An old NFL coach who really didn't like me much used to revel in discounting my statistical analyses.

''Statistics,'' he would say, ''are for losers.''

He was fired shortly thereafter, and presumably has spent much of his free time learning his way around a calculator.

Of course, there was some truth in his statement. Statistics can be manipulated to strengthen just about any argument. Take USA Today, for example. The Nation's Newspaper recently published its annual list of NBA player ratings. The formula is to take the 29 starters (or regulars) at each position, rank them in order in every statistical category, and allocate points based on that positioning. Whoever winds up ranking the highest in the most categories - and therefore compiles the fewest points - wins.

Sounds good on paper, right? But the problem is this: none of the statistical categories are weighted by priority. In other words, rebounds are just as important to point guards as assists. Three-point percentage is just as important to centers as rebounds or blocked shots. Scoring means no more to a shooting guard or small forward then blocked shots.