Date:         Mon, 17 Jul 2006 15:13:41 -0400
Reply-To:     "Gerstle, John (CDC/NCHSTP/DHAP) (CTR)" <yzg9@CDC.GOV>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "Gerstle, John (CDC/NCHSTP/DHAP) (CTR)" <yzg9@CDC.GOV>
Subject:      Re: Making really nice just about ready to go tables
Comments: To: Jennifer Sabatier <plessthanpointohfive@GMAIL.COM>
Content-Type: text/plain; charset="US-ASCII"

Jennifer,
This is a quick answer to your question. It can be done easily, with
some work on the frontend.  Once you have a template, you can re-use it
with minimal modifications (using macro code as needed, especially with
titles).

I'm going to skip the data step code and show some ODS and PROC
REPORT...

First, I'd run a handful of PROC FREQS, output the resulting tables
using ODS OUTPUT, then piece this dataset together.

i.e.:
ods output onewayfreq=demog;
proc freq data=foo; table (gender age education)*group / missing; run;
ods output close;

...if group is your Control and Experimental groups. (You'll need to
transpose the frequency or percents values by group to get the dataset
below.)

Next, build the dataset you want to report out - something like so
(suggested variable names in the first row - look at the ODS dataset via
PROC PRINT to fix this the way you want it...):

Varname Varvalue        control_pct control_num_denom exp_pct
exp_num_denom
Gender          Male         52.6               (312/593)
54.4            (331/608)
Gender          Female       47.4       (281/593)       45.6
(277/608)
Age             < 15 years    8.2       (48/588)        7.5
(45/600)
Age             15 - 45 years 79.3      (466/588)       76.7
(460/600)
Age             > 45 years    12.6      (74/588)        15.8
(95/600)
Etc...

Assign the formatted value to Varvalue, or define a rather complicated
format to cover all your demographic values.
I'd create a character variable, control_num_denom, that concatenates
the parentheses, numerator and denominator values for each demographic
variable value.

Then the untested, simple (skips a lot of the bells and whistles) PROC
REPORT code, with ODS...

Ods html file="<path>\Table 1.xls" stle=journal;
Title "Table 1.  Demographic characteristics (n = 1205).";
Proc report data=builtabove nowdw ;
        Columns (" " Varname  Varvalue)
                  ("Control/(n = 594)"  control_pct control_num_denom)
                  ("Experimental Group/(n = 611)"        exp_pct
exp_num_denom);
        Define varname / " " group ;
        Define varvalue / " " ;
        Define control_pct / "%" format=8.1;
        Define control_num_denom / "(frequency/total)";
        Define exp_pct / "%" format=8.1;
        Define exp_num_denom / "(frequency/total)";
        Break after varname / skip;
Run;
Ods html close;

Look up the vast plethora of PROC REPORT, TABULATE, with ODS papers at
www.lexjansen.com


John Gerstle, MS
Biostatistician
Northrop Grumman
CDC Information Technological Support Contract (CITS)
NCHSTP \DHAP \HICSB \Research, Analysis, and Evaluation Section
Centers for Disease Control and Prevention

"As far as the laws of mathematics refer to reality, they are not
certain; and as far as they are certain, they do not refer to reality."

                  Albert Einstein, addressing the Prussian Acadamy of
Science, Berlin , Jan 27, 1921

"Boss. We've got cats."  "Meow"
-----Original Message-----
From: owner-sas-l@listserv.uga.edu [mailto:owner-sas-l@listserv.uga.edu]
On Behalf Of Jennifer Sabatier
Sent: Monday, July 17, 2006 2:36 PM
To: SAS-L@LISTSERV.UGA.EDU
Cc: Jennifer Sabatier
Subject: Making really nice just about ready to go tables

I know that this is just about impossible, getting tables out of SAS
that
look just about ready for submission to a journal.  But, perhaps I can
get
close.

I'm familiar, though not expert with using PROC REPORT and PROC
TABULATE.  I
know some of this can be done using a data step and put statements and
that's where I am really inexperienced.

Here's what I want my table to look like (a sample from a table I made
after
doing a lot of goofing around in EXCEL and WORD, taking 4 times the
amount
of time I'm sure it should).  I hope it posts right!

Table 1.  Demographic characteristics (n = 1205).
                       Control         Experimental Group
                     (n = 594)            (n = 611)
Gender                   %     (frequency/total)  %
(frequency/total)
     Male          52.6% (312/593)    54.4% (331/608)
     Female          47.4% (281/593)    45.6% (277/608)
Age
     < 15 years           8.2% (48/588)     7.5% (45/600)
     15 - 45 years  79.3% (466/588)    76.7% (460/600)
     > 45 years          12.6% (74/588)    15.8% (95/600)
Education
     None          57.1% (335/587)    53.4% (316/592)
     Grades 1 - 3   4.6% (27/587)     7.9% (47/592)
     Grades 4 and 5   4.3% (25/587)     4.7% (28/592)
     Grade 6          34.1% (200/587)    34.0% (201/592)





The variables are coded, ie, Gender = 1 or 2, Age = 1, 2, or 3, etc.
Missing variables are coded as . and are excluded from the calculation
of
percent (as you can see, the percentages come from the denominator of
what
data was available).  I specifically want to include the correct
denominator, because that's how I am representing the missingness.  The
total n for each group is listed in the column label.

What can I use, data stepping, any proc, whatever, to get a table in
this
format, with percentage signs, the parentheticals, then slash for the
fraction?

If anyone has any code that comes close and I can modify it would save
me
HOURS of time that I could spend on actual analysis of other projects,
instead of goofing around with tables.

And if you think I'm just a whiner who needs to realize that spending
HOURS
making tables for publication is just how it has to be, feel free to
tell me
that to.  At least then I can move on to the next shortcut hunt...

Thanks,

Jen