On Thu, 28 Jun 2007 06:14:18 -0700, SK <skauchali@GMAIL.COM> wrote:

>Hi there, this is a newbie-type question. > >I have a longitudinal dataset of daily feeding records of infants from >birth to 270 days (9 months). Infant feeding (category) for the each >day is recorded at either exclusive breast =1), mixed feeding (=2), >and exclusive formula (=3). I wish to count the proportion of days a >child receives either 1, 2 or 3. > >Data structure and intended variable after manipulation (propfeed): > >id Feed >1 1 >1 1 >1 2 >1 3 >2 1 >2 1 >2 1 >2 1 >3 3 >3 3 >3 3 >etc > > >Desired table: >id propfeed1 propfeed2 propfeed3 >1 0.5 0.25 0.25 >2 1.0 0 0 >3 0 0 1.0 >etc. > >Could someone help guide me through this problem? >thank you, > >SK

Is the purpose here to generate a report, or a data set to be used in subsequent processing?

For a report, consider PROC TABULATE. Your data are well organized and would seem to need no pre-processing.

Load:

data have; input id Feed ; cards; 1 1 1 1 1 2 1 3 2 1 2 1 2 1 2 1 3 3 3 3 3 3 ;

Create the report:

proc tabulate data=have formchar="|-+-+|+|+-+"; class id feed; table pctn<feed> , id , feed='PropFeed' ; run;

Output:

PctN +-------------------------------------------------------------+ | | PropFeed | | |--------------------------------------| | | 1 | 2 | 3 | |----------------------+------------+------------+------------| |id | | | | |----------------------| | | | |1 | 50.00| 25.00| 25.00| |----------------------+------------+------------+------------| |2 | 100.00| .| .| |----------------------+------------+------------+------------| |3 | .| .| 100.00| +-------------------------------------------------------------+