Why use bivariate and then multivariate to test variables with P value of 0.10?

[runnyeggsham]

Hi guys, I am confused by a paper that is using bivariate analysis to variables and if the P value is 0.10 or less, they use those variables in a multivariate analysis.

I only took intro biostats and I dont think we even covered bivariate/multivariate.

So this is what I learned from the net and guessing. Please correct me if I am wrong.
Bivariate is a special case of multivariate.
Multivariate is used to remove confounders (variables).
P values are usually set at </= to 0.05 but the authors here arbitrarily chose 0.10.

They used bivariate to test which variables are significant so that they can have a more robust model when running only those significant variables in the multivariate model?

I guess I just don't see why they need the redundancy.

[Supun]

Hi guys, I am confused by a paper that is using bivariate analysis to variables and if the P value is 0.10 or less, they use those variables in a multivariate analysis.

I only took intro biostats and I dont think we even covered bivariate/multivariate.

So this is what I learned from the net and guessing. Please correct me if I am wrong.
Bivariate is a special case of multivariate.
Multivariate is used to remove confounders (variables).
P values are usually set at </= to 0.05 but the authors here arbitrarily chose 0.10.

They used bivariate to test which variables are significant so that they can have a more robust model when running only those significant variables in the multivariate model

I guess I just don't see why they need the redundancy.

Well, before using the multivariate statistics you have to use bivariate stat to screen your variables, the significant value to use when you are doing it is usually higer than your normal 0.05 values some even use more than 0.10, yes it is true bivariate is a special case of multivariate including 2 variables, how ever if your dont screen your variables your multivariate regression model becomes cumbersome.

Supun