different parameter estimates from vague JAGS model and OLS

2013-02-05
2013-02-06
  • Andrew Rogers

    Andrew Rogers - 2013-02-05

    When running WinBUGS I tended to get virtually identical regression parameter estimates for a completely vague WinBugs model as an OLS model. Should this also be true within JAGS? If so, is there a common reason and/or any suggestions why I am not seeing this?
    The basic model is

    model {
    for (i in 1:100) {
    vol[i] ~ dnorm(p[i], tau)
    p[i] <-
    beta02 * dist[i] +
    beta03 pos[i] +
    beta04
    price[i] +
    beta05 tv[i] +
    beta06
    spon[i] +
    beta07 rain[i] +
    beta08
    plus100[i] +
    beta09 pricess[i] +
    beta10
    pricet[i] +
    beta11 pricec[i] +
    beta12
    pricecarl[i] +
    beta13 xmas[i] +
    beta14
    prexmas[i] +
    beta15 olympics[i] +
    beta16
    eurocup[i]
    }

    beta02 ~ dnorm(0, 0.000001)
    beta03 ~ dnorm(0, 0.000001)
    beta04 ~ dnorm(0, 0.000001)
    beta05 ~ dnorm(0, 0.000001)
    beta06 ~ dnorm(0, 0.000001)
    beta07 ~ dnorm(0, 0.000001)
    beta08 ~ dnorm(0, 0.000001)
    beta09 ~ dnorm(0, 0.000001)
    beta10 ~ dnorm(0, 0.000001)
    beta11 ~ dnorm(0, 0.000001)
    beta12 ~ dnorm(0, 0.000001)
    beta13 ~ dnorm(0, 0.000001)
    beta14 ~ dnorm(0, 0.000001)
    beta15 ~ dnorm(0, 0.000001)
    beta16 ~ dnorm(0, 0.000001)

    tau ~ dgamma(0.05, 0.05)
    sigma <- 1/sqrt(tau)
    }

    n.chains = 2
    
     
  • Martyn Plummer

    Martyn Plummer - 2013-02-06

    That is not a syntactically valid model. It is also not reproducible, as nobody else has access to your data.

    If you want help, then please send a reproducible example (data + model + scripts) with a summary of the results you get from OLS and from JAGS.

     

Log in to post a comment.

Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:

JavaScript is required for this form.





No, thanks