1500 words

Introduction

Spring training is ramping up to prepare MLB players for the beginning of the season at the end of the month. (Go Yankees, Yankee fan for 30+ years.) To celebrate, I’m going to discuss sabermetrics and psychometrics and why sabermetrics > psychometrics. The gist is this: Sabermetrics and sabermetricians are actually measuring aspects of baseball performance (since there are observable physical events that occur, and then the sabermetricians think of what they want to measure and then use tangible values) while psychometricians aren’t measuring anything since there is no specified measured object, object of measurement and measurement unit for IQ or any psychological trait. I will mount the argument that sabermetricians are actually measuring aspects of baseball performance while psychometricians aren’t actually measuring aspects of human psychology.

Sabermetrics > psychometrics

Psychometrics is the so-called science of the mind. The psychometrician claims that they can measure the mind and specific attributes of individuals. But without a specified measured object, object of measurement and measurement unit for any psychological trait, then a science of the mind just isn’t possible. Psychometrics fails as true measurement since it doesn’t meet the basic requirements for measurement. When something physical is measured—like the length of a stick or a person’s weight—three things are needed: a clear object (a person or stick); a specific property (length or weight); and a standard unit (inches or kilograms). But unlike physical traits, mental traits aren’t directly observable and therefore, psychometricians just assume that they are measuring what they set out to. People think they because numbers are assigned to things, that psychometrics is measurement.

---

Sabermetrics was developed in the 1980s, pioneered by Bill James. The point of sabermetrics is to used advanced stats to analyze baseball performance to understand player performance and how a manager should build their team. We now have tools like Statcast where the exit velocity is measured once a player hits a ball, and we can also see the launch angle of the ball after it leaves the bat. It clearly focuses on measurable, tangible events which can then be evaluated more in depth when we want to understand more about a certain player.

For instance, take OBP, SLG, and OPS.

OBP (on-base percentage) is the frequency by which a player reaches base. This could be due to getting a hit, drawing a walk or being hit by a pitch. The OBP formula is: OBP = hits + walks + hit by pitch / at bats + walks + hit by pitch + sac flies. While batting average (BA) tells us how often we would expect a hitter to get a hit during a plate appearance, OBP incorporates walks which are of course important for scoring opportunities.

SLG (slugging) measures the total bases a player earns per at bat, while giving extra weight to a double, triple and homerun. SLG shows how well a batter can hit for extra bases, which is basically an aspect of their batting power. (That’s is also isolated power or ISO which is SLG – BA.) The formula for SLG is total bases / at bats.

OPS (on-base plus slugging) is a sum of OBP and SLG. It combines a player’s ability to get on base with their power through their SLG. There is also OPS+ which takes into account the ballpark’s dimensions and the altitude of the stadium to compare players without variables that would influence their performance in either direction.

When it comes to balls and strikes there is a subjective element there since different umpires have different strike zones and therefore, one umpire’s strike zone will be different from another’s. However, the MLB is actually testing an automated ball-strike system which would then take out subjectivity.

There is also wOBA (weighted on base average) which accounts for how a player got on base. Homeruns are weighted more than triples, doubles, or singles since they contribute fully to a run. Thus, wOBA is calculated from observable physical events. wOBA predicts run production and is testable against actual scoring.

We also have DRS (defensive runs saved) which attempts to quantify how many runs a particular defenders defense saved which takes into account the defender’s range of his throw, errors and double play ability. It basically is a measure of how many runs a defender cost or saved his team. So a SS who prevents 10 runs in a season has a DRS of +10. (This is similar to the ultimate zone rating—UZR—stat.) Both stats are derived from measurable physical events.

Each of the stats I discussed measure specific and countable actions which are verifiable through replay/Statcast which then tie directly to the game’s result (runs scored/prevented). Advanced baseball stats now have tools like Statcast which analyzes player and ball data during the game. Statcast takes out a lot of subjectivity in certain measurements, and it makes these measurements more reliable. Statcast captures things like exit velocity, launch angle, sprint speed and pitch spin rate. It can also track how far a ball is hit.

The argument that sabermetrics > psychometrics

(P1) If a field relies on quantifiable, observable data (physical events), then its analyses are more accurate.
(P2) If a field’s analyses are more accurate, then it is better for measurement.
(C) So if a field relies on quantifiable, observable data (physical events), then it is better for measurement.

Premise 1

Sabermetrics uses concrete numbers like hits, RBIs and homeruns. BA = hits / at bats, so a player who has 90 hits out of 300 at bats has a .300 average. When it comes to psychometrics mental traits cannot be observed/seen or counted like the physical events in baseball. So sabermetrics satisfies P1 since it relies on quantifiable, observable data while psychometrics fails since it’s data isn’t directly observable nor is it consistently quantifiable in a verifiable way. It should be noted that counting right or wrong answers on a test isn’t the same. A correct answer on a so-called intelligence test doesn’t directly measure intelligence, it’s supposedly a proxy which is influenced by test design and exposure to the items in question.

Premise 2

A player’s OBP can reliably indicate their runs scored contribution which can then be validated by the outcomes in the game. Psychometrics on the other hand has an issue here—one’s performance on a so-called psychometric test can be influenced by time or test type. So sabermetrics satisfies P2, since it’s accurate analyses enhance its measurement strength while psychometrics does not less accurate analyses along with not having the basic requirements for measurement then mean that it’s not measurement proper, at all.

Conclusion

Sabermetrics relies on quantifiable, observable data (P1 is true), and this leads to accurate analyses making it better for measurement (P2 is true), so sabermetrics > psychometrics since there are actual, quantifiable, observable physical events to be measured and analyzed by sabermetricians while the same is not true for psychometrics.

Since only counting and measurement qualify for quantification because they provide meaningful representations of quantities, then sabermetrics excels as a true quantitative field by directly rallying observable physical events. The numbers used in sabermetrics reflect real physical events and not interpretations. Batting average and on-base percentage are calculated directly from counts without introducing arbitrary scaling, meaning that a clear link to the original quantifiable events are maintained.

Conclusion

Rooted in data and observable, physical events, sabermetrics comes out the clear winner in this comparison. Fields that use quantifiable, observable evidence yield better, clearer insights and these insights then allow a field to gauge its subject accurately. This clearly encompasses sabermetrics. The data used in sabermetrics are based on quantifiable, observable data (physical events).

On the other hand, psychometrics fails where sabermetrics flourishes. Psychometrics lacks observable, quantifiable substance that true measurement demands. There is no specified measured object, object of measurement and measurement unit for IQ or any psychological trait. Therefore, psychometrics can’t satisfy the premises in the argument that I have constructed.

Basically, psychometricians render “mere application of number systems to objects” (Garrison, 2004: 63). Therefore, there is an illusion of measurement for psychometrics. The psychometrician claims they can assess abstract constructs that cannot be directly observed while also using indirect proxies like answers to test questions—which are not the trait themselves. There is no standardized unit in psychometrics and, for example for IQ, not true “0” point. Psychometricians order people from high to low, without using true countable units.

If there is physical event analysis then there is quantifiable data. If there is quantifiable data, then there is better measurement. So if there is physical event analysis, then there is better measurement. Thus, if there is no physical event analysis, then there is no measurement. It’s clear which field holds for each premise. The mere fact that baseball is a physical event and we can then count and average out certain aspects of player performance means that sabermetrics is true measurement (since there is a specified measured object, object of measurement and measurement unit) while psychometrics isn’t (no specified measured object, object of measurement and measurement unit).

Thus, sabermetrics > psychometrics.