Changing probability cutoff in binary logistic regression

[So Ri Park]

Hello everyone?

I am running binary logistic regression. After running it, I used the lstat command and found out that my sensitivity rate is too low. Using the lsens command, I figured out that probability cutoff should be around .25.

My question is: I would like to run logistic regression with probability cut off at .25 instead of traditional .50, but I do not know how to do this. Can please any advise me of how to change probability cutoff in logistic regression by any chance?

Thank you very much!

[Clyde Schechter]

Unless you are running very ancient version of Stata (before version 9), the -lstat- command is obsolete. Its function has been replaced by -estat classification-. This newer command has a -cutoff()- option for which you can specify whatever cutoff you want. So, instead of -lstat-, run -estat classification, cutoff(0.25)-.

[So Ri Park]

Hi Clyde?

Thank you very much for your reply. I am also aware of -estat classification, cutoff(0.25) command, but as far as I understand, it just shows sensitivity/specificity rates if the cutoff level were .25. As far as I understand, it does not mean that the probability cutoff that is applied to regression analysis has changed to .25, does it? What I want to do is actually run logistic regression with probability cut off changing from traditional .5 to .25.

Can you please help me on this matter by any chance?

Thank you trillion!

[Clyde Schechter]

the probability cutoff that is applied to regression analysis has changed to .25,

This does not make sense. Logistic regression analysis itself has no cutoffs of any kind. It estimates a probability, as a continuous variable, between 0 and 1, for each outcome. Any cutoffs are the result of later commands such as -estat classification-.