diggerfoot
Humanity Hiker
- Joined
- Oct 1, 2011
- Messages
- 1,550
- Reaction Score
- 8,695
I cringe a little when I see the old "damn lies and statistics" chestnut, just as I do when I see statistics presented as conclusive evidence of something. Philosophies of science and knowledge were a big part of my academic training, along with my ongoing "research" into life. There are two important parts of information, though constructed different by different philosophies and (in my opinion) not the only parts to consider. These are validity and reliability.
Statistics provide reliable information, people can agree on what they mean, but what they mean may not be valid to what is being considered. Essentially every objective statistic has subjective criteria behind its choice. Should Stewart's unique blocks and assists statistic be valued more or less than Moore's blocks and steals statistics? Should Taurasi's number of clutch shots (neither Moore nor Stewart played in many close games, Stewart never played in a close game we won) be valued over other statistics? I think most people agree that you cannot compare Stewart v Jefferson based on statistics, no matter how reliable those statistics are.
I tend to favor validity over reliability, but there are pitfalls there as well. Taurasi did something the others did not, lead an inexperienced squad to a championship. This is an extremely valid criterion for comparison. Different players may have different roles in different systems for different teams that all affect the stats, but it's certainly valid to assume you want them to lead teams to championships regardless of their different situations. However, I'll be the first to admit that my criterion (yes, I favor Taurasi) is not reliable. We are talking about a sample size of one. Stewart never even had the chance to lead an inexperienced squad, while Moore only failed at it once. If you put each player into that type of situation ten times, maybe Taurasi only does it once and she happened to get lucky (I don't believe this, but you get my drift). Meanwhile, maybe Moore ends up leading an inexperienced team to a championship nine times out of ten but was unlucky that she only had one opportunity at it. My claim would be bolstered if there was the statistic born out of a decent sample size behind it.
Thus there should be a degree of humility behind either reliable or valid claims, since it is rare to have large measures of both. Just my two cents for now.
Statistics provide reliable information, people can agree on what they mean, but what they mean may not be valid to what is being considered. Essentially every objective statistic has subjective criteria behind its choice. Should Stewart's unique blocks and assists statistic be valued more or less than Moore's blocks and steals statistics? Should Taurasi's number of clutch shots (neither Moore nor Stewart played in many close games, Stewart never played in a close game we won) be valued over other statistics? I think most people agree that you cannot compare Stewart v Jefferson based on statistics, no matter how reliable those statistics are.
I tend to favor validity over reliability, but there are pitfalls there as well. Taurasi did something the others did not, lead an inexperienced squad to a championship. This is an extremely valid criterion for comparison. Different players may have different roles in different systems for different teams that all affect the stats, but it's certainly valid to assume you want them to lead teams to championships regardless of their different situations. However, I'll be the first to admit that my criterion (yes, I favor Taurasi) is not reliable. We are talking about a sample size of one. Stewart never even had the chance to lead an inexperienced squad, while Moore only failed at it once. If you put each player into that type of situation ten times, maybe Taurasi only does it once and she happened to get lucky (I don't believe this, but you get my drift). Meanwhile, maybe Moore ends up leading an inexperienced team to a championship nine times out of ten but was unlucky that she only had one opportunity at it. My claim would be bolstered if there was the statistic born out of a decent sample size behind it.
Thus there should be a degree of humility behind either reliable or valid claims, since it is rare to have large measures of both. Just my two cents for now.