On small samples, briefly

Published by on July 17, 2012 | 20 Responses

Small sample size. It’s the phrase that’s often beaten to death in the statistical community. This being a statistical-leaning blog, I thought it would be a good idea to briefly touch on the topic. “SSS,” in its simplest form, merely means that we need more data before we can draw a conclusion. How much data you need can vary between what you’re trying to measure, but the general idea is more data is always better.

The biggest problem with SSS is when people use statistics in a misleading way to try and make an argument. You’ll often see this practiced by fans and sportswriters alike that are grasping for straws.

To think of this another way, let’s do the old player comparison routine and look at some numbers of two different, and unnamed, players.

Player A: 44 PA, 2 BB, 12 SO, 7 H, 0 2B, 0 3B, 0 HR, .167/.205/.167
Player B: 48 PA, 4 BB, 15 SO, 8 H, 2 2B, 1 3B, 0 HR, .186/.250/.279

So, yeah. Player A and Player B both looked pretty bad over this stretch. Neither hit much (batting average-wise), got on base much, or hit for much power. Player A is Buster Posey from May 1st to May 12th this year; Player B is Brandon Belt from June 25th to July 15th. I don’t think anyone is making the argument that Buster Posey sucks. And yet, the noise on the topic of Brandon Belt, and whether or not he’s any good, seems to get louder each day. Most of that noise seems to be centered around SSS tomfoolery.

Over a full season of baseball, you’ll get more noise than clarity; and, in my humble opinion, it’s one of the reasons why I love baseball so much. You have to look at the big picture. On a game-by-game basis, nothing much really matters. Heck, in a 10-15 game sample nothing is going to be significant. That’s not to say that watching your favorite player go 4-4 isn’t something to behold, because it is, but rather when you try to craft narratives off a series of chaotic events, that are so noisy by nature, you’re going down the path towards disingenuousness. And really, we’re better than that.

So, the next time you read your favorite sportswriter droning on about Player A, or Player B, or Player C and how they’ve either become the greatest (or worst) player ever, and they’re using the smallest of samples to make their arguments, you can, and should, just move on.

Article written by

12 comments
  Livefyre
Newest | Oldest

This comment has been deleted

This comment has been deleted

ariess
ariess

Did anyone ask Jeff Kent if he thought that hitting either in front of or behind

Barry Bonds made him a better  hitter?

AdamDoctolero
AdamDoctolero

 @ariess We're talking about two different things here.  Having Bonds in the lineup around him did not make Jeff Kent a better hitter, it only changed the context in which he hit more often than not.  Opportunity and ability are two different things.

ariess
ariess

 @AdamDoctolero  I think we are word smithing here.  If Kent was an excellent fastball hitter, and he received more fastballs hitting next to Bonds, and he also faced more pitchers in the stretch when batting after Bonds, his hitting results were better.  It does not change his talent, it just allowed him a better situation with which to express those talents. It is not as if Kent were Felipe Crespo, a hitter who could rip on anybody's fastball, but could not hit a breaking pitch.  Even if he hit in front of Bonds, they would still be tossing him breaking balls, because it became established that he could not hit them.  Kent was not lost with breaking pitches, but he took full advantage of his place in the lineup.  For me I think it may have actually made him a better hitter, Because: when he left the giants he remained a decent hitter without Bonds in the lineup.  I am not sure that we disagree, I just think the definition of "better hitter" was not clear.

ariess
ariess

Speaking of stats and sample size, ie; half season, ESPN has OWAR for hitters

and WAR for pitchers listed.  Vogelsong is 2.8WAR and Cain is 2.3 WAR, but

Explain how Melky at 3.5 WAR is worth more wins than Hamilton or Braun, who

both have a lower OWAR per ESPN's calculation than Melky.  There are some other

numbers that just glare out at you.

AdamDoctolero
AdamDoctolero

I want Belt to play everyday and he should be hitting 1st or 2nd to take advantage of the one thing he does well consistently at this point, which is get on base.  However, there is no hard evidence to suggest that "lineup protection" means anything.  It is a purely anecdotal argument to explain why guys who hit in the first four (or so) spots in a lineup accumulate counting stats or hit for a high average.  You could put Brandon Crawford in the 3rd spot and yeah, he may pick up some extra RBIs there (which are meaningless), but he would still be the below-average hitter we know him to be.  This is true in the other direction as a player like Buster Posey is not going to hit a ton more HRs or hit for a much higher average, just because he has Sandoval hitting after him as opposed to say, Ryan Theriot.

ariess
ariess

 @TSF  @AdamDoctolero Belt could lead off!

AdamDoctolero
AdamDoctolero

None of which changes the fact that there is zero actual evidence that shows who bats in front or behind you in the order matters.  The overarching point here is that what Bochy is doing right now is sub-optimal as he is not taking full advantage of what his hitters do best, and he is giving a larger number of PAs overall to an inferior hitter like Theriot.

 

Ultimately we're arguing the same thing though, as I think we both agree Belt needs to be hitting higher up in the lineup.

 

AdamDoctolero
AdamDoctolero

 @TSF I assume this response was for campanari as I totally agree.

AdamDoctolero
AdamDoctolero

None of which changes the fact that there is zero actual evidence that shows who bats in front or behind you in the order matters.  The overarching point here is that what Bochy is doing right now is sub-optimal as he is not taking full advantage of what his hitters do best, and he is giving a larger number of PAs overall to an inferior hitter like Theriot.

 

Ultimately we're arguing the same thing though, as I think we both agree Belt needs to be hitting higher up in the lineup.

TSF
TSF

 @AdamDoctolero You are missing the point. Belt's biggest asset right now is his patience, but the Giants don't need a high OBP to the exclusion of all else from first base. They need a power hitting, or high average 1B first and foremost, and Belt isn't either yet. Shifting Belt's position in the lineup is to better take advantage of his knack for taking a walk, in the vein of moving Angel Pagan out of leadoff into the middle of the order will make the team more productive overall, and Belt will stay in the lineup more consistently.

 

Batting Belt 7th is daring him to drive in runs, because a light hitting shortstop and the pitcher can't be trusted to do it. Putting Belt at 6th (in front of Pagan) or 2nd (in front of Cabrera) allows Belt to walk all he wants and not worry about driving in runs. 2nd is ideal because if Blanco gets on, and steals second base, Belt will bat in a low pressure situation since he can't hit into a double play. If Blanco doesn't get on, Belt becomes the on base man for Melky.

 

The justification for battling Belt so low is really, really weak. "Theriot feels comfortable" or "Belt needs to grow up fast" is bullshit. The Giants aren't doing Belt or themselves a favor by not maximizing the skills of their players.

TSF
TSF

This is some sticky territory, Chris. While there's little doubt that writers and fans love to use isolated events to cherry pick a preconceived narrative, "small sample size" ends up really subjective. Baseball generally can't wait for "statistical significance" unless a player is good. Few...perhaps only one contending team could stomach the likes of 2011 Aubrey Huff or 2012 Tim Lincecum. Similarly, someone in the Mariners' organization is going to be held responsible for giving 1200+ PAs to Justin Smoak.

 

While it's obvious that Belt is easily the best option at 1B in the entire Giants system, how he's being used usually makes him look worse than he really is. Gregor Blanco has Melky Cabrera to bat him in, while Belt has Brandon Crawford and/or the pitcher. Batting him second, or with Pagan behind him, might make Belt a part of more rallies and calm people down. He doesn't necessarily have to continue those rallies, but he'll at least get them started.

marcello
marcello

Throughout his entire minor league career he hit RHP better than LHP.