People really don't want to understand what Matt Pfeffer said instead of taking the face value of his conclusion. He did exactly what it was supposed to do if you want to 'learn' things from data perspective.
1- You have to have enough data. This is where he said at this point of his career (with his experience), Shea Webber data is enough to use to study as it. It means, he played enough years, had different partners, coaches, systems. It's not all possibilities but pretty good set of data to study from.
2- When you have enough data, you try to elaborate one set of pattern which are useful to you. In case of a player in a team sport, especially hockey where the team concept is much stronger than other sports, the logic pattern is the 'influence' on the game. The influence of each of the player presence on ice with 9 different players. What else you can learn really ?
People who laugh because they think he used +/- . Of course not, of course he would not do a presentation to the management group if he wasn't absolutely convinced of what he found. What he found using different approaches from his expertise field of data mining. He came to the conclusion that Webber presence on ice, after all those years of hockey with different set of 'unknown' parameters is neutral. It means that, within data realm, Shea Webber is 'average' in term of influence on the game. It's not positive, not negative.
Even solely use the +/- has a meaning in a long period. If a player is always in the - after ten years in the league. You can laugh all you want but he's probably not a good defensive player. Use one stat is always flawed especially if you don't possess enough data. When you have enough data, you crunch it intelligently then something will come out.
What Matt Pfeffer should have said that would make people swallow it easier is that Shea Webber data suggest that he's an average 1st or 2nd D. This will include the fact that he probably played more minutes vs 'harder' competition.
People who laugh at data study have absolutely no idea how powerful it is. Facebook and Google did a lot of research in this field with the best of researchers and the results are stunning or scary. One example, one of the study (Oxford) showed that if they collected around 70 'likes' that people use on their Facebook, they can trace a better personality picture of the person than his/her own mother. The experts, the I know better because I am the expert (the mom in this case), mentality is a human flaw because human always think they know it all. There're also countless example that 'experts' are absolutely wrong in their views but then they will say "error is human" to explain it all.
