I just figured something out about the sas random number functions which perhaps must others already know, but which cased me some amount of problems. Apparently each random number function (RANUNI, RANNOR, RANPOI, etc.) creates and uses its own seed value. Therefor, one must be very careful about using different random number functions in the same data step.

For example, in the following data step random Poisson variates, Y, are generated and assigned to on of two groups based on the value of a uniform variate. Both function calls are initialized with a negative seed so that the time of day is used as the initial seed. Since the initial seeds are the same and since the ranpoi function either calls the ranuni function or uses the same routine, in each iteration the two seed values are the same. Therefore high values of the Poisson variate occur with high values of the uniform variate, and low values occur with low values. The result of one run produced 5014 observations of Group A with a mean of 7.487, and 4989 observations of Group B with a mean of 12.442;

data ONE; do i=1 to 10000; Y = ranpoi(-1, 10); x = ranuni(-1); if x<.5 then group = 'A'; else group = 'B'; output; end;

The way around this is to use the random number call routines. But even here one has to be careful not to use different seed variable for the different routines (unless you specify an initial and different seed for each rather then using the time of day) or else the same results occurs. One must use a single seed variable for all random number call routines used within a data step.

Just thought this might be useful.

--------------------------------------------- R. Blake Weiler Curriculum in Ecology CB# 3275, 299 Wilson Hall Univ. of North Carolina Chapel Hill, NC 27599-3275

rbw.wilson@mhs.unc.edu 919/962-5930