+ Reply to Thread
Results 1 to 4 of 4

Thread: Need critic and help on survival model

  1. #1
    Points: 9, Level: 1
    Level completed: 17%, Points required for next Level: 41

    Posts
    2
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Need critic and help on survival model




    Hey all,

    I am in need of some help or critic regarding my model for a research paper and i've lost my confidence about the regression by now...

    I am using cox hazard model to analyze firm termination that is dependent of various factors.

    A: censor: 0 firm not default (ongoing), 1 firm default
    B: t=time in days since startup

    Covaratates:
    Size (approx by sales, continue nr)
    industry check (whether it is service industry or not, 0-1 value)
    number of locations the firm has (cont..)
    risk index of country of the firm

    I use SPSS, as i never have learned to work with other programs

    I do the following for my regressions:
    A is the status, B is the time, the rest i put in the covariate box and just run regression. Is this correct?

    -Should i aditionally devide the groups that failed, eg do i need an extra "dependent factor" (like normally you test death/alive on drug=yes/no)
    -should i test every covariate seperately?

    Ive been advised to use this cox regression, however I am not complete certain I understand its use in my model whereas i would in the case of drug testing..

    Any comments would help me and i'd be very thankful!

  2. #2
    TS Contributor
    Points: 12,501, Level: 73
    Level completed: 13%, Points required for next Level: 349

    Posts
    951
    Thanks
    0
    Thanked 103 Times in 100 Posts

    Re: Need critic and help on survival model

    Quote Originally Posted by Nowak View Post
    -Should i aditionally devide the groups that failed, eg do i need an extra "dependent factor" (like normally you test death/alive on drug=yes/no)
    It depends on the data. eg if you find a variable that has a significant interaction effect. [eg drug use could affect other covariates like hours of sleep, or drug use could be tied up with age]

    Quote Originally Posted by Nowak View Post
    -should i test every covariate seperately?
    If it helps you understand your data better, you should do it. This test would be most likely for your own knowledge, not something you would put in a paper.

    Quote Originally Posted by Nowak View Post
    Ive been advised to use this cox regression, however I am not complete certain I understand its use in my model whereas i would in the case of drug testing..
    You're using the Cox regression because of the element of time. You don't need to specify the underlying hazard function. What other type of analysis would you do?
    All things are known because we want to believe in them.

  3. #3
    Points: 9, Level: 1
    Level completed: 17%, Points required for next Level: 41

    Posts
    2
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Need critic and help on survival model

    Normally, i use logistic regressions models which are more used in economics. In that case i would look at the predictor failure status and not so much at time.


    But since I do not really have a second underlying difference my fomula becauses:

    fail time = H (o)^t * exp (Bsize*firm1size+Bindustry*industryfirm 1........)

    Isn't by that case a cox regression with covariates in spss rather easy and straightforward?

    I do feel i have to check for some collinearity though of several factors, next challenge

  4. #4
    TS Contributor
    Points: 12,501, Level: 73
    Level completed: 13%, Points required for next Level: 349

    Posts
    951
    Thanks
    0
    Thanked 103 Times in 100 Posts

    Re: Need critic and help on survival model


    Quote Originally Posted by Nowak View Post
    But since I do not really have a second underlying difference my fomula becauses:

    fail time = H (o)^t * exp (Bsize*firm1size+Bindustry*industryfirm 1........)

    Isn't by that case a cox regression with covariates in spss rather easy and straightforward?
    I don't use SPSS. However there is an SPSS sub-forum here, where you could ask your question if you need to.

    I asked what other type of analysis you would do, because there are other types. Cox regression assumes proportional hazards. There are other analyses where you treat time more like a part of the outcome, rather than as an independent variable. Cox regression is very popular in this instance though.
    All things are known because we want to believe in them.

+ 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