I had a question recently in the “Hey, Bill” section about whether the tendency of the Gold Glove voters to award the Gold Glove to guys who are big names because they are big hitters has gotten worse in recent years.   I suppose the first thing I should say in response to this is that the Gold Glove voting system has never been a good one.  I generally defend the Gold Glove voting, because

            1)  I have always thought it was useful to have an actual record of the subjective judgments of the time, even when those subjective judgments are whacko, and

            2)  Some people overstate how bad the voting is.

            The Gold Glove selections aren’t systematically bad, but they are spotty.   It is an extremely bad voting system, that leads frequently to poor selections, and leads sometimes to absolutely ridiculous voting selections.  It always has. 

• In 1958, an American League Gold Glove in the outfield went to Norm Siebern, who was with the Yankees and hit .300, but who was not a good outfielder.  Siebern embarrassed the then-new award by making a costly mistake in the outfield in the third game of the ’58 World Series, then making two costly mistakes in the fourth game, quickly becoming notorious for his defense.  He spent almost all of the rest of his career as a first baseman.
• From 1962 to 1977, Jim Kaat won a Gold Glove every year whether he deserved it or not.   Kaat was a wonderful fielder, a big guy who was nonetheless extremely quick and quite graceful—but some of his Gold Glove selections are absurd.  In 1965 he led American League pitchers in errors, with 6, fielding .929 (which actually was up a point from 1964, when he had fielded .928)—but won the Gold Glove anyway.   In 1969 he led all major league pitchers in errors with 8, fielding .826 (!), but won the Gold Glove anyway.  In 1972 he missed over the half the season with an injury, fielded .923—and won the Gold Glove anyway.   In 1977 he fielded .897—but won the Gold Glove anyway.  
• In 1975 Joe Rudi won a Gold Glove in the outfield although he had played only 44 games in the outfield, having played first base most of the year (91 games at first.)
• In 1999 Rafael Palmeiro won the Gold Glove at first base although he had been a DH most of the season, playing just 28 games at first.

These Gold Glove selections are beyond defense; they are simply and obviously wrong.   Usually the poor selections—like those this year of Michael Young and Nate McLouth—are less definitively wrong.    Until the last few years our ways of measuring defense were (a) not very reliable and (b) not widely circulated.   It may be that the Gold Glove awards are no worse than ever, but that they appear worse because the public knows much more now, and is less accepting of awards being given to less-qualified players. 

The essential problem with the Gold Glove vote is that it is an unconstrained plurality with few parameters.   An unconstrained plurality is always at risk of delivering an irrational result, because in certain circumstances it allows fifteen or twenty percent of the voters to determine the outcome.  The Gold Glove voting system is founded on two principles:

      1)  If we get the right people to vote, the system will work, and

      2)  The right people to vote on this award are the coaches and managers.

      Both of these principles are nonsense.  Coaches and managers don’t really know who is a good fielder and who isn’t.   Some do and some don’t, but it isn’t essential to their job to know that.   Many of them don’t take the vote seriously, in the sense that they don’t do anything to prepare for the vote.   And the idea that if you get the right people to vote the system will work is childish nonsense, ignorant of history.  

      That’s not to say that the system always fails.   The system gets a reasonable result 80, 85% of the time.   The problem is that the system CAN fail horribly, so it does sometimes.   There’s no second tier to the vote; there’s no preliminary screening or meaningful eligibility criteria, no review.  It’s just. . .we vote, we announce the winners.   Even if it’s Nate McLoud or Rafael Palmeiro.

      Anyway, the question before the house is whether the tendency to select the best hitters as the Gold Glove winners has gotten worse over time.   The answer is that it has not.  The Gold Glove winners have always been generally very good hitters; in fact, there has never a year in which the Gold Glove winners were not, in general, above-average hitters.   

      I studied the issue in this way.   First, I took from each season the players with the most plate appearances, up to a number equivalent to ten times the number of teams in the majors—that is, 160 players from 1957-1960, 180 in 1961, 200 from 1962-1968, 240 from 1969-1976, 260 from 1977-1992, 280 from 1993-1997, and 300 from 1998 to the present.    It works out loosely to “everybody with 250 or more plate appearances”.  The most plate appearances by any player excluded from the study was 276, by Johnny Edwards in 1973, and the fewest plate appearances by any player who was included in the study, other than the 1981 and 1994 seasons, was 230, by several different players (there are several different years in which the cutoff ends at 230.   In the strike years the cutoff goes as low as 164.) 

      Anyway, we have ten players per team per season. . .not ten players from each and every team, but ten on average in every year.   The regulars.   I then sorted these players by OPS, and coded the top 10% as “10”, the next 10% as “9”, the next 10% as “8”, etc.    I then looked at the scores of the Gold Glove winners (ignoring the pitchers, of course.)

      That may sound easy, but remember:  I’m not a programmer.   It took me about ten hours.   Anyway, the one season in which the Gold Glove winners were the best hitters, relative to norms, was. ..well, actually, there were two seasons.  1967 and 1984.  These are the Gold Glove Teams for those two seasons:    

     The average performance group for every major league season, by this method, is 5.500; it cannot go higher or lower. 

      The average batting performance group of the Gold Glove teams in these two seasons was 8.19.  The average was high in 1967 because 1) good-hitting catchers won the Gold Glove in both leagues, 2) a very good-hitting shortstop won the Gold Glove in the American League, and 3) the NL’s Gold Glove shortstop also had a good year with the bat.  This left the second basemen, Knoop and Mazeroski, as the only below-average hitters on the team.   The NL’s Gold Glove first baseman, Wes Parker, was a famous good-field, no-hit player, but his .704 OPS for 1967 was 40 points above the major league average.    In a year in which only 16 major league regulars hit .300—one per team—six of the sixteen won Gold Gloves.

      In 1967 nine Gold Glove winners where in the top 20% of all hitters; in 1984 ten were.  Three Gold Glove winners of 1984 were below-average hitters—Bob Dernier, Andre Dawson and Ozzie Smith—but none of them were much below average.   .300 hitters and power hitters dominate both teams.

      The weakest-hitting Gold Glove teams of all time were in 1962 and 1964.  The teams above had an “average offensive performance group” of 8.19; in 1962 the average was 6.06, and in 1964 it was 6.13. 

    In 1962 (which was a hitter’s year) only three high-impact hitters won Gold Gloves, those being the Hall of Fame outfielders Mantle, Mays and Kaline, while two guys won Gold Gloves who hit in the .240s with little power and few walks (Virdon and Aparicio.)   In 1964, although four high-impact hitters won Gold Gloves, they were joined by an outfielder who hit .208 (Jim Landis), a first baseman who drove in only 17 runs (Vic Power), and five more below-average hitters.

      But, in fact, the differences between the 1964 team—one of the weakest-hitting of all time—and the 1967 team—one of the strongest—are not that great, nor are these years separated in time by a margin large enough for us to describe them as coming from different eras with different attitudes about who should win the Gold Glove, the Oscar, the Emmy or the hand of the maiden.   Even the weakest-hitting Gold Glove team, the 1962 team, had a performance group average of 6.06, as opposed to a league average of 5.50.   Even in 1962 half of the Gold Glove winners hit .290 or better.  These disparate votes, really, are just something that happened.  Vic Power in ’64 wound up with a Gold Glove in ’64 because he was a brilliant defensive first baseman, and even though he was on the way out by ’64, no one else had stepped forward to claim that position, so some of the voters went back to Power.   It doesn’t really say ANYTHING about the era or the beliefs of Gold Glove voters; it is just something that happened.

      The “average Offensive Performance Group” of all 52 Gold Glove teams is summarized in the chart below:

       These are the 2008 Gold Glove teams:

 Only one true top-level hitter—David Wright—won a Gold Glove in 2008.   Numerous other top-level hitters could and perhaps should have won—Kevin Youkilis (10), Albert Pujols (10), Alex Rodriguez (10), Mark Teixeira (10), J. D. Drew (10), Chase Utley (10), Josh Hamilton (10), Nick Markakis (10), Matt Holliday (10).  Evan Longoria (9).   I personally would have voted for Youkilis over Pena, Pujols over Gonzalez, Markakis over Ichiro, and Matt Holliday over Nate McLouth—but they didn’t win.

            People say that Michael Young won the AL Shortstop Award with his bat, but look at who didn’t win:  Aviles (8), Peralta (7), and Jeter (6).  And, if the voters are just going for the best hitters, what about Chase Utley?   Utley is clearly the best-hitting second baseman in the NL, and he is an outstanding defensive player as well—but he didn’t win.   The vote went to a much weaker hitter.  

      We could do a trend-line analysis of this data in different ways.  If you take a ten-year moving average, the lowpoint of that line would be 1978, and the highpoint would be 2000.  From 1969 to 1978 more weak hitters won Gold Gloves than in any other ten-year period.   From 1991 to 2000, fewer weak hitters won Gold Gloves than in any other period.  In other words, the tendency to give Gold Gloves to good hitters was stronger in the 1990s than in any other time period.   Since 2000 this ten-year average has trended downward:

     The overall average for the 52-year-history of the award is 7.09.   Both the 2008 figure (7.19) and the ten-year moving average (7.16) are very near the historic norms. 

      I do believe that the numbers have trended downward since 2000 in part because of the Rafael Palmeiro fiasco in 1999.   I think that that woke people up a little bit, and got them to take their votes a little bit more seriously.   However, in general, there is little evidence of an actual change in how voters are influenced by hitting stats in selecting Gold Glove winners.  It has always been a valuable and flawed process in which big hitters do well; it is still a valuable and flawed process in which big hitters do well.