+ Reply to Thread
Results 1 to 3 of 3

Thread: multivariate ordered probit regression---use standard bivariate normal distribution?

  1. #1
    Points: 342, Level: 6
    Level completed: 84%, Points required for next Level: 8

    Posts
    9
    Thanks
    0
    Thanked 0 Times in 0 Posts

    multivariate ordered probit regression---use standard bivariate normal distribution?




    Hello.

    My two previous posts (1 & 2) have about 100 views each, but no replies, so I must have done a poor job asking the question. So, here is the general question: how can I run a run a multivariate (more than one non-independent, response variables) ordered probit regression model? I've had success doing this in the univariate case using the vglm() function in the VGAM package. For example:

    unvar.prob<-vglm(y~x,cumulative(link="probit",parallel=FALSE,reverse=TRUE))

    This would fit a unrestricted cumulative probit model (or a "thresholds of change" model) with the ordered response variable treated as a latent trait modeled with a standard normal distribution. What I'd like to do is include an additional, ordinal response variable, so that the two (non-independent) response variables are treated as latent traits and modeled with a standard bivariate normal distribution.

    The VGAM package includes the function binormal(), which seems like it should be of some use here, but if so, I don't know where/how it fits into the model. I've included the results of running the model with the two response variables separately. Any suggestions are welcome, and I've attached the data file, if that helps. Thanks.

    --Trey

    Code: 
    unvar.prob1<-vglm(pube3~age,data=refdata,cumulative(link="probit",parallel=FALSE,reverse=TRUE))
    > unvar.prob1
    Call:
    vglm(formula = pube3 ~ age, family = cumulative(link = "probit",
    parallel = FALSE, reverse = TRUE), data = refdata)

    Coefficients:
    (Intercept):1 (Intercept):2 age:1 age:2
    -1.65895567 -2.14755951 0.06688242 0.04055919

    Degrees of Freedom: 1492 Total; 1488 Residual
    Residual Deviance: 1188.909
    Log-likelihood: -594.4543

    Code: 
    unvar.prob2<-vglm(auric4~age,data=refdata,cumulative(link="probit",parallel=FALSE,reverse=TRUE))
    > unvar.prob2
    Call:
    vglm(formula = auric4 ~ age, family = cumulative(link = "probit",
    parallel = FALSE, reverse = TRUE), data = refdata)

    Coefficients:
    (Intercept):1 (Intercept):2 (Intercept):3 age:1 age:2
    -2.07719235 -2.43422370 -2.99123098 0.07319632 0.05133132
    age:3
    0.03797696

    Degrees of Freedom: 2238 Total; 2232 Residual
    Residual Deviance: 1583.47
    Log-likelihood: -791.7348

    *************
    Trey Batey--Anthropology Instructor
    Division of Social Sciences
    Mt. Hood Community College
    Gresham, OR 97030
    Attached Files

  2. #2
    Devorador de queso
    Points: 95,540, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Awards:
    Posting AwardCommunity AwardDiscussion EnderFrequent Poster
    Dason's Avatar
    Location
    Tampa, FL
    Posts
    12,930
    Thanks
    307
    Thanked 2,629 Times in 2,245 Posts

    Re: multivariate ordered probit regression---use standard bivariate normal distributi

    Wish I could help - I'm not too well versed in many multivariate methods though and a multivariate ordered probit isn't something I've had the pleasure to attempt to do.
    I don't have emotions and sometimes that makes me very sad.

  3. #3
    Points: 342, Level: 6
    Level completed: 84%, Points required for next Level: 8

    Posts
    9
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: multivariate ordered probit regression---use standard bivariate normal distributi


    Hi, Dason.

    I appreciate your taking the time to reply. It's no big thing, if you can't help; it's not seeing any replies that is a killer! Hopefully, someone will come across the original post and have some suggestions. Thanks, again.

    --Trey

+ Reply to Thread

           




Tags for this 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