Multiple Imputation MI

[Garance Delagneau]

Hi all ,

Background:

I have a dataset with 698 participants. I'm interested in doing multivariate regression analysis.

I have 24 potential covariates (4 of them are continuous, 13 binomial, 7 categorical), 3 dependent variables (continuous: PL, VS, CE) and 1 independent variable (categorical - 6 categories; traj_alcohol). I have missing data on some of my variables (i.e., my potential covariates). I would like to perform multiple imputations and also to conduct univariate analyses with my potential covariates, DV and IV to know which one to include in my final regression model.

My two questions are:

1. I have used a chained impuattion method but get an error message and I'm not too sure how to handle it.

Command: mi impute chained (regress) mat_ment_health parity folate fad (logit) pregnancy_planning SRI preeclampsia preterm SGA drug child_sex DASS_stress DASS_depr DASS_anx antidep_combined post_smoking (mlogit) psy breastfeed parenting pren_smoking BMI diet ax = traj_alcohol, add(1)

Error: Performing chained iterations ...
mi impute logit: perfect predictor(s) detected
Variables that perfectly predict an outcome were detected when logit executed on the observed data. First, specify mi
impute's option noisily to identify the problem covariates. Then either remove perfect predictors from the model or
specify mi impute logit's option augment to perform augmented regression; see The issue of perfect prediction during
imputation of categorical data in [MI] mi impute for details.
error occurred during imputation of mat_ment_health parity folate fad pregnancy_planning SRI preeclampsia preterm SGA drug
DASS_stress DASS_depr DASS_anx antidep_combined post_smoking psy breastfeed parenting BMI diet ax on m = 41
r(498);


2. Should I impute before or after doing my univariate analyses?

Thank you!

[Rich Goldstein]

re: your first question - your situation is not entirely clear to me but it appears you want to add the "augment" option; see

help mi_impute_chained

your second question is entirely unclear: what "univariate analyses" are you referring to?

[Garance Delagneau]

Thank you that worked!
I'm referring to correlations, chi-square analyses, or regression depending on whether my variables are continuous or categorical (i.e., a chi-square between two categorical variables; regression between a dependent and categorical variables, and correlations between two continuous variables). I want to do these analyses to see which potential covariates are associated with my IV or DVs to see which ones I should include in my final multivariate regression analysis.

[Garance Delagneau]

Sorry for all the questions, but I'm also wondering if I should include my DVs and IV in the registration and imputation steps, and participants with imputed outcomes in the final analyses. I found contradictory information on different papers

[Garance Delagneau]

(I'm new to stata and multiple imputations so I'm a bit lost)

[Garance Delagneau]

I also have this error message that keeps popping up:
variable name: missing imputed values produced
This may occur when imputation variables are used as independent variables or when independent variables contain missing values. You can specify
option force if you wish to proceed anyway.

When I remove that variable from the model, I have a similar message with another variable.

This is my entire code:
mi set mlong

mi register imputed post_smoking preeclampsia preterm pregnancy_planning DASS_stress DASS_anx DASS_depr parity parenting diet mat_ment_health folate fad ethnicity psy breastfeed antidep_combined SRI SGA BMI ax drug VS PL CE child_sex pren_smoking traj_alcohol

mi impute chained (regress) fad mat_ment_health parity folate VS PL CE (logit) ethnicity pregnancy_planning SRI preeclampsia preterm SGA drug DASS_stress DASS_depr DASS_anx antidep_combined post_smoking (mlogit) psy breastfeed parenting BMI diet ax = traj_alcohol child_sex pren_smoking, add(43) augment noisily

I've attached screenshots of the errors and of my variables

Thank you!