Commentary & Analysis

Finetuning the Baseball Hall of Fame Elections

By Aaron Hamlin & Andrew Jennings, Ph.D.

[This was originally published at Deadspin. When reposting, please reference this page.]

Admission into the Baseball Hall of Fame has been called baseball’s highest honor. Lately, it’s also been baseball’s most controversial.

The controversy has led to increased attention for the election process itself, which is what we’ll examine here. We’re going to talk about why the Hall of Fame shouldn’t restrict the number of picks that voters can choose. We’ll also look at the Hall of Fame’s minimum support requirement for players to stay on the ballot in subsequent years. We’ll leave to others the question of who is allowed to vote.

Here’s the rundown of how the Hall of Fame’s election process works. Every winter, the Hall of Fame hands ballots to members of the Baseball Writers’ Association of America (BBWAA) who have been members at least 10 years. Each member may choose up to 10 players but no more. Any player chosen on at least 75% of the ballots is inducted into the Hall of Fame. On the other hand, for a player who doesn’t get inducted, the show’s not over. A player can stay on the ballot as many as 15 times so long as each year at least 5% of voters pick that player on their ballot.

Let’s look at some data. In this year’s election, slightly more than half of the 571 voters submitting ballots used all ten of their picks. The average was 8.4 picks. The average number of candidates on the ballot over the last five years was 31.8. Over the last 79 years, the Hall of Fame has admitted (counting the three from this year) only 115 of the over 18,000 players to set foot on the field. Minding the three years when there was no vote, that’s about one and a half players inducted per year.

So we have many candidates on the ballot, and a lot of voters use all their allowed picks. The fact that over half of voters use all their picks is a strong indication that voters would choose more players if they could. But the current rules tell them they can’t. Indeed, many critics have taken issue with this 10-pick limit. We take issue as well.

One can see the rationale behind the Hall of Fame’s 10-pick rule. By giving fewer picks, fewer players are going to be able to reach the 75% threshold. This increases the Hall of Fame’s exclusivity. While exclusivity is important here, limiting voters is not the way to accomplish it. In effect, you’re telling voters that you don’t care about all their recommendations.

Regardless of context, whether you’re limiting voters to choose one candidate for an elected office, or you’re limiting voters’ picks to ten to induct players into the Hall of Fame, you’re distorting voter opinion. Just as some political candidates get artificially lower support when the ballot limits voters to choosing one candidate, some baseball players are also going to get artificially lower support when the Hall of Fame limits voters to choose only ten players.

When you limit what voters can convey on their ballot, you shut out valuable information, information you need to make an intelligent collective decision. Further, in this context, limiting voters makes it so that election results for any individual player are dependent on the other players on the ballot. This is a clear source of avoidable error.

What do we mean by this dependence and avoidable error? We mean that because of the 10-pick limit, voters have to factor in who else is on the ballot as well as how other voters will vote.

Say there is a player who is a shoo-in for the Hall of Fame and you agree with the player’s merit. So do you pick this player, someone likely to get inducted even without your vote? Remember that there are many other players you’d like to vote for, too, and the Hall of Fame limits you to only ten picks. If you choose the shoo-in, then you could accidentally keep another player you really like from joining the Hall of Fame. If you decide not to choose the shoo-in, then that shoo-in player gets an artificially lower approval percentage. And if too many people play this tactical game, then the “shoo-in” player may not even get inducted at all.

Voting mishaps can occur on the other end of the spectrum as well. Think of a longshot player that you don’t think will get in. Despite these underdog odds, you think this player deserves to be in the Hall of Fame anyway. But when you have a limit on how many players you can pick, what do you do? If you choose the longshot, then you forfeit the chance to support another player you want to get in. On the other hand, if you decide not to choose the longshot, then that player loses support. If the longshot player turns out to have more support than you thought, then holding out your support could deny that player entry into the Hall of Fame. Your holdout can even keep that player from meeting the minimum 5% threshold for next year’s ballot access.

If we had to point to a recent player that has suffered at the hands of the 10-pick limit, it would be the hapless Craig Biggio. Biggio is a four-time Golden Glove winner, five-time Silver Slugger winner, and seven-time All-Star player. This year, he missed the cut-off by a mere two votes. So we have to look to the 144 voters that didn’t choose Biggio. How many of them actually wanted to choose him but couldn’t because they had already maxed out their ballot? If the answer is at least two, then Biggio should have hit the 75% threshold. As more and more noteworthy players bump elbows on the ballot, this issue will only become more prevalent.

Another peculiarity to notice is the fluctuation in support that players receive from year to year. This is a red flag that election results are dependent on the other players listed on the ballot. This year, Lee Smith, a right-hander with a 95 MPH fastball, lost nearly 18 percentage points of the support he had last year. Was there a Lee Smith scandal only the voters were aware of?

From 2010 to 2012, Cincinnati’s legendary shortstop Barry Larkin (and now Hall of Famer) received a much more positive fortune. His support jumped an incredible 35 percentage points. But lacking some kind of public relations campaign, this makes little sense.

All this is a side effect of the Hall of Fame limiting its voters. The players on the ballot have long since finished all their games. There’s no new information to figure in. So why would so many voters change their mind from year to year? The clear explanation is that voters are forced to misrepresent their actual opinion in order to accommodate a limiting ballot.

So how can the folks at the Hall of Fame stop limiting their voters? Easy, they . . . stop limiting their voters. Let the BBWAA (or whomever else they decide to let vote) choose as many candidates as they want and remove that 10-pick limit. What you actually want from voters is for them to consider each player independently and indicate whether they believe that player deserves to be in the Hall of Fame. And letting voters choose as many players as they want creates that independence. It should simply never matter what other players are on the ballot.

We know what you’re thinking now. Won’t permitting voters to choose as many players as they want make it easier to get into the Hall of Fame? With the same 75% cutoff, perhaps it will be a bit easier. But it won’t be the flood of players that you’re so worried about. Moreover, there are still other more reasonable means to control entry.

Consider that even without the 10-pick limit, the Hall of Fame holds a high bar. Three of every four BBWAA voters must agree that a particular player belongs in the Hall of Fame. And it’s really doubtful that removing the 10-pick limit will cause voters to become too free-spirited with their Hall of Fame selections. One indicator that such outlandish voting is unlikely is from Deadspin’s own readership poll.

Deadspin polled its readers and used that information to fill out its acquired Hall of Fame ballot. But Deadspin didn’t restrict its readers to only 10 picks. Rather, readers could choose as many players as they wanted. Consequently, players’ support was 16% higher on average. (Readers also appeared quicker to forgive players that allegedly used performance enhancing drugs.) In the end, with a 75% threshold, Deadspin readers would have elected the same three players as the BBWAA voters, plus two more: 10-time Silver Slugger Mike Piazza and the pitiable Craig Biggio. While that’s more than the three the BBWAA selected, there was no flood of players hitting the 75% mark.

If you let voters choose as many players as they want, then the Hall of Fame is still likely to remain a home reserved for an elite 1% – even if you stick with the 75% threshold. In any case, there are simple enough adjustments to preserve the Hall of Fame’s exclusivity should it become too permissive.

To keep its exclusivity, the Hall of Fame could simply raise the entry threshold from 75% to 80% or 85%. And experimenting to determine what that threshold is easy enough. The BBWAA can just rerun the 2014 election with a mock ballot where voters are permitted to choose as many players as they want.

For completeness, we’ll consider how letting voters choose all the players they want affects the 5% ballot access threshold. It’s true that without the 10-pick limit more players will be able to hit the 5% necessary to stay on next year’s ballot. Fortunately, when you allow voters to choose as many players as they wish, a crowded ballot doesn’t really complicate anything. That’s because voters don’t need to worry about which other candidates are on the ballot or making sure they don’t go over some arbitrary pick limit. Rather, voters consider each player independently. Other voting methods, however, such as ranking methods or methods with arbitrary pick limitations, become exponentially more complex for voters as the ballot gets longer. Unnecessary complexity is another good reason to stay away from such voting methods.

That said, even with the simple choose-as-many-as-you-want voting method that we recommend, 50+ players on the ballot may be a bit much for voters. So let’s see if it’s safe to increase the elimination threshold.

Because voters can look at players independently when the 10-pick limit is removed, a player’s approval fluctuation from year to year should be much smaller. (Technically, one would say that we’re reducing error.) Because the players’ approval ratings are now independent of each other, we can use statistical confidence intervals to tell us the range for each player’s true approval percentage[1].

For example, let’s take a player that gets approved by 65% of the 571 eligible BBWAA voters. We can calculate with 99.9% confidence that the player’s true approval is between 58.4% and 71.6%. So, considering only statistical error, a player with an approval as high as 65% in a given year is still extremely unlikely to be have a true approval of 75%. These technical indicators suggest that a 5% minimum is extraordinarily generous. Really, even a threshold of 55% would be plenty high enough to rule out the possibility of someone getting eliminated unfairly due to random error.

Random error aside, there are other factors that can cause approval percentages to change over time, factors that may cause fluctuation outside of calculated confidence intervals. Some voters may hesitate to pick a player new to the ballot; they may feel more comfortable waiting a few more years before evaluating his legacy. Another factor may be the elections themselves. Voters may use the feedback from election results to alter their opinion in later elections. Yet the most significant factor nowadays may be a change in public opinion over controversial issues – think performance enhancing drugs. A shift in opinions on sensitive issues creates the potential for a significant increase in approval rating over the years.

For all of these reasons, the Hall of Fame should certainly avoid setting the threshold too high. While the 5% minimum is clearly too low, a threshold of 20% would provide an adequate buffer for statistical error as well as evaluations that may change over time.

In all, the Hall of Fame has a real opportunity to simplify their election process while solving some pestering issues. The arbitrary and error-inducing 10-pick limit is a poor way to create exclusivity. Instead, voters must be able to choose as many players as they want. This freedom preserves the integrity of voter opinion while keeping the process simple. And exclusivity can easily be preserved under this approach by raising the induction threshold.

This alternative approach we’re proposing (letting voters choose as many players as they want) is actually based on a voting method called approval voting. While our organization, The Center for Election Science, promotes this method for government elections, approval voting can be used in a variety of other contexts. We believe everyone deserves to have smart elections – from local government all the way to the Baseball Hall of Fame.

[1] The “true approval percentage” is, technically, the approval percentage in a hypothetical poll of all voters ever qualified to vote on on Hall of Fame admission, even those that do not yet exist and those who no longer exist. (The technical term for this group is a superpopulation.) By assuming that the current voting pool is representative of that hypothetical group, we can calculate confidence intervals for this true approval percentage.