Exact logistic regression continous variable

gretaolsson — Wed, 10 May 2017 05:17:17 GMT

Hi,

I want to execute Exact logistic regression on a sample with a continuous variable. The sample size is 100 and below I show an extract from the data:

"","num","y","x"
"1",1,1,0.242106882663965
"2",1,0,-1.0215795935404
"3",1,1,-0.166510155016807
"4",1,0,-0.979868182931502
"5",1,1,0.078228780249738
"6",1,0,-0.37946718709135
"7",1,1,2.22519466020747
"8",1,0,0.0303031993682403
"9",1,0,-1.82711480734735
"10",1,0,1.1355902983012

I use the following code:

proc import out = filename
          file = "/folders/myfolders/sasuser.v94/Btest.csv"
          dbms = CSV
          replace;
run;

data want;
  set filename;
  sampleid=ceil(_n_/100);  /* Assign sampleid as record blocks of 10 */  
run;


proc logistic data=want desc;
  by sampleid;
  model y = x;
  exact x / estimate = both;
run;

(The beginning of the code I have because later on I want to have a csv file with several samples to be analyzed.)

And I get the following message:

An exception was thrown while sending a packet to the peer.
What am I doing wrong?

Re: Exact logistic regression continous variable

Ksharp — Wed, 10 May 2017 12:22:19 GMT

Try EXACTONLY option.

proc logistic data=want exactonly desc;
  by sampleid;
  model y = x;
  exact x / estimate = both;
run;

topic Exact logistic regression continous variable in Statistical Procedures

Exact logistic regression continous variable

Re: Exact logistic regression continous variable