0
$\begingroup$

I am doing critical appraisal of a paper. The study has missing covariate data. The missing data is listed but it does not state what categories the data is missing? Is it appropriate to impute missing covariate data or was the study correct in just listing the amount missing and continuing with analysis.

Independent variable (self-reported sleep duration) Dependent variable (Hypertension) Missing covariates include alcohol consumption, caffeine, smoking, CVD, insomnia symptoms, depressive symptoms.

$\endgroup$
1
  • 1
    $\begingroup$ Welcome to Cross Validated! What do you mean by "ignore"? - using complete cases only? And what was the analysis they continued with? - a multiple regression? I think more details are needed for more than a generic answer to be given here. $\endgroup$ Mar 22, 2017 at 14:41

1 Answer 1

1
$\begingroup$

If you data is missing at random (MAR), then the answer is possibly and you should be careful to adjust for the fact that you imputed the data (by using multiple imputation, say), which will inflate the variance adequately.

Otherwise, if you cannot assume that your covariates are missing at random, things can go wrong. Consider a simple case: heavy drinkers do not reply to your study and you have no patient with high alcohol consumption in your study. Suppose that alcohol affects sleep past a certain consumption level and bias your results. You would not detect this effect.

$\endgroup$
1
  • $\begingroup$ Thank you for your help! I basically need to figure out if it is missing at random and go from there. $\endgroup$ Mar 22, 2017 at 15:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.