quantile regression using qreg2 command with clustering

[Diya Bhandari]

Hi all,

I am currently running a quantile regression model using the command qreg2 to report median difference with 95% CI.

My outcome is continuous and predictor is binary (treatment 1 or treatment 2). The reason I am using quantile regression is because the outcome data is skewed and when running the Shapiro-Wilk test for normality after normal regression p was reported as <0.05.

My code adjusts for minimisation factors (age, multiple and centre) and a cluster to allow for the correlation between multiple births. I wanted to know if the way I've built this model logically and statistically makes sense, and/or if there is any other command that is a better fit for my data.

Code in Stata:
qreg2 outcome i.treatment i.multiple age i.centreID, cluster (MultipleClusterID) quantile(0.5)

Any advice is appreciated

[Joao Santos Silva]

Dear Diya Bhandari,

I do not know what you are doing, so I do not now if your model makes sense, but the command looks fine. Note, however, that the reasons you give for using quantile regression are not very convincing.

Best wishes,

Joao