To satisfy your curiosity further--I was presenting things here in a
format that would make sense in a typical Stata dataset. In fact my
variables weren't really "Age", "Sex" and "Weight" but more along the
lines of category identifiers. I was trying to get the data into the
format:
Individual_ID Category_ID Data_point
Then you could index by both Individual_ID and Category_ID in Access
and define relationships to another table with category level data for
example--which is a fairly standard approach in Access.
I was not unaware that--as presented--this was a pointless data
structure in Stata. I was just trying to keep things simple and not to
give too many unnecessary details. But I just ended up just being
puzzling!
Karin
2009/1/14 Nick Cox <[email protected]>:
> Thanks for satisfying my curiosity. I have never used Access but I am
> very surprised to hear that it requires this format. Be that as it may,
> my not so hidden point was that your new data structure is, for Stata
> purposes, immensely inferior to the previous one.
>
> Nick
> [email protected]
>
> K Jensen
>
> I was using Stata to prepare data for a prewritten program--not
> mine--in another package (MS Access) that expects the data in this
> format. I suspected--correctly!--that it would be easier to do this in
> Stata than in Access. Access would require either a silly number of
> UNION queries or a VBA programming loop that would take many more
> lines.
>
> I am still lost in admiration for the neatness of Ulrich's solution.
>
> 2009/1/13 Nick Cox <[email protected]>:
>
>> Ulrich gave a good answer to your question.
>>
>> Could you please explain why this data structure is any sense better
>> for analysis?
>>
>> K Jensen
>>
>> 2009/1/13 Ulrich Kohler <[email protected]>:
>>> Am Dienstag, den 13.01.2009, 16:49 +0100 schrieb Ulrich Kohler:
>>>> Am Dienstag, den 13.01.2009, 15:37 +0000 schrieb K Jensen:
>>>> > Hi
>>>> >
>>>> > I would like to move from the following format
>>>> > ID Age Weight Height
>>>> > 1 23 140 62
>>>> > 2 64 96 72
>>>> > 3 49 123 67
>>>> >
>>>> > To this format:
>>>> > ID Var_name Var_value
>>>> > 1 Age 23
>>>> > 1 Weight 140
>>>> > 1 Height 62
>>>> > 2 Age 64
>>>> > 2 Weight 96
>>>> > 2 Height 72
>>>> > 3 Age 49
>>>> > 3 Weight 123
>>>> > 3 Height 67
>>>> >
>>>> > How can I do this in Stata? None of reshape, stack or xpose are
>>>> > exactly what I need.
>>>> >
>>>> > In reality I have over 100 variables rather than three--so I would
>>>> > like to avoid having to list the variables by name if possible.
>>>>
>>>>
>>>> local i 1
>>>> foreach var of varlist Age-Height {
>>>> local lab`i' `var'
>>>> ren `var' var`i++'
>>>> }
>>>> reshape long var, i(ID) j(Var_name)
>>>> lab val Var_name names
>>>> forv j = 1/`--i' {
>>>> lab def names `j' "`lab`j''", modify
>>>> }
>>>
>>>
>>> Or even shorter by using a String-Variable for Var_name:
>>>
>>> foreach var of varlist Age-Height {
>>> ren `var' var`var'
>>> }
>>> reshape long var, i(ID) j(Var_name) string
>>
>> That's _so_ neat--thankyou Ulrich! Exactly what I needed.
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/