ERROR ..... BINOMIAL LEVEL= value

[malls]

Hi,

I am trying to generate frequencies binomial CI for my response data using below code.

Sadly i'm getting errors when i have only one category(resp='0') in response variable for some categories of by variable.

Can anyone help me with this please ?

thanks

proc sort data=wis.liv OUT= wis.liv_freq; by lev cph ; run;

ods output Binomial=Binomial;

proc freq data = wis.liv_freq;

tables SURG / binomial(level=2);

by lev cph;

run;

ERROR: The BINOMIAL LEV= value is greater than the number of levs for variable SURG.

NOTE: The above message was for the following BY group:

      lev=1 cph=h2

ERROR: The BINOMIAL LEV= value is greater than the number of levs for variable SURG.

NOTE: The above message was for the following BY group:

      lev=1 cph=h3

ERROR: The BINOMIAL LEV= value is greater than the number of levs for variable SURG.

NOTE: The above message was for the following BY group:

      lev=2 cph=h4

ERROR: The BINOMIAL LEV= value is greater than the number of levs for variable SURG.

NOTE: The above message was for the following BY group:

      lev=2 cph=h7

ERROR: The BINOMIAL LEV= value is greater than the number of levs for variable SURG.

NOTE: The above message was for the following BY group:

      lev=3 cph=h12

ERROR: The BINOMIAL LEV= value is greater than the number of levs for variable SURG.

NOTE: The above message was for the following BY group:

      lev=3 cph=h13

ERROR: The BINOMIAL LEV= value is greater than the number of levs for variable SURG.

NOTE: The above message was for the following BY group:

      lev=3 cph=h19

ERROR: The BINOMIAL LEV= value is greater than the number of levs for variable SURG.

NOTE: The above message was for the following BY group:

      lev=3 cph=h22

ERROR: The BINOMIAL LEV= value is greater than the number of levs for variable SURG.

NOTE: The above message was for the following BY group:

      lev=3 cph=h23

ERROR: The BINOMIAL LEV= value is greater than the number of levs for variable SURG.

NOTE: The above message was for the following BY group:

      lev=3 cph=h24

ERROR: The BINOMIAL LEV= value is greater than the number of levs for variable SURG.

NOTE: The above message was for the following BY group:

      lev=3 cph=h25

ERROR: The BINOMIAL LEV= value is greater than the number of levs for variable SURG.

NOTE: The above message was for the following BY group:

      lev=3 cph=h27

ERROR: The BINOMIAL LEV= value is greater than the number of levs for variable SURG.

NOTE: The above message was for the following BY group:

      lev=3 cph=h29

ERROR: The BINOMIAL LEV= value is greater than the number of levs for variable SURG.

NOTE: The above message was for the following BY group:

      lev=4 cph=h39

ERROR: The BINOMIAL LEV= value is greater than the number of levs for variable SURG.

NOTE: The above message was for the following BY group:

      lev=4 cph=h51

ERROR: The BINOMIAL LEV= value is greater than the number of levs for variable SURG.

NOTE: The above message was for the following BY group:

      lev=4 cph=h58

NOTE: The SAS System stopped processing this step because of errors.

WARNING: The data set WORK.BINOMIAL may be incomplete.  When this step was stopped there were 16

         observations and 7 variables.

[PaigeMiller]

You can't compute anything from data where your variable SURG takes on only the value '0' and no other value.

[malls]

program works except for some combinations.

i'm trying to find a way for SAS to compute freq for valid combinations without using the WHERE CLAUSE 16 times 

[PaigeMiller]

You would have to weed out these levels where there is only one value in a data step (or PROC) before your run PROC FREQ.

[malls]

many thanks 

i'll try this 

[SAS_Rob · Accepted solution]

A suggestion would be to take advantage of the WEIGHT statement with the ZEROS option.  You could begin by taking your existing data set and assigning a count of 1.

data wis.liv_freq;

set wis.liv_freq;

ct=1;

run;

Once you do that, you could then create a data set that has all the combinations of the BY variables and the SURG variable and assign a count of 0 to each observation.

data all;

do lev=  *assign the levels of lev here;

do cph=...;

do surg=...;

ct=0;

output;

end;

end;

end;

Now if you stack those two data sets together and sort them

data wis.liv_freq2;

set wis.liv_freq all;

proc sort data=wis.liv_freq2;

by lev cph ;

run;

Then you should be able to use the WEIGHT statement in Proc FREQ.

proc freq data = wis.liv_freq2;

weight ct/zeros;

tables SURG / binomial(level=2);

by lev cph;

run;

[malls]

Both methods works perfectly 

thanks all