First-Year Candidates’ Effect on the Rest of the HoF Ballot

This year’s Hall of Fame ballot had a very weak pool of first-year candidates. Bernie Williams was the leading vote getter with just 9.6% of the vote, and the only member to break the 5% cut off to stay on the ballot. At the same time the 14 returning candidates saw their vote total increase by an average of 7.1%, and five had increases of over 10%. Many have suggested that there is a relationship between these two facts; that is, with few good first-year candidates to vote for there were extra votes for the returning candidates.

Last year David Roher at Deadspin/Harvard Sports Analysis Collective noted that over the 2000s the average number of votes per HoF ballot was fairly constant, between 6.6 and 5.35. This would suggest that in years with strong first-year candidates there would be fewer votes for returning candidates and vice versa. I wanted to more explicitly test this relationship and see whether it extended further back than just the 2000s.

I looked at every Hall of Fame vote from 1967, when the current voting rules were put in place. Along the x-axis is the average number of first-year candidates voted for. Along the y-axis is the average change in vote share for returning candidates compared to the previous year (here an increase from 60% to 65% would be denoted by 0.05).

There is quite a strong, and significant, relationship. In years when ballots have about zero first-year players the returning pool sees about a 5% increase. On the other hand when there are over three first-year players on the average ballot returning players see a 5% drop in their vote. This suggests that voters — consciously or not — do have a rough number of players they want to include on their ballot, and thus the number of first-year candidates they vote for affects the rest of their ballot.

Obviously this could spell trouble for the current ballot as the 2013 first-year pool is loaded (Barry Bonds, Roger Clemens, Sammy Sosa, Mike Piazza, Craig Biggio, Curt Schilling), and could receive higher vote share than the 1989 (Johnny Bench, Carl Yastrzemski, Gaylord Perry, Fergie Jenkins) or 1999 (Nolan Ryan, George Brett, Robin Yount, Carlton Fisk) pools, thought the PED issue could complicate things. Going from a historically poor first-year pool to a historically great one, returning candidates could see double-digit drops in their vote totals.




Print This Post



Dave Allen's other baseball work can be found at Baseball Analysts.


7 Responses to “First-Year Candidates’ Effect on the Rest of the HoF Ballot”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. Barkey Walker says:

    The PED users will appear on all the same 25% of ballots that vote for all of the PED users. This will amount to about 1 vote per first-year player and should not be a significant uptick.

    Vote -1 Vote +1

    • Detroit Michael says:

      Bonds and Clemens were amply qualified for the Hall of Fame even if one somewhat discounts their performance for PED use, so I think they’ll get more support for the Hall than McGuire for example. In addition, Piazza, Biggio and Schilling don’t have specific PED allegations affecting their candidacies.

      In short, I think you are wrong that this wave of candidates will not cause an uptick in the number of average votes per ballot.

      Vote -1 Vote +1

      • Barkey Walker says:

        I think most writes will reason, “Clemens and Bonds used, so they don’t get in.” Plus, Curt, “216 wins” Shilling isn’t that great a traditional candidate, and Biggio isn’t exactly a lock either. Finally, you can knock Piazza off the list too because he had a performance enhancing mustache. ;)

        Vote -1 Vote +1

  2. Mac says:

    Fun project here. I’m curious where you got the data for average number of first-year players voted for?

    The one thing (it’s a tad nit-picky) bugging me in this article is the linear trendline. Just eyballing the chart, the data seems much better suited for a linear fit, which to me is more logical. One or two popular new candidate don’t rock the boat very much, but man those star-studded years can really have a profound effect.

    There’s all kinds of fun additional analysis that could dig deeper on this. Here’s one: Do the votes for the longer-tenured returning players rebound the year after huge rookie classes or is it a lingering effect?

    Vote -1 Vote +1

    • Dave Allen says:

      Mac, yeah the fact that all the years with more than two players per ballot fall below the trend line — and pretty far below the trend line — does make make it seem that the effect is not just linear. Meaning, like you say, it is more pronounced in the start-studded years.

      Your question is interesting, I will poke around in the data and see whether I can find anything worth a post.

      Vote -1 Vote +1

      • Barkey Walker says:

        It looks like you did this with R. If you do

        lm1 <- lm(y ~ x)
        plot(lm1)

        It will show you the influential points plot and I'll bet 1999 and 1989 are very influential.

        The stats 101 thing to do would be to rerun without 1999 and 1989 and see what happens to the line.

        Vote -1 Vote +1

      • Dave Allen says:

        Barkey,

        Interesting I didn’t know about that way to plot a linear model in R. Very cool. It highlights three years as influential or outliers or whatever 1992, 2000 and 1989.

        Vote -1 Vote +1

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Current ye@r *