Hello I am attempting to run a structural break Weibull duration model with two parameters for some social science (cabinet) duration data. I have a small number of observations (13) and 6 covariates (though my problem exists for any number of covariates I try). I can compile the model, load the data, and load/gen inits without issue. I cannot, however, avoid numerical overflow (undefined real result) traps (I cannot even obtain one update without trap). As per the winBUGS manual, I have tried: 1) using specified/reasonable priors (I obtained these from a slightly simpler model, with more observations, for a lower number of covariates), as in my inits below 2) standardisation of covariates to have mean 0 and standard deviation 1 (as in my data below) 3) put reasonable bounds on the distribution of Y (the duration) 3) using UpdaterSlice rather than DFreeARS My model, inits and data follow. Any help or advice is grately appreciated. Arthur model { for(i in 1:N){ Y[i] ~ dweib(r, mu[i])I(150,1900) log(mu[i])<-m[i]## regression structure for covariates m[i]<-alpha0 +alpha[1]*atts[i] + alpha[2]*maj[i] + alpha[3]*fract[i] + alpha[4]*pol[i] +alpha[5]*seats[i] +alpha[6]*votes[i] + J[i]*beta0 + J[i]*beta[1]*atts[i] + J[i]*beta[2]*maj[i] + J[i]*beta[3]*fract[i] + J[i]*beta[4]*pol[i] +J[i]*beta[5]*seats[i] +J[i]*beta[6]*votes[i] J[i] <- step(yr[i]-cp-eps); } for(i in 1:g){alpha[i] ~ dnorm(0,1.0E-6)} for(i in 1:g){beta[i] ~ dnorm(0,1.0E-6)} for(i in 1:T) { priort[i] <- punif[i]/sum(punif[]) } alpha0 ~ dnorm(0,1.0E-6) beta0 ~ dnorm(0,1.0E-6) cp ~ dcat(priort[]) r~dgamma(1,1.0E-3) eps<-0.5 } inits{ list(alpha=c(-19,9,-0.01,300,0,0), beta=c(14,-16,0.04,265,0,0), alpha0=18, beta0=-20, cp=10 ) } ##data ##all covariates have mean=0, sd=1 list(Y=c(1673, 604, 1307, 1595, 1832, 531, 1540, 1349, 219, 1610, 1497, 1463, 934), atts=c(-0.716435813139072, -0.301342083389757, -0.794265861574658, -1.20935957567396, -1.15747295288861, 0.217524978626755, - 0.768322561902083, -0.379172147384419, 2.00761678167354, 1.46280633367525, 0.269411861231133, 0.658562071238475, 0.710448969507407), maj=c(-0.716435822143307, -0.301342086325116, -0.794265923501635, -1.20935967496984, -1.15747286455923, 0.217525183617782, - 0.768322506576223, -0.379172172124362, 2.00761682583084, 1.46280622527176, 0.269411734209298, 0.658562273171539, 0.710448808098491), fract=c(-0.645030899468411, -0.261636394513209, -0.516382537217905, -0.777860722846884, -1.34066922032107, -0.463776103071283, - 0.88638719710475, -0.267924814410628, 1.83344004750904, 1.65120105592168, 0.637093921578761, -0.0898466717851229, 1.12777953572978), pol=c(1.58671725865179, -0.919619660823142, -0.697697870724915, 0.117234223980945, 0.825104341051577, -0.851776144092821, 0.754916942620654, -0.414667525585798, -1.50903175716309, -0.894441095902639, - 0.269628145402186, 1.49352949302410, 0.779359940365521), seats=c(0.45495277259305, 0.391879881604168, 0.605308920144101, 0.822155559975886, 1.31015637515749, 0.618054174378085, 0.85668034587467, 0.319084368420061, -1.59604245101827, -1.73772821556916, - 0.627773501089918, -0.117524511080664, -1.29920371938950), votes=c(-0.629533336732271, -0.272644920429723, -0.528175842000219, -0.783375286316831, -1.34539469287088, -0.477046851761567, - 0.885361776206008, -0.272476190766197, 1.78484981014816, 1.66957095628813, 0.647686868196159, -0.067302233736413, 1.15920349618766), N=13, T=13, yr=c(1,2,3,4,5,6,7,8,9,10,11,12,13), punif=c(1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1), g=6)) ------------------------------------------------------------------- This list is for discussion of modelling issues and the BUGS software. For help with crashes and error messages, first mail [log in to unmask] To mail the BUGS list, mail to [log in to unmask] Before mailing, please check the archive at www.jiscmail.ac.uk/lists/bugs.html Please do not mail attachments to the list. To leave the BUGS list, send LEAVE BUGS to [log in to unmask] If this fails, mail [log in to unmask], NOT the whole list