Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

coding variables by dates?

0 views
Skip to first unread message

Su

unread,
Nov 13, 2009, 10:42:52 PM11/13/09
to
hi folks,

I have a dataset with about 3000 observations on 30 participants (they
responded to short questionnaires about 3-6x/day for 2 weeks). I'm
trying to classify the days as "binge-eating" or "non-binge-eating"
days but I can't figure out if there is a way to do this with syntax,
or if I will manually have to enter this info.

To simplify, let's say that the variables are: participant number,
date of record, and type of eating behavior. So I have 3-6 records for
each day per participant. Initially when I coded the info, I
mistakenly just identified whether or not the record was a binge-
record. However, I would like to generalize to that entire day of
records. Could you please let me know if you have any suggestions!

thanks so much!
SA

JKPeck

unread,
Nov 14, 2009, 9:18:37 AM11/14/09
to

Use AGGREGATE with the participant number as the break variable and
max of the binge dummy as the summary.
Choose to add the aggregated value to each record rather than creating
a new dataset. You can do all this from the Aggregate dialog box.

HTH,
Jon Peck

Art Kendall

unread,
Nov 14, 2009, 10:56:43 AM11/14/09
to
You may end up going back to the dialog box until you have sort of finalized what you want to do. But after that don't forget to exit the dialog box via <paste>.

Art
0 new messages