Batting Average is Not a Flawed Statistic

It has become common in the baseball community to bash batting average, to say it is a flawed statistic, for a few reasons:

  • It does not recognize how different types of hits (singles, triples, etc) contribute differently to a player’s value or team’s performance
  • It counts at-bats and not plate appearances (and thus  pretends sac flies, walks, errors, hit-by-pitches, fielder’s choices, etc do not happen)
  • Seemingly significant differences between averages (such as between .290 and .300) are, in reality, the difference of only a few hits

This post was inspired by a blog post that Joe Posnanski wrote about F.C. Lane, a baseball enthusiast in 1916, who argues how BA is useless. Here is Joe summarizing Lane’s four-punch syllogism:

Punch 1. Question: Is a scratch single with nobody on base worth as much as a grand slam?

Punch 2: Batting average says yes.

Punch 3: Phelon says batting average can’t be improved upon.

Punch 4: Phelon is saying that a scratch single with nobody on base is worth as much as a grand slam.

I understand where Joe/FC are going with this, and I agree that they are headed in the right direction, and I agree with their destination. My problem here is with the wording of punch (premise) 2.

Batting average does not “say” that a bases-empty single is worth as much as a grand slam. Batting average is not a person with ideas and a mouth. It cannot say anything. It is a statistic, hits divided by at-bats, invented by humans about the performance of other humans. It describes how often a batter has hit safely.

These humans are the ones that say things. Many humans use batting average to do so. Therefore, it is people who cite BA as an important or relevant stat that are saying all types of hits are equal.

The full syllogism becomes: “Batting average does not measure the differences between types of hits. Many people feel that batting average is a useful indicator of a player’s performance. People who feel this way are arguing, indirectly, that all hits are worth the same, e.g. that a single is as important as a grand slam. However, all hits are not the same. Therefore, batting average is less useful than other stats in determining a player’s worth.”

Wordier, for sure, but more precise 🙂

It is impossible for a statistic to be flawed. It is only possible for a statistic to not measure what we want it to measure. Sometimes it takes decades before we figure out whether our statistics are useful — hence the current revolution underway in baseball.

When a statistic turns out to be not as useful as we’d though, the statistic can’t change. It’s just math based on a batter’s performance. Instead, we must change how (or whether) we use it to measure something. In that sense, a statistic can be not useful or irrelevant — but never flawed in and of itself.

Leave a Reply