Originally Posted by scott91575
and for his career Howard is .987 OPS w/RISP and .951 without RISP. Not a huge difference, and over the next few years it will probably get closer as the sample sizes increase. That is why RISP is not used in Sabermetric statistics. RISP stats trend towards a players overall stats, and the overall stats have a larger sample size. So why use a smaller sample size when you have a larger one? No stats are perfect, but there is a reason RISP is ignored in Sabermetric stats. The sample sizes each year are small in comparison to the overall stats, and as historical stats have proven over a long enough time RISP stats are very similar to overall stats.
WAR is not a perfect stat, but way better than batting w/RISP for part of a year.
Howard is actually only a .879 OPS with bases empty. He has a .951 OPS overall. It's pretty hard to say based on the size of the sample, about 1750 AB's with the bases empty and 1800 PA's with runners on base and 1100 with RISP, that Howard's numbers with RISP will decrease towards his numbers with nobody on base.
Dunn has an even larger sample size himself. Using RISP numbers inside of 1 season is not useful because the sample size is usually relatively small, but over the course of someone's career the sample sizes are plenty large to realize some noticable trends.
I wasn't advocating that a slash line w/RISP is more important or better than WAR. I was just using Dunn and Howard as an example to explain one of the fallacy's with WAR when comparing two players. Dunn has shown an ability to walk with runners on, but not hit. Whereas Howard as shown an ability to elevate his game which doesn't necessarily show up in WAR.