+ Reply to Thread
Results 1 to 7 of 7

Thread: Writing down firth logistic regression equation

  1. #1
    Points: 2,295, Level: 28
    Level completed: 97%, Points required for next Level: 5

    Posts
    14
    Thanks
    1
    Thanked 0 Times in 0 Posts

    Writing down firth logistic regression equation




    Hi,

    This is a really simple question that I'm sort of embarrassed I can't answer - how would you write down a firth logistic regression? Now to clarify, I know how to RUN one in Stata and I know how to interpret the coefficients, but how would you actually write one down, like in a paper? Not sure what the formula is and papers that use firth logistic regression is an extreme rarity in my field.

    Also I should mention that I'm having the output transformed into odds ratios.

  2. #2
    Omega Contributor
    Points: 38,334, Level: 100
    Level completed: 0%, Points required for next Level: 0
    hlsmith's Avatar
    Location
    Not Ames, IA
    Posts
    6,998
    Thanks
    398
    Thanked 1,186 Times in 1,147 Posts

    Re: Writing down firth logistic regression equation

    I think you are looking for an nice clean way to write out the model equation or formula for the odds ratio, which I don't think are available. It is a bias correction in the likelihood, so I think the two above mention equations remain the same, but the interworkings for coefficient calculation are slightly tweaked.


    I am a little like you, in that I know when to use it and I interpret results the same as before but document the use of the correction. The best that I can do is provide the link below and say perhaps some people (unsure?) add a subscript documenting its use (e.g., ORFirth: 2.4 95% CI: 1.2, 3.9), imagine I know Latex and the firth is a subscript.


    http://stats.stackexchange.com/quest...tic-regression


    PS, We can see if anyone else chimes in, and please update you post if you figure out an acceptable solution.
    Stop cowardice, ban guns!

  3. #3
    Points: 2,295, Level: 28
    Level completed: 97%, Points required for next Level: 5

    Posts
    14
    Thanks
    1
    Thanked 0 Times in 0 Posts

    Re: Writing down firth logistic regression equation

    Hi,

    So would it be OK to write logitFirth(Reward=1)= (and then the actual equation) with Firth obviously written as a subscript? Is that what you mean? Because that I can do :P

    Thanks for the help.

  4. #4
    Omega Contributor
    Points: 38,334, Level: 100
    Level completed: 0%, Points required for next Level: 0
    hlsmith's Avatar
    Location
    Not Ames, IA
    Posts
    6,998
    Thanks
    398
    Thanked 1,186 Times in 1,147 Posts

    Re: Writing down firth logistic regression equation

    Yeah, that would be my applied approach. Since the Firth approach is a little too complicated to paraphrase in a cute little package, at least in my opinion.


    "Actual equation" you mean, say - the model's form. So a bunch of betas with variable terms?
    Stop cowardice, ban guns!

  5. #5
    Points: 2,295, Level: 28
    Level completed: 97%, Points required for next Level: 5

    Posts
    14
    Thanks
    1
    Thanked 0 Times in 0 Posts

    Re: Writing down firth logistic regression equation

    Yeah, like
    logit(reward=1) = B1Treatment1 + B2Treatment2 + B3Age1823 etc (obviously I'd use Beta and subscripts for the numbers but you get the point). I did an experiment with two treatments.

  6. #6
    Omega Contributor
    Points: 38,334, Level: 100
    Level completed: 0%, Points required for next Level: 0
    hlsmith's Avatar
    Location
    Not Ames, IA
    Posts
    6,998
    Thanks
    398
    Thanked 1,186 Times in 1,147 Posts

    Re: Writing down firth logistic regression equation

    That is what I would do, as long as you don't need anything in addition.
    Stop cowardice, ban guns!

  7. #7
    Points: 2,295, Level: 28
    Level completed: 97%, Points required for next Level: 5

    Posts
    14
    Thanks
    1
    Thanked 0 Times in 0 Posts

    Re: Writing down firth logistic regression equation


    Yeah, it's just for the results section to show everyone that I used logistic regression with the firth penalization, the actual code that you'd punch in to replicate the results I have in the appendix.

+ Reply to Thread

           




Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats