The linear regression test, lregr, matches two lists of random fractions from zero to one. The first list is generated completely, before the second list is generated. This means that each list represents a different portion of the random number cycle.
The result is expected to have a slope as close to zero as possible, using the least squares method of calculation. The slope is then compared to the Student's T Distribution to see if its slope is subject to chance.
The coefficient of correlation usually has the same result as the linear regression in the student's t distribution. This happens even though the slope is different from the coefficient of correlation.
If the linear regression passes the student's t test, then the two lists of numbers are statistically independent, even though they are generated by the same random number generator.
When you choose the size of each list with the size parameter, the central limit theorem is in effect.
The syntax for lregr is:
lregr size
lregr 17
Table of Contents
Job and Record Level Definitions
Field Level Definitions
Datagen Syntax
Supporting Routines
FAQ
Copying