical) covariate using a prefix (e.g. "i." or "b1.") Stata returns r(119):
no; data are mi set
Use mi fvset to set or query these data; mi fvset has the same syntax as fvset.
Perhaps you did not type fvset. Some commands call fvset to obtain information about the
settings. In that case, that command is not appropriate for running directly on mi data.
Use mi extract to select the data on which you want to run the command, which is probably
m=0.
r(119);
Use mi fvset to set or query these data; mi fvset has the same syntax as fvset.
Perhaps you did not type fvset. Some commands call fvset to obtain information about the
settings. In that case, that command is not appropriate for running directly on mi data.
Use mi extract to select the data on which you want to run the command, which is probably
m=0.
r(119);
It is also unclear to me why it errors only for melogit models with factor variables (and not logit).
I went through all steps of my data management again and found out that the error already occurs after merging all different original SHARE datasets (9 waves of +- 10 separate questionnaire modules). Is it possible that the error results of one of the original merged datasets being mi set by accident?
And is there a silver bullet to undo this in my final dataset or do I need to check all 90 original datasets one-by-one? As far as I am aware the orginial data sources do not include any _mi variables neither.
Comment