# Does cox proportional hazard model control for type 1 error?

#### hehe1223

##### New Member
I am a greenhorn to survival analysis.
Just wondering if there are many covariates, the likelihood that one of them is significant should be quite high.

#### Mean Joe

##### TS Contributor
If you use 20 covariates, then you have to be wary that you could be finding one significant covariate just by random luck or something. The way I've seen it done, is if you find 3 significant covariates where 1 of them is "out of the blue" ie no one else has reported it, then you'd have trouble convincing people since you have so many covariates. But if they have all been previously reported as well, then it's not really questioned so much.