This is an old topic, but I was just revisiting it. I'm a bit of a statistician, and the data presented would not be enough to figure out how random the NEWID() function is. You would expect each number to be selected about the same frequency as all others, but if you were trying to figure out if it was random this way only, a non-random loop would appear to be random because you would end up with exactly the same frequency for each value.
You need some method of looking at the ordering of values. For example, how often a number is preceded by exactly the same the number.
Article: Using NEWID to Randomly Sort Records
Posted - 04/16/2002 : 15:42:42
