Proc lifetest question

[Niugg2010]

I used two method trying to get quartile estimates: one is using weight statement and another is not. I realize the results are a little differenct.  

Below is my code and results.

Can someone help me figure out why it is different, which result is correct, and how to make the results same. Thanks

data x1;
wgt=1;arm=1;
input subjid aval cnsr;
datalines;
1 3.2 0
2 3.7 0
3 6.4 0
4 7.2 0
5 7.3 1
6 9.1 0
7 9.1 1
8 9.2 0
9 11.4 0
10 13.6 0
;

proc lifetest data=x1 ;
time AVAL*CNSR (1);
strata arm/adjust=tukey;
run;
proc lifetest data=x1 ;
weight wgt;
time AVAL*CNSR (1);
strata arm/adjust=tukey;
run;

Below is the result without weight statement. 

Below is the result with weight statement. 

[OsoGris · Accepted solution]

Sometimes using WEIGHT=1 for all observations can produce an error. I would try the FREQ statement.

[Niugg2010]

Thanks. I tried FREQ statement and it produced the same results.

I using Weight statement, becuase sometime all group censor varlue is 1, if without weight the code will produce warnings:
WARNING: The likelihood ratio test for strata homogeneity is questionable since some strata have no events.
Like below example: if without Weight statment or Freq statement, both produce WARNING

if with Weight statement, there is no warning. but the result is different from others.

data x1;
wgt=1;
input subjid arm aval cnsr;
datalines;
1 1 3.2 0
2 1 3.7 0
3 1 6.4 0
4 1 7.2 0
5 1 7.3 1
6 1 9.1 0
7 1 9.1 1
8 1 9.2 0
9 1 11.4 0
10 1 13.6 0
5 2 2.3 1
6 2 3.1 1
7 2 6.1 1
8 2 5.2 1
9 2 12.4 1
10 2 11.6 1
;

proc lifetest data=x1 ;
time AVAL*CNSR (1);
strata arm/adjust=tukey;
run;
proc lifetest data=x1 ;
weight wgt;
time AVAL*CNSR (1);
strata arm/adjust=tukey;
run;

proc lifetest data=x1 ;
freq wgt;
time AVAL*CNSR (1);
strata arm/adjust=tukey;
run;

[Niugg2010]

By the way, can you please give me some details, in what conditions Weight statement may produce errors? Since I used Weight statement a lot, and nerver realize this issue.  Thanks