questions about logistic regression with missing data in covariates

[Laura Lombardi]

Hello everybody !

Enjoy to meet you !

I have a little question concerning logistic regression in plink 1.9.In my dataset, I have some individuals that have missing data for age and I wondered how plink handle this issues in logistic regression ? Software eliminates the individual or not ?

Because, when I see the p-values of my association is quite similar, but not identical...and I lost a signal... So, I want to know how PLINK handle missing data in logistic and linear regression and if it is better or not to integrate age as covariate in my model...

Thanks a lot for your help !

Best,

Laura 

[Christopher Chang]

Samples with a missing covariate value are excluded from PLINK's regression analyses.  So, yes, if there's too much missing data, you are probably better off excluding the covariate unless you have a good-enough way of imputing missing values.