Date: Mon, 21 Apr 2008 14:21:04 -0400
Reply-To: Sigurd Hermansen <HERMANS1@WESTAT.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Sigurd Hermansen <HERMANS1@WESTAT.COM>
Subject: Re: What happened to my special missing values?
In-Reply-To: <OFF580544F.672BFF83-ON85257432.0060CFDD-85257432.0062D99B@ria.buffalo.edu>
Content-Type: text/plain; charset="windows-1255"
Joe:
The same FORMAT statement that works in a Data step also works similarly in a PROC FREQ.
data want;
missing U;
input x;
cards;
1
2
U
2
1
;
run;
PROC FORMAT;
VALUE fynunk 1 = 'Yes'
2 = 'No'
.U = 'Unknown';
RUN;
PROC FREQ DATA = want;
TABLES x / MISSPRINT;
format x fynunk.;
RUN;
SAS actually has more than one way to trigger printing of formatted values, but the FORMAT statement seems simplest.
S
-----Original Message-----
From: owner-sas-l@listserv.uga.edu [mailto:owner-sas-l@listserv.uga.edu] On Behalf Of hoffman@ria.buffalo.edu
Sent: Monday, April 21, 2008 1:59 PM
To: SAS-L@LISTSERV.UGA.EDU
Subject: What happened to my special missing values?
Dear SAS-L ers:
I am an experienced SPSS user but novice at SAS doing some data
analysis on a large survey data set, using SAS 9.1.3 for Windows. Many
variables in this data set had missing values that were coded using
numeric values like "9" to indicate "unknown" responses. I wanted to
preserve these as special missing values (SPSS would call them "user
missing" values), so I created special formats for them, re-coded numeric
missing values to SAS special missing values (e.g., change "9" to ".U"
which has a format that labels this value as "Unknown"). An example of my
code is shown below. Initially this worked fine (see first output sample),
but I later re-ran PROC FREQ and found that my special missing values had
apparently disappeared, and were all displaying as regular missings (see
second output sample). I reviewed my steps and found that I did one other
file operation in between these two PROC FREQ runs: I had sorted my data
set on some other variables and then replaced the data set with the sorted
data set. But why would that have caused my special missings (e.g., ".U")
to apparently disappear?
I'd like to get these special missings to display correctly on PROC FREQ
output again, but I don't want to have to re-do this formatting for many
variables in the data set.
Thanks for any ideas.
Sincerely,
Joe Hoffman
LIBNAME library 'C:\SDATA';
OPTIONS FMTSEARCH=(library);
PROC FORMAT LIBRARY=library;
VALUE fynunk 1 = 'Yes'
2 = 'No'
.U = 'Unknown';
RUN;
DATA SDATA.want;
SET SDATA.have;
if s4aq1 = 9 then s4aq1 = .U; /* Unknown */
LABEL s4aq1 = 'Ever had 2-week period when felt sad, blue,
depressed, or down most of time';
FORMAT s4aq1 fynunk.;
RUN;
PROC FREQ DATA = SDATA.want;
TABLES s4aq1 / MISSPRINT;
RUN;
The SAS System 13:25
Monday, April 21, 2008 1
The FREQ Procedure
Ever had 2-week period when felt sad, blue, depressed, or down
most of time
Cumulative
Cumulative
S4AQ1 Frequency Percent Frequency Percent
ャャャャャャャャャャャャャャャャャャャャャャャャャャャャャャ
Unknown 892 . . .
Yes 12785 30.30 12785 30.30
No 29416 69.70 42201 100.00
Frequency Missing = 892
Later, after sorting the data set and re-saving it, I get this:
The SAS System
The FREQ Procedure
Ever had 2-week period when felt sad, blue, depressed, or down
most of time
Cumulative
Cumulative
S4AQ1 Frequency Percent Frequency Percent
ャャャャャャャャャャャャャャャャャャャャャャャャャャャャャャ
. 892 . . .
Yes 12785 30.30 12785 30.30
No 29416 69.70 42201 100.00
Frequency Missing = 892
===================================
Joe Hoffman
Data Analyst
Research Institute on Addictions
State University of New York at Buffalo
1021 Main Street
Buffalo NY 14203
phone (716) 887-2219
FAX (716) 887-2510
e-mail hoffman@ria.buffalo.edu ===================================
