[R] Tobit analysis

Andrew Barnes A.Barnes at ed.sac.ac.uk
Wed Jul 16 17:26:08 CEST 2003

Having read previous correspondance on this topic, am I right in using a 
gaussian distribution for a tobit model, one article suggests a normal distribution?

Also, I want to censure at the upper bound, so, using the survival5 package I use:

survreg(Surv(y,y<c,type="right")~x) for a censored regression.

Could anybody who's had experience of this, confirm whether I'm in the ballpark?

Grateful thanks
Andrew Barnes
Dr Andrew Barnes
Agricultural Management Economist

Modelling and Strategy Group
Land Economy Research Department
Research Division
West Mains Road

Tel: 0131 535 4042    E.Mail:  A.Barnes at ed.sac.ac.uk
Fax: 0131 667 2601    Mobile:  07811 935 022

Website: http://www.sac.ac.uk/management/external/default.htm

More information about the R-help mailing list