I don't think this is the only issue. There
is at least one more, that -mvis- sets
a characteristic, which must be set otherwise.
As mentioned more than once on this list recently,
Patrick Royston, the author of these commands,
is not a member of Statalist. That reduces (although
not to zero) the chance of getting an authoritative
answer from a posting to Statalist.
In this case, I forwarded the original question
to Patrick, who will no doubt reply in due course.
Nick
[email protected]
Fred Wolfe
> >A colleague of mine imputed data using SAS's mi command. The
> dataset,
> >containing m=5 imputations, I translated into a .dta format using
> >StatTransfer. I wondered if Stata's micombine command might
> be able to get
> >the regression coefficients. I created the two variables _i
> and _j that
> >normally would be present had we used Stata's mvis command
> instead of
> >SAS's mi command and ordered the dataset as mvis would have.
> >
> >I still get the following error message after using micombine.
> >
> >. micombine regress s_mh dum1grp dum2grp gesl lft cd4 vload,
> impid (_j)
> >no row-identifier variable found - data may have incorrect format
> >r(198);
> >
> >What format does Stata expect? Any help to sort this out
> will be greatly
> >appreciated.
>
>
> Here is an example of the format from a data set of mine that worked
> without problem.
>
> +---------------------+
> | _i _j sex |
> |---------------------|
> 1. | 1 1 Female |
> 2. | 1 2 Female |
> 3. | 1 3 Female |
> 4. | 1 4 Female |
> 5. | 1 5 Female |
> |---------------------|
> 6. | 2 1 Female |
> 7. | 2 2 Female |
> 8. | 2 3 Female |
> 9. | 2 4 Female |
> 10. | 2 5 Female |
> |---------------------|
>
> In the final data set there are 5 combined data sets (_j) and
> (_j=1 to 5).
> Within each data set, _i is the observation number.
*
* 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/
