Testing HoF Polling for Recency Bias

June 30, 2019

I had the notion that we could look at the first few four-player polls Bill has run as his preliminary HoF polling network (which he explains https://www.billjamesonline.com/a_hall_of_fame_polling_network/here)  and test these sequences arranged by alphabetical order (as arbitrary an arrangement as you could wish for) to see if recency bias appears in them.

If the result were a perfect confirmation of recency bias, all of them would yield the most recently retired player getting the nod for the HoF every time, and the most ancient of the retirees to finish in fourth place.

This is what basically happens, and where it doesn’t, it comes purdy darn close.  There are, in the first ten groupings, three which follow perfectly (in reverse) the order of retirement. I list their names in the order of the top- vote-getter first, the second guy second, etc., and the retirement numbers in the order of the most recent retirement a "1" and the least recent a "4."

Player

RETIRED

Beltre

1

Beltran

2

Belle

3

Bell

4

 

Player

RETIRED

K. Brown

1

K. Boyer

2

John Beckwith

3

Pete Browning

4

 

Player

RETIRED

Clemens

1

W. Clark

2

Jack Clark

3

Cicotte

4

 

1-2-3-4 is the order in which they finished in the polling and the (reverse) order of their retirements. This is fairly impressive, but wait.  In 6 out of the 10 groupings, the most recently retired guy won. In 8 out of the 10 groupings, the most anciently retired guy finished 4th.

But wait. Of the four instances in which most recently retired guy DIDN’T win, he finished second three times. This is to say that the most recent retiree won 6 times, finished second three times, and finished third once (Bartolo Colon).  The most recent retiree in these ten random groupings NEVER finished last.

But wait. Of the two instances in which the most ancient retired guy DIDN’T come in last, he came in third. Which is to say, that the guy who retired the longest ago NEVER finished in first or second place.

There were two instances of the third-most recently retired guy winning the heat: Dick Allen and Norm Cash finished first. But in both instances the guy who finished right behind them was the most recently retired guy (Bobby Abreu and Ron Cey).

So I would say, based on this preliminary study, that recency bias is driving Bill’s first ten HoF polling groups. Whether it’s driving these polls 90 MPH on a straightaway or backing them slowly out of the driveway remains to be seen.

If this pattern continues, the validity of such polling is seriously being undermined by recency bias, I would say, and would need to be corrected for. HOW to correct is delicate, since there may be valid reasons for the recency bias that would make it a valid bias, albeit perhaps not on a one-to-one scale. The most valid reason would be that more recent players are simply better players, a case I’ve been making here for quite some time.  Another valid line of reasoning would be the argument (that Bill has made on occasion) that players from the late 19th and early 20th century have had their shots at election to the HoF, and failed repeatedly—they are personally beyond caring, and their heirs and fans lack a direct personal connection to the dead players, so who is served by electing them instead of electing more recent players who would enjoy the honor? Not saying I buy either of these arguments absolutely, but there is a certain compelling logic to them.

The complete ranking in the first ten HoF polls is as follows:

 

PLAYER

RETIRED

Allen

3

Abreu

1

Bando

2

Allison

4

 

 

Beltre

1

Beltran

2

Belle

3

Bell

4

 

 

Barry Bonds

2

Berkman

1

Blue

3

Bobby Bonds

4

 

 

K. Brown

1

K. Boyer

2

John Beckwith

3

Pete Browning

4

 

 

Robinson Cano

1

Campaneris

3

J. Carter

2

Camilli 

4

 

 

Cash

3

Cey

1

Cedeno

2

Clarkson

4

 

 

Clemens

1

W. Clark

2

Jack Clark

3

Cicotte

4

 

 

Cone 

2

Concepcion

3

Colon

1

Cruz

4

 

 

Damon

1

W. Davis

2

Dahlen

4

Dark

3

 

 

Edmonds 

1

Delgado 

2

Donaldson

4

Elliot 

3

 

 

I don’t necessarily expect this pattern to continue this strong, but I do expect it to continue. I could get more precise, measuring the total percentages of votes rather than the ranks, and the years of retirement rather than the order, but first I’d like to see if this crude measurement continues in further polls. (Technically, I’m going by their last game played, rather than by the year of their retirements—1988 was both Concepcion’s and Cruz’s last season, but Cruz stopped playing in July, Concepcion in September, so Davey C. gets the "3" and Jose C. gets the "4.")  It’s entirely possible that the guy who retired longest ago would win a few polls if he has strong stats and the luck to draw three complete bozos who retired more recently. Put Mickey Mantle up against Rick Manning, Jeff Manto, and Kurt Manwaring and he just might eke out a win (eke!! A mouse!!), so as the sample size of these polls expands, I wouldn’t be astonished to see a 4-3-2-1 finish on occasion. But as the sample size expands, you should be getting more pure 1-2-3-4 finishes as well. We shall see.

 
 

COMMENTS (4 Comments, most recent shown first)

DaveNJnews
The Joe Mauer vote would seem to provide another bit of strong evidence of recency bias, though the Minnie Minoso vote does not.
10:59 AM Jul 1st
 
Steven Goldleaf
Sure it could, Dave. I suggested as much when, out of the blue, Bill posted this afternoon on Twitter that he was surprised to find these HoF polls displaying such obvious recency bias. As I wrote above, he tends not to see the influence of bias nearly as strongly as I do, and we tweeted back and forth a bit, agreeing that it might be true (that these initial polls tend to be won by the most recent retiree) but that it might also be valid: modern players are better players than old-timers, generally, and the early 20th century has been picked pretty clean, etc.

If we look at only the top three, rather than four, or the top two, I think this pattern will continue, and that spells bias more directly. That is, if we're voting on a player who retired in 2005 and another who retired in 1999, and the 2005 guy is still winning most of the heats, then the argument about real old-timers goes away, don't it?
5:25 PM Jun 30th
 
DaveNJnews
Could part of this pattern be a reflection that the Hall has already inducted the overwhelming majority of players who played before 1950/1960? The pickings from those old decades are pretty slim.
4:32 PM Jun 30th
 
MarisFan61
I think maybe most people are wanting first to see what Bill is going to do with it.....
(I am.)​
2:35 PM Jun 30th
 
 
©2021 Be Jolly, Inc. All Rights Reserved.|Powered by Sports Info Solutions|Terms & Conditions|Privacy Policy