Balance Table and Randomization

[Blanca Jara]

Hi, I am working with data of a random experiment. How ever, in my case of study, in a moment I lose information about 247 individuals. How can it affect to my results of OLS? How can determinated with a Balance Table that the randomization is still working? (what type of information can I use in the balance table?)

Thanks for your answer.

B.

[Clyde Schechter]

I think it is more important for you to think about why and how you lost the information about the 247 participants. Even a complete balance table will never assure you that the remaining participants in the study arms are also comparable on unobserved attributes. So you need to try to figure out what kinds of bias on unobserved variables might have been introduced by the particular mechanism that led to the loss of data on the 247.

That said, you still need to do a balance table to supplement your hand-waving about the mechanism of data loss. You should include in the balance table all variables in your data set (other than study id numbers and comment-like text variables).