Dear Statalist,
I used dtobit (*! version 1.0.2 03aug2004) and dtobit2 to compute
marginal effects after tobit and encountered a bug in this version of
dtobit when using the option -censor-. When I set trace on, I get the
follwing error:
-----------------------------------------------------------------------------
begin dtobit.Display1 ---
- args bm vm title level censor discret
- local c colsof(`bm')
= local c colsof(__00000L)
- if `level'<10 | `level'>99 {
= if 95<10 | 95>99 {
local level 95
}
- tempname Z xmat rate
- if "`censor'" == "" {
= if "" == "" {
- mat `xmat' = e(at)
= mat __00000O = e(at)
matrix has missing values
}
-------------------------------------------------------------------------------
end dtobit.Display1 ---
after running (note that I changed all lines in dtobit.ado and dtobit2.ado
with _b[_se] to _b[/sigma] to have it run in version 9.2):
*
sysuse auto, clear
replace mpg = 20 if mpg <= 20
tobit mpg length weight, ll
dtobit, censor
dtobit2 mpg length weight, ll(20)
*
it stops with dtobit. But when I add the option -brief- everything works:
*
sysuse auto, clear
replace mpg = 20 if mpg <= 20
tobit mpg length weight, ll
dtobit, brief censor
dtobit2 mpg length weight, ll(20)
*
It happens when the program -display1- is called. In the presumably wrong
version there was:
177 Display1 `B' `V' "`title'" "`level'"
178 local title "Unconditional Expected Value"
Display1 `dfdx_u' `V_u' "`title'" "`level'" "`censor'" /*
*/ "`discret'"
I changed that and inserted "`censor'" and "`discret'" in line 177/178:
177 Display1 `B' `V' "`title'" "`level'" "`censor'" /*
178 */ "`discret'"
179 local title "Unconditional Expected Value"
Display1 `dfdx_u' `V_u' "`title'" "`level'" "`censor'" /*
*/ "`discret'"
That seems to fix it - do you agree or did I miss something? In any case I
have not fully understand why this error occurs. Any idea?
Johannes
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/