Cross Validated
2026-11-16 14:22 UTC
By Martina
AI-113-20261116-social-media-a8d167b0
MICE imputation of continuous outcome in model that includes interaction effect between categorical variables
I have a multilevel dataset. My statistical model includes the following predictors: Age group: young vs. old (between-subjects factor) Day: first vs. second (within-subjects factor) Age group x Pulse My outcome has missing data, and it is continuous. No data is missing at the predictor level. The data is in long format I am using mice for imputation purposes. I understand that interaction terms need to be included in the imputation process. I do not understand how to code the interaction as it has no missing data and it is between two categorical variables. I started with an imputation with main effects only. impmethod Should I use formulas to set the interaction instead of modifying the predictor matrix? If yes, how do I specify which variables I want to include random slopes, etc? I understand that formulas overwrite the predictor matrix.
I have a multilevel dataset. My statistical model includes the following predictors: Age group: young vs. old (between-subjects factor) Day: first vs. second (within-subjects factor) Age group x Pulse My outcome has missing data, and it is continuous. No data is missing at the predictor level. The data is in long format I am using mice for imputation purposes. I understand that interaction terms need to be included in the imputation process. I do not understand how to code the interaction as it has no missing data and it is between two categorical variables. I started with an imputation with main effects only. impmethod Should I use formulas to set the interaction instead of modifying the predictor matrix? If yes, how do I specify which variables I want to include random slopes, etc? I understand that formulas overwrite the predictor matrix.
Full article content could not be extracted automatically. Read the original below.
Source:
Cross Validated
· stats.stackexchange.com