Date:         Thu, 24 Feb 2011 11:46:40 -0600
Reply-To:     "Pashi, Arthur" <Arthur.Pashi@DHSS.MO.GOV>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "Pashi, Arthur" <Arthur.Pashi@DHSS.MO.GOV>
Subject:      Calculating Mean Days
Content-Type: text/plain; charset="us-ascii"

Dear SAS-L users,

My survey data set looks like this:

Obs    asthma    physhlth  menthlth  actlimit

1    2-No         0        0      2-No
2    1-Yes       30        6      1-Yes
3    2-No        15        0      1-Yes
4    2-No         0       10      2-No
5    2-No         2       20      1-Yes
6    2-No         5        1      2-No
7    2-No         .        0      2-No
8    2-No        14        0      2-No

 Note: asthma = has asthma (1-Yes), and no asthma (2-No)
       physhlth = number of days with unhealthy physical health in past 30 days
       menthlth = number of days with unhealthy mental health in past 30 days
       actlimit = frequency (days) of activity limitation

I want to calculate the following:


1.  Given asthma equal 1-yes, compute



a.  Mean physically unhealthy days

b.  Mean mentally unhealthy days

c.  Mean physically or mentally unhealthy days

d.  Mean days of activity limitation


2.  Given asthma equal 2-No, compute



e.  Mean physically unhealthy days

f.  Mean mentally unhealthy days

g.  Mean physically or mentally unhealthy days

h.  Mean days of activity limitation

I have used proc means as follows:

Data =  ;
  Set    ;
  if 1 le physhlth le 30 then phdnoasth = 1;
  else if physhlth = 0 then phdnoasth = 0;
  label phdnoasth= 'Mean physically unhealthy days in past 30';
run;

proc surveymeans data = ast.asthma07 mean clm;
  strata vstrtum ;
  weight finl_wt;
  var phdnoasth mentdnoasth phmtdnoasth actlimnoasth;
  domain asthma;
run;

Instead of mean numbers of days, my results are rather percentages.

Any help is welcome.

Thank you

Arthur K. Pashi