I figured this out - I needed to remove the COMPUTE Pdate = 0. Thanks to David for directing me to the right places to figure this out.
DG
From:
Doug Harden <[hidden email]>
To:
[hidden email]
Date:
04/17/2013 04:44 PM
Subject:
Re: Identify
Break in Service Use
Sent by:
"SPSSX(r)
Discussion" <[hidden email]>
Thanks David,
I went on the list archives. I found the
syntax and adapted it to my variables. But I have a follow up question.The
syntax returns a very large and unusable number in the first instance of
the date diff comparison. See the first instance of diff is 155307. How
do I rewrite the syntax so it doesn't do that? The rest of the values makes
sense. Anything greater than 31 is a value I can note as a noncontinuous
value and flag it.
Here is my syntax
COMPUTE Pdate = 0.
SORT CASES BY USEID Sdate (A).
IF (USEID = lag(USEID,1)) PDate=lag(StartDate,1).
FORMATS PDate (DATE11).
COMPUTE diff=Datediff(Sdate,PDate,"Days").
Here is the outcome
| USEID
| TYPE
| Sdate
| diff
|
|
4567
| 4E
|
1/1/2008
|
155307
|
|
4567
| 4E
|
2/1/2008
|
31
|
|
4567
| 4E
|
3/1/2008
|
29
|
|
4567
| 4E
|
4/1/2008
|
31
|
|
4567
| 4E
|
5/1/2008
|
30
|
|
4567
| 4E
|
7/1/2009
|
426
|
|
4567
| 4E
|
8/1/2009
|
31
|
|
4567
| 4E
|
9/1/2009
|
31 |
Locked
