www.cmm.bristol.ac.uk/forum

It's possible that you are encountering a bug that we have fixed in our development version since we last put a release on ssc. Could you please enter the command:
prior to the -runmlwin- command and let us know what the Stata output is around the type mismatch error?

Hi

I entered the "set trace on" command and again the previous cod. The output contain many things that not familiar for me;
like this
...
local cmdname `s(cmd)'
= local cmdname runmlwin
- local fullcmdname `s(cmdname)'
= local fullcmdname runmlwin
- local props : properties `cmdname'
= local props : properties runmlwin
- local ismi : list posof "mi" in props
- if (!`ismi' & "`cmdok'"=="") {
= if (!0 & ""=="") {
- di as err "{bf:mi estimate}: command not supported"
mi estimate: command not supported
- di as err "{p 4 4 2}{bf:`fullcmdname'} is not officially "
= di as err "{p 4 4 2}{bf:runmlwin} is not officially " ....

Looking back at your original syntax the reason for the error mi estimate: command not supported appears to be that you have a small typo in your -mi- command. Where you have written cmdoc it should be cmdok.

Could you try the command without the -mi- suffix after running so that I can attempt to determine the cause of the type mismatch error?

Hi
I entered the commans and got outpots as fallow:

set trace on
runmlwin V_outcome5 V_SEXID cons, level2( TheKey: cons) level1( V_id:) discrete(dist(multinomial) link(ologit) denom(cons) basecategory(3)) nopause

....
= if "" == "" & "" ~= "" & "" ~= "" local mlwinpath
- if "`mlwinpath'" == "" & "$MLwiNScript_path" ~= "" & "`batch'" ~= "" local mlwinpath $MLwiNScript_pat
> h
= if "" == "" & "" ~= "" & "" ~= "" local mlwinpath
- if "`mlwinpath'" == "" & "$MLwiN_path" ~= "" local mlwinpath $MLwiN_path
= if "" == "" & "C:\Program Files (x86)\MLwiN trial\i386"\mlwin.exe" ~= "" local mlwinpath C:\Program F
> iles (x86)\MLwiN trial\i386"\mlwin.exe
type mismatch


Bst wishes

It looks as if you have an extra quotation mark in your MLwiN_path global. To correctly set this for the version that you are using the command should be:

Hi dear Chris

I did what you said, but with following syntax I got out pouts as below and "variable cons not found
(error in option denominator())" error.


runmlwin V_outcome5 V_SEXID cons, level2( TheKey: cons) level1( V_id:) discrete(dist(multinomial) link(ologit) denom(cons) basecategory(3)) nopause
...
- marksample touse, novarlist
- if ("`discrete'"~="") {
= if ("dist(multinomial) link(ologit) denom(cons) basecategory(3)"~="") {
- local 0 , `discrete'
= local 0 , dist(multinomial) link(ologit) denom(cons) basecategory(3)
- syntax , Distribution(string) [Link(namelist min=1 max=1) DEnominator(varlist numeric) Extra Offset(v
> arname numeric) Proportion(varname) Basecategory(numlist integer min=1 max=1) MQL1 MQL2 PQL1 PQL2]
variable cons not found
(error in option denominator())
local validdistributions normal binomial poisson nbinomial multinomial
local checkdistribution :list distribution & validdistributions
if "`checkdistribution'"=="" {
display as error "Invalid distribution(). Valid distributions are: normal, binomial, poisson, nbinomi
> al, multinomial"
exit 198

Thanks for trying this. Can you also run the code that George provided earlier again to see whether you still have the same problem with running that?

yes,
unmlwin V_outcome5 cons, level1(V_id:) discrete(dist(multinomial) link(ologit) denom(cons) basecategory(3 )) nopause

= if ("dist(multinomial) link(ologit) denom(cons) basecategory(3 )"~="") {
- local 0 , `discrete'
= local 0 , dist(multinomial) link(ologit) denom(cons) basecategory(3 )
- syntax , Distribution(string) [Link(namelist min=1 max=1) DEnominator(varlist numeric) Extra Offset(v
> arname numeric) Proportion(varname) Basecategory(numlist integer min=1 max=1) MQL1 MQL2 PQL1 PQL2]
variable cons not found
(error in option denominator())
local validdistributions normal binomial poisson nbinomial multinomial
local checkdistribution :list distribution & validdistributions

As you are getting the same error there I'm not really sure what is going on. The message in question comes from Stata's -syntax- command (http://www.stata.com/help.cgi?syntax), which we simply call and suggests that Stata is unable to find a numeric variable called cons in your data. Can you post the output of:
so that we can check that there is nothing strange in your variable names or contents. If not then I can only suggest running:
to ensure that your version of Stata is up to date.

summariz V_outcome5 V_SEXID2

Variable | Obs Mean Std. Dev. Min Max
-------------+--------------------------------------------------------
V_outcome5 | 20819 1.090975 .3103937 1 3
V_SEXID2 | 20837 .9151509 .2786636 0 1