Abstract
Media bias is well documented in the industrial relations domain. This paper extends this research by exploring whether union participation among former professional baseball players affects the likelihood of moving through two stages of the selection process for the Major League Baseball Hall of Fame (HOF), a selection process controlled by sports media. At the HOF ballot inclusion stage, union activism increased position players’ and pitchers’ likelihoods of inclusion, regardless of time period. Conversely, at the HOF voting stage, position players who were union representatives during labor-management conflict (i.e., strikes and lockouts) received significantly less votes than non-activists, while position players who were representatives during labor-management cooperation received significantly more votes. Union activism did not affect pitchers at this stage. We conclude that union activists can be subject to negative media bias during labor-management conflict that, in turn, negatively affects post-employment outcomes.
Introduction
It is well established that media bias plays a part in coverage of the industrial relations and collective bargaining domain. Studies have shown media coverage of union activity emphasizes union corruption, strike activity, and violence. This negatively affects public opinion about unions, harming employee support for union representation (Bruno 2009; Harmon and Lee 2010; Puette 1992; Schmidt 1993). In short, media coverage of labor relations issues largely centers on the adversarial elements of labor-management relations. Moreover, historical reviews, while lacking consensus on the antecedents of media bias, find that the union movement and high-ranking union officials have often been recipients of negative, unfair, and biased journalism (Gilpin 2020; Glende 2020; Richards 2008). Media bias against union activists may be particularly salient in contexts where there is significant labor-management conflict—such as work stoppages. In other words, those active in the union during work stoppages may be a target for media scrutiny.
Most studies examining media bias focus on media content. Unlike prior studies, we investigate media bias in terms of actions taken by journalists outside of their role as creators of media content. Specifically, we examine two stages of the selection process for candidates into the National Baseball Hall of Fame (HOF), the most prestigious award given to former players in Major League Baseball (MLB). 1 The decision-making body for the HOF is sports journalists who are members of the Baseball Writers’ Association of America (BBWAA). Among the former players eligible for selection is a cohort of retired players who were player representatives (i.e., union stewards) for the MLB Players Association (MLBPA). In this paper, we identify and differentiate between two time periods that capture significant differences in the tenor of labor-management relations—that is, the level of conflict between baseball ownership and the MLBPA.
Examining media bias using HOF data is not unusual. Researchers have investigated for racial bias among sports journalists in the baseball HOF selection process. The concern about bias in the HOF process is understandable, given that the process is often criticized for its subjective nature (James 1994). 2 Members of the Writers’ Association have occasionally used the HOF selection process in a vindictive way (Wulf 1988). While the results from these studies are sensitive to model specification, there is evidence of racial and ethnic bias in the HOF process (Findlay and Reid 1997; Desser, Monks and Robinson 1999; Jewell, Brown and Miles 2002). However, racial and ethnic bias does not appear to delay induction into the HOF nor alter the composition of the HOF (Findlay and Reid 1997; Jewell 2003; Mills and Salaga 2011).
No studies have investigated bias towards union activists in the HOF selection process. At the organizational level of analysis, studies show that work stoppages receive considerable media attention—with much of that coverage placing unions, relative to management, in an unfavorable light. Prior work suggests press coverage of collective bargaining views unions negatively, is slanted favorably toward management, and focuses excessively on union-related conflict (Bruno 2009; Flynn 2000; Oshagan and Martin 1999; Erickson and Mitchell 1996). However, at the individual level of analysis, there is a lack of empirical research examining how the media views active and vocal union members, especially during work stoppages. Do negative media portrayals of organized labor negatively affect individual union activists, particularly during periods of labor-management conflict?
In this paper, we separately analyze two stages of the HOF selection process: (a) placement on the HOF ballot and (b) the percentage of votes received on the HOF ballot. In so doing, we distinguish between two periods: (a) periods of labor-management cooperation, and (b) periods of labor-management conflict, based on when labor strikes and lockouts occurred in MLB. The first strike occurred in 1972 over pension funding, with forty seven out of forty eight player representatives voting to strike and eighty six games being lost in the 1972 season due to the work stoppage (Fagan 2018). Subsequently, team owners engaged in lockouts in 1973 and 1976, and further MLBPA strikes occurred in 1980, 1981, and 1985. Another team owner lockout took place in 1990. The last work stoppage involved a 232-day MLBPA strike that overlapped both the 1994 and 1995 seasons (Fagan 2018). Thus, we classify years of labor-management conflict as follows: 1972, 1973, 1976, 1980, 1981, 1985, 1990, 1994, and 1995. We identify players who were team representatives during those years as representatives during labor-management conflict. Team reps during all other years in our analysis are classified as reps during labor-management cooperation, given the absence of strikes and lockouts in those years.
Our analysis produced two key results. First, in the ballot inclusion stage, union activism increased position players’ and pitchers’ likelihood of inclusion by 4 percent to 5 percent, regardless of time period. These results suggest that union activism, even for those who serve during periods of conflict, helps players get onto the ballot. However, our second key finding suggests that in the HOF voting stage, these activism benefits disappear for position players who serve as reps in periods of conflict. While position players who serve as team reps during cooperation still experience positive outcomes in this stage (i.e., increased HOF votes), position players who are reps during strikes and lockouts receive 46.7 percent to 49.8 percent fewer HOF votes from baseball writers than their non-activist counterparts. Pitchers, in contrast, were not affected by their union activism at the HOF voting stage.
Media Bias and Organized Labor
There has been a longstanding concern about press hostility to organized labor. For example, Upton Sinclair accused the press of turning public opinion against labor and striking workers: “whenever there is a show-down between labor and capital, the press is openly or secretly for capital” Sinclair (1936, 346). Former U.S. Secretary of Labor Willard Wirtz noted that “It is a significant fact that particularly in this labor relations area … good news, reversing the adage, is no news. A strike is invariably the subject of extended coverage … The peaceful signing of a new collective bargaining agreement … usually on an inside page. I count it a matter of serious concern that the public receives, in the press … a seriously distorted picture of American labor and labor relations” (Zagoria 1967, 43). More recently, Puette (1992) identified significant negative bias in newspaper coverage of organized labor—emphasizing union corruption, strike activity, violence, and links to organized crime, while finding a much lower likelihood of reporting on strike-free negotiations and settlements, labor-management cooperation, or union philanthropic activities. Schmidt (1993) looked at media coverage of unions from 1946 to 1985 and found that it overstated the extent of strike activity, which in turn had a negative effect on public opinion of unions. Jarley and Kuruvilla (1994) also found linkages between public disapproval of unions and negative media coverage of union activity. Carreiro (2005) examined media coverage of anti-union discharges and found that these events are often ignored or treated as isolated incidents. Indeed, several researchers have found that press coverage of collective bargaining has an unwarrantedly negative view of unions, is slanted favorably toward management's position, and focuses excessively on union-related conflict (Bruno 2009; Flynn 2000; Oshagan and Martin 1999; Erickson and Mitchell 1996).
Several explanations have been offered for media antagonism toward unions and the labor movement. From a political economy perspective, the media functions to serve the dominant class and its supportive economic institutions, which is manifested in media ownership, the media's for-profit nature, and the central role of advertisers (Herman and Chomsky 1988; Harmon and Lee 2010). “These varying influences are reflected in media content, which is dominated by the interests of the business and political elite. The media are used as a tool to attempt to perpetuate hegemonic messages that are meant to universalize dominant interests and obscure or marginalize alternatives” (Lewis and Proffitt 2013, 302).
The decline in media resources devoted to covering unions and collective bargaining, such as a decline in dedicated labor reporters and an increase in business reporters, provides evidence of a historic yet ongoing process by which labor has been marginalized in popular media (Zagoria 1967; Schmidt 1993; Park and Wright 2007). One result is that business journalists often cover labor-management stories. Some assert that this shift may exacerbate bias because business reporters suffer from conflicts of interest since they may be reluctant to “antagonize their {business} sources or they might get cut off from all information” (McChesney 2004, 69).
Zagoria (1967) suggests that bias may arise because media companies receive significant advertising revenue from business interests and almost none from unions. Moreover, the media is also an employer, and from that perspective, the media may be reluctant to give unions recognition or credit (Zagoria 1967). Bruno (2009) engaged in a ten-year content analysis of industrial relations coverage of the Chicago Tribune and found that the paper established a negative tone in coverage of the labor movement. The author chose to focus on the Chicago Tribune because of the history of highly contentious labor relations between the paper and its employee unions—an exemplar of the fact that media companies are employers.
Martin (2004), using media framing theory, views media hostility to labor as a function of common media frames—one of which is the emphasis on the priority of consumerism. The consumerist perspective is implicitly pro-employer. A work stoppage is framed based on the impact the dispute has on consumers, not on the underlying employee problems or management actions driving disputes. For example, in the case of a flight attendant strike, the media's attention focused on stranded consumers and not on flight attendant working conditions or airline executive compensation (Martin 2004). Media coverage of the 1994−1995 strike in MLB was also framed from a consumerist perspective. For example, Martin (2004) found that the use of replacement players by owners was characterized in the media as teams providing fans with an inferior product, rather than as part of an aggressive management strategy to weaken and undermine the MLBPA (Martin 2004). In a study focusing on media coverage of the 2011 National Football League lockout, Lewis and Proffitt (2013) determined that media coverage framed the work stoppage as an action denying fans of entertainment because of a dispute between “millionaires and billionaires.”
Baseball and Collective Bargaining
The sports domain, especially professional baseball, has provided a valuable context for researchers studying a variety of organizational phenomenon, such as pay variability and team performance (Bloom 1999), pay equity (Werner and Mero 1999), managerial attributes and performance (Kahn 1993), and racial discrimination (Anderson and La Croix 1991; Hanssen and Anderson 1999; Phillips 1983). Professional baseball is attractive from a research perspective because of the availability of objective data capturing a variety of variables (e.g., compensation, player mobility, player position, individual and team performance, and customer activity). Most of these studies focus on the outcomes associated with the actions and policies adopted by team management and ownership.
Moreover, professional baseball provides a robust labor-management context in which to explore media bias. The ebb and flow of unionization among professional baseball players has been well documented, along with the formation of the MLB Players’ Association (MLBPA) in 1954 (Dworkin 1981). From 1954 to 1966, the MLBPA was akin to a company union reliant on baseball team owners for financial and administrative support (Helyar 1994). This changed in 1966 with the hiring of Marvin Miller as the Executive Director of the MLBPA. Under Miller, the MLBPA asserted independence from the owners and negotiated its first collective bargaining agreement (Baseball Owners Raise Minimum Salary $3,000 1968). Since 1966, there have been eight lockouts and/or strikes. Thus, there has been considerable variance in the tone of labor-management relations in professional baseball.
Miller instituted a governance structure with the executive board as the decision-making body of the MLBPA. The executive board consists primarily of player representatives. The players on each team select a player representative and an alternate player representative for two-year terms. According to Michael Weiner, former Executive Director of the MLBPA, beyond executive board duties and involvement in collective bargaining negotiations, the player representative meets with players on the team and addresses their team specific concerns (M. Weiner personal communication, July 20, 2011). Player representatives perform a role analogous to shop stewards in other workplaces. Like shop stewards, player representatives have concerns regarding retaliation by team management that could negatively impact their careers (Korr 2002; Miller 2004). 3
Baseball, Labor, and Media Bias
The media is an important stakeholder in labor-management relationships in professional sports. This is most obvious in cases where media companies have ownership stakes in professional franchises (e.g., the Chicago Tribune and Chicago Cubs; Bruno 2009) or more recently where the MLB has created its own media conglomerate—MLB Advanced Media. However, even without direct ownership interests, the media is a key stakeholder because of extensive financial partnerships with team ownership primarily through broadcast rights (Lewis and Proffitt 2013). Therefore, baseball media may have negative attitudes towards union members who are activists during work stoppages, due to the impact a work stoppage has on sport journalists and media. For example, the 1994−1995 baseball strike had a significant negative impact on baseball broadcast-related journalists—no games, no broadcasting (Wanta and Kunz 1997). In terms of newspapers, the strike also resulted in larger papers cutting the print space devoted to sports, as well as causing an overall loss of newspaper circulation (Wanta and Kunz 1997). Thus, sport journalists likely suffer significant collateral damage during a work stoppage.
Some baseball journalists have been explicit about their hostility toward collective bargaining and players actively asserting their rights through the MLBPA (Swanson 2016). In 1967, Writers’ Association member Furman Bisher wrote, “… professional baseball players are going to have to decide if they are common laborers or professional men … I cannot see a major league baseball player demeaning himself to the status of a unionized laborer …. The players should be happy with the privilege to be major leaguers … and they should have enough pride in their job to realize they should not associate with unions” (Korr 2002, 44). Marvin Miller believed that most sports writers covering labor relations in MLB were biased in management's favor (Miller 2004). 4 However, there were writers with supportive views of the MLBPA (Kates 2016). For example, Red Smith characterized the owners’ treatment of the players as “slave trade” (Smith 1972, 29). Nonetheless, numerous sportswriters wrote articles antagonistic to union activities in baseball as the union gained power as an independent bargaining entity under Miller's leadership (Korr 2002). However, during the initial five-year period of 1966−1971, no work stoppages occurred in MLB.
This changed in 1972, when the first MLBPA strike occurred over pension funding. Forty seven out of forty eight player representatives voted to strike, and eighty six games were lost in the 1972 season due to the work stoppage (Fagan 2018). Subsequently, team owners engaged in lockouts in 1973 and 1976, and further MLBPA strikes occurred in 1980, 1981, and 1985. Another team owner lockout took place in 1990. The last work stoppage involved a 232-day MLBPA strike that overlapped both the 1994 and 1995 seasons (Fagan 2018). We argue that MLB players who were union representatives during these strikes and lockouts faced increased levels of scrutiny and potential bias from sportswriters responsible for HOF voting, as these years were characterized by contentious relations between the union and team owners. Conversely, we argue that players who were representatives in other years (with no strikes or lockouts) face less media bias. This should then lead to differential effects on the HOF voting process. In the next section, we outline our analytical framework for examining this relationship.
Methods
Sample
Our analysis focuses separately on two stages of the baseball writers’ voting process for selecting retired players to the HOF: placement on the HOF ballot and the percentage of votes received on the HOF ballot. In analyzing media bias, we focus on the percentage of HOF ballot votes received, rather than HOF induction, for two reasons. First, only a small proportion of our sample was inducted (12%). It is difficult to establish the existence of media bias against union activists with such a low base rate of induction. Second, as HOF voting is a binary choice (yes or no) for each Writers’ Association voter, focusing on the total number of “yes” votes received for each player is a better method for analyzing the existence of media bias against union activists. By examining groups of writer votes, we examine media bias in the context of voter behavior, rather than aggregate outcomes (e.g., HOF induction).
Our sample consists of 1,781 former players (1,066 position players and 715 pitchers) who retired between 1966 and 2009 and who also met the eligibility requirements for placement on the HOF ballot between 1972 and 2015, identified using Baseball-Reference.com. The analysis begins with the year 1966 because it is widely recognized as the inaugural year for the MLBPA. A player active in the union in 1966 and retiring at the end of the 1966 season would be eligible for placement on the HOF ballot in 1972. Likewise, a player active in the union and retiring in 2009 would be eligible for the ballot in 2015. Multiple sources were used to identify player representatives. 5
Baseball HOF Voting Process 6
The Board of Directors of the HOF authorizes baseball writers to conduct an annual vote by its members to determine the election of former professional baseball players to the HOF. To vote, a Writers’ Association member must have been a writer, columnist, or sports editor from a newspaper, wire service, or web site that covered MLB for at least 10 years. Individual voting decisions are anonymous unless an individual voter decides to disclose their vote. The number of Association members casting ballots has steadily increased over time, from 396 ballots returned in 1972 to 549 ballots returned in 2015.
For a former player to be inducted into the HOF they must pass a three-step process. First, to be considered for inclusion on the ballot, the former player: (a) must have been active as a player in the major leagues within 20 years of election; (b) must have completed at least ten seasons in the major leagues; (c) must have been retired as an active player for 5 years prior to election; and (d) may not be on the ineligible list. 7
Once a former player is deemed eligible, a Writers’ Association Screening Committee determines whether they should be selected for placement onto the HOF ballot. 8 If a player is eligible for the first time, they must be nominated by members of the Screening Committee to be placed onto the ballot. If a retired player is not placed on the HOF ballot in his first year of eligibility, the Screening Committee will not reconsider placing that player on future ballots (Muder 2013). A retired player that appeared on the ballot and did not receive enough votes for induction to the HOF, but received sufficient votes to pass a minimum requirement, could remain on the ballot for a maximum of 15 years. Those candidates not meeting voting thresholds each year are eliminated from future HOF ballots.
The third and final step of the process is the vote by all active members of the Writers’ Association on the candidates listed on the finalized ballot. Voters must follow two rules. First, voters are notified that voting shall be based upon the player's record, playing ability, integrity, sportsmanship, character, and contributions to the team(s) on which the player played. Second, each voter may select as many as ten candidates on the ballot. If a candidate receives votes on less than 5 percent of returned ballots, the candidate is eliminated from future consideration for the HOF. If a candidate receives votes on 75 percent of the returned ballots, the candidate is inducted into the HOF. 9
Models
Prior studies (Findlay and Reid 1997; Desser, Monks and Robinson 1999; Jewell, Brown and Miles 2002; Jewell 2003) focus exclusively on position players (e.g., catchers, first basemen, second basemen) and omit pitchers from their analyses. In this paper, we include pitchers and estimate models for position players and pitchers separately, for two reasons. First, position players and pitchers are in fundamentally different roles with different responsibilities (Werner and Mero 1999). Pitchers are responsible for using their pitching skills to get opposing batters “out” (i.e., causing batters to not be able to successfully reach base). Common pitching statistics include the number of games their teams win and lose while they are pitching and the number of runs other teams score while they are pitching. In contrast, statistics for position players are primarily based on their offensive ability as batters. Common statistics for position players include batting average (i.e., their ability to successfully hit the ball and reach base) and runs scored (Werner and Mero 1999).
Second, position players and pitchers have naturally differing degrees of visibility. Position players play more frequently. In contrast, due to the strain pitching exerts on throwing arms, pitchers play less often. As part of a starting rotation, starting pitchers (i.e., the pitcher a particular team begins a game with) only pitch every five games on average. Relief pitchers can pitch more often than starting pitchers, because they are responsible for “relieving” starting pitchers and pitching less innings; however, relief pitchers still only pitch every few games on average. Thus, position players (especially those eligible for the HOF ballot) are required to face the media more often than pitchers, as media availability on a given day is most often determined by who plays in that day’s game. It may be the case that the added exposure and visibility received from union activism combined with the heightened media availability requirements of playing nearly every game makes position players more prone to media scrutiny than pitchers. These differences necessitate the analysis of position players and pitchers separately.
The first stage of the HOF voting process, placement on the HOF ballot, is a binary outcome: either the player is placed on the HOF ballot, or he is not. Consistent with prior studies, (e.g., Findlay and Reid 1997; Desser, Monks and Robinson 1999), we estimate probit models for the probability of appearing on the HOF ballot. The dependent variable is the outcome of the ballot inclusion decision; explanatory variables include whether the player was a player representative, whether they were a representative during the period of labor-management conflict, and measures of player demographics, position, performance, awards, and visibility.
In the second stage, HOF voting is also a binary response (yes or no) for each BBWAA voter. However, individual votes are unobserved; rather, we observe the total number of “yes” votes received for each player. Therefore, for each player, the observed second-stage dependent variable is the percentage of yes votes received out of total votes. This type of data is referred to as grouped or proportions data, and following prior literature (e.g., Findlay and Reid 1997; Desser, Monks and Robinson 1999; Jewell, Brown and Miles 2002), we convert these proportions to log-odds ratios. Further, as most players who are elected to the HOF are elected in their first ballot, we concentrate on and estimate models for each player's first HOF ballot (Findlay and Reid 1997; Desser, Monks and Robinson 1999; Jewell, Brown and Miles 2002).
For observations in which the percentage of votes received is zero (i.e., players who received no HOF votes), this percentage cannot be transformed into a log-odds ratio because the natural log of zero cannot be computed. Jewell, Brown and Miles (2002) treat this transformation problem as sample selection and follow Findlay and Reid (1997) in employing Heckman's two-stage correction for sample selection to account for no-vote players, who are likely fundamentally different from players who receive votes. This study follows Findlay and Reid (1997) and Jewell, Brown and Miles (2002) in employing Heckman's two-stage correction. 10
The first stage of the Heckman two-stage model is, by definition, a probit estimation of the determinants of a player receiving at least one vote. In the second stage, the dependent variable is the percentage of votes received, and only players who received votes are included. 11 As Heckman models are on firmer ground with exclusion restrictions (Jewell, Brown and Miles 2002), three variables are included in the first stage that are excluded in the second stage; these variables are expected to be correlated with the probability of receiving at least one vote. First, for position players, we include a dummy variable equal to one if a player was an MLB manager, and zero otherwise (Jewell, Brown and Miles 2002). 12 Next, we include two variables that capture postseason play: postseason appearances and World Series championships. Simply appearing in the postseason or winning the World Series, without good performance, is unlikely to directly impact overall HOF ballot voting; however, the increased visibility from simply appearing in the postseason or winning the World Series is likely to affect a player's probability of receiving at least one vote.
Measures 13
Our first independent variable of interest is Union Rep, an indicator with a value of one if a player was a union representative, and zero otherwise. This leads to our independent variable of interest, Union Rep During Conflict. This captures the effect of being a union rep during the following years: 1972, 1973, 1976, 1980, 1981, 1985, 1990, 1994, and 1995, the years of MLB strikes and lockouts. It has a value of one if a player was a player representative during these years, and zero otherwise. This then captures the relative effect of labor unrest on a player representative's likelihood of being placed on the HOF ballot and subsequent percentage of HOF votes received.
Next, we include Wins Above Replacement (WAR) period, an indicator with a value of one if a player retired in the period 1999−2009. WAR is a wholistic performance statistic (used as a control in this study) that measures a player's overall contribution to their team. A player's WAR value is defined as the number of additional wins his team achieved over and above the number of expected wins the team would have if that player were substituted with an average, or “replacement-level,” player. Recent work (Sanders et al. 2021) has argued WAR's usage and rapid popularization as a statistic informed baseball MVP voters from 2004 onward. We identify the WAR period from 1999 onward because while WAR had an immediate impact on MVP voting beginning in 2004, we must account for the fact that 2004 is when players who retired in 1999 became eligible for the HOF. We also include this time period control because prior work (Miceli and Volz 2012) suggests that performance standards have become stricter over time. As a result, all else equal, it may have become more difficult for players with similar statistics to be placed on the ballot and get into the HOF over time.
The explanatory variables used for ballot selection and proportions of yes votes are grouped into demographics, position, performance, accomplishments and recognition, and visibility. Demographic variables for both position players and pitchers include indicator variables for players’ race or ethnicity, which are Black, Latin, and Black*Latin. 14 , 15 For position players, a series of position indicator variables are included for the primary fielding position played: catcher, first baseman, second baseman, third baseman, and shortstop, with outfield as the reference category. For pitchers, as there are only two position categories (starting pitcher and relief pitcher), a single starter indicator variable is included, with a value of one indicating a starting pitcher and a value of zero indicating a relief pitcher.
Next, we include WAR as an overall measure of performance, and we also include recognition variables. For position players, these measures included whether the player has received the most valuable player award and Gold Glove awards (awards given to players at each position who were the best defensive players). For pitchers, we focus on Cy Young awards (recognition as the best pitcher in each season). 16 Lastly, we include measures of the player's level of exposure and visibility, as these could also influence BBWAA voters: whether they played at least 10 percent of their games in both the American and National Leagues (i.e., played in both leagues), the number of teams they played for, and their longest tenure (in years) with a particular team. For example, we include longest tenure as a control because players who were more loyal to a given team during their career (or even spent their entire career with one team) may be subsequently rewarded in HOF voting.
Results
We first calculated means, standard deviations, and correlations for all variables. We find no evidence of high collinearity between explanatory variables in our analyses. For the full position player specification, we find a mean Variance Inflation Factor (VIF) of 1.57; for the full pitcher specification, we find a mean VIF of 1.87. Both values suggest low collinearity and are below commonly used VIF thresholds. Given that the lowest possible VIF value for a given variable is 1, these mean VIF values indicate that collinearity was not an issue in our analyses. 17
Position Player Estimates
Table 1 presents linearized probit estimates for placement onto the HOF ballot for position players, calculated at the sample mean. Their standard errors are estimated using the delta method. In each table, three models were run. Model 1 includes the Union Rep indicator, the WAR period indicator, and controls for demographics, performance, and awards and recognition. The second model, Model 2, adds in Union Rep During Conflict. Lastly, Model 3 adds in controls for visibility. First, in Model 1, we find being union representative significantly increased position players’ likelihoods of inclusion on the HOF ballot. We find additional evidence of this in Models 2 and 3, and the non-significant Union Rep During Conflict coefficient suggests time period does not matter. Thus, in the ballot inclusion stage, union representation benefits position players by increasing their likelihood of being placed on the HOF ballot. Specifically, all else equal, being a union representative increases position players’ likelihood of ballot inclusion by 4 percent to 5 percent.
Effects on Hall of Fame Ballot Inclusion (Position Players).
N = 1,627. Delta-method standard errors in parentheses.
*Significant at the .10 level (two-tailed test).
**Significant at the .05 level (two-tailed test).
***Significant at the .01 level (two-tailed test).
As for the controls, as expected, WAR exhibits a positive and significant effect on ballot inclusion. In addition, we find consistent evidence that awards and recognition (i.e., MVP awards and Gold Glove awards) play key roles in determining the likelihood of ballot inclusion; across both models, winning MVP and Gold Glove awards significantly increase position players’ likelihoods of being placed on the HOF ballot. Also, in Model 3, we find evidence that players who (a) played for more teams and (b) had longer tenures with a particular team had higher likelihoods of ballot inclusion.
The Heckman selection model estimates for position players are found in Table 2. The second-stage dependent variable, the proportion of HOF votes received, was transformed from a proportion into a log-odds ratio. Aside from the inclusion of a control for the year in which a player appeared on their first HOF ballot in the Heckman selection models, our specification is consistent, by player group, across the probit and Heckman models. 18
Effects on Proportion of Hall of Fame Votes Received (Position Players).
N = 493. Standard errors in parentheses.
*Significant at the .10 level (two-tailed test).
**Significant at the .05 level (two-tailed test).
***Significant at the .01 level (two-tailed test).
In Model 1 of Table 2, we do not find evidence that status as a player rep affected HOF voting overall. However, this does not tell the full story: in Models 2 and 3, we find consistent, contrasting effects of being a union rep based on the tone of labor-management relations, suggesting the existence of media bias. Position players who were union activists during labor-management conflict (i.e., strike and lockout years) received significantly less HOF votes than non-activists. Based on the log odds coefficients for Union Rep During Conflict in Models 2 and 3, position player activists during strikes and lockouts received 46.7 percent to 49.8 percent fewer votes than non-activists, which was statistically significant. 19 In contrast, activists during labor-management cooperation received significantly more votes than their non-activist counterparts. This is consistent with the notion of media bias against players who were union representatives during work stoppages: for position players, we find evidence that union activism acts as a penalty if the activism takes place during labor-management conflict and, conversely, can act as a benefit if it takes place during labor-management cooperation.
It must be noted here that we are not able to disentangle the role of being a union rep from the person who takes on that role. That is, we cannot rule out the possibility that there are statistically unobservable differences between players who choose to involve themselves in union activities and those who do not. For example, players who value their reputation and image may be less likely to serve as a union rep during labor-management conflict (or leading up to the conflict). However, we are not able to control for player personality characteristics in this analysis, and we acknowledge this as a limitation of the paper.
As for the controls, we find that performance (WAR), awards (MVP and Gold Glove awards), and visibility (number of teams and longest tenure) are significant predictors of HOF voting, as was also the case with the ballot inclusion stage in Table 1. For position players, the key difference between the ballot inclusion and HOF voting stages is that race becomes a factor in HOF voting. Black players received significantly more HOF votes. However, the Black*Latin interaction was negative and significant in all Table 2 models, consistent with prior work (e.g., Jewell, Brown and Miles 2002). Scholars have identified the intersection of race, language and nationality as important to understanding the complexity of racial discrimination in professional baseball (Burgos 2007, 2009), and the Black*Latin variable highlights this complexity. While Black players who were not born in a Latin American country received more HOF votes (as shown by the Black coefficients across all Table 2 models), Black players who were born in a Latin American country received significantly fewer votes (as shown by the Black*Latin coefficients in Table 2).
Pitcher Estimates
Estimates for pitchers are found in Tables 3 and 4. Table 3 presents linearized probit estimates for placement onto the HOF ballot for pitchers, also calculated at the sample mean, with delta-method standard errors. As was the case with position players, we find that union representation significantly boosts chances of HOF ballot inclusion, regardless of time period. Union representation, all else equal, boosts pitchers’ chances of ballot inclusion by 4 percent to 5 percent, identical to position players.
Effects on Hall of Fame Ballot Inclusion (Pitchers).
N = 1,174. Standard errors in parentheses.
*Significant at the .10 level (two-tailed test).
**Significant at the .05 level (two-tailed test).
***Significant at the .01 level (two-tailed test).
Effects on Proportion of Votes Received (Pitchers).
N = 382. Standard errors in parentheses.
*Significant at the .10 level (two-tailed test).
**Significant at the .05 level (two-tailed test).
***Significant at the .01 level (two-tailed test).
For the controls, consistent with position players, we find that pitcher awards and recognition (i.e., Cy Young awards) significantly increase the likelihood of ballot inclusion, as does visibility (i.e., played in both leagues, longest tenure, number of teams). In addition, higher WAR values boost likelihoods of ballot inclusion. A key difference is that for pitchers, retiring during the WAR period (1999 − 2009) significantly reduces these odds; retiring during this period reduced likelihoods of ballot inclusion by 10 percent to 11 percent. This is consistent with prior work suggesting that performance standards have become more stringent over time (Miceli and Volz 2012). In addition, compared to relief pitchers, starting pitchers faced reduced likelihoods of inclusion; starters were 7 percent to 8 percent less likely to be included on HOF ballots.
Lastly, unlike position players, we find racial and ethnic effects for pitchers in the ballot inclusion stage. Black pitchers, all else equal, were 8 percent to 9 percent more likely to be included, while pitchers born in Latin American countries were 14 percent less likely to be included. Taken together with the Black*Latin results for position players in Table 2, our results suggest that Latin-American players are the group most likely to face discrimination in the HOF voting process.
Turning to the Heckman model for pitchers (Table 4), in Model 1, we do not find evidence that union rep status affected HOF voting overall. This is consistent with the finding for position players in Model 1 of Table 2. However, unlike position players, we do not find sufficient evidence that being a union rep during conflict harms HOF voting outcomes for pitchers. While the coefficients are negative, they are not statistically significant. The evidence of media bias during conflict against pitcher representatives is then weaker than position player representatives, showcasing the necessity of analyzing position players and pitchers separately. Yet taken together with the position player results from Table 2, the results show that in periods of labor-management conflict, depending on playing position, union representatives can be subject to media bias such that their status as a player representative can act as a penalty with respect to HOF voting.
Discussion and Conclusion
Research has shown media bias in coverage of industrial relations and collective bargaining and that media coverage of union activity largely centers on adversarial elements of labor-management relations (e.g., Puette 1992; Bruno 2009; Harmon and Lee 2010; Lewis and Proffitt 2013). While the existence of media bias in coverage of industrial relations activity is well established, we know far less about its effects at the individual level. To date, no empirical research has analyzed the role labor-management conflict plays in shaping journalists’ opinions toward union activists and the potential effects on activist outcomes resulting from those opinions. In this paper, we begin to address these questions by examining two stages of the selection process into the HOF, estimating separate models for position players and pitchers.
This analysis produced two key results. First, in the ballot inclusion stage, union activism increased position players’ and pitchers’ likelihood of inclusion by 4 percent to 5 percent, regardless of time period. These results suggest that union activism, even for those who serve during periods of conflict, helps players get onto the ballot. However, our second key finding suggests that in the HOF voting stage, these activism benefits disappear for position players who serve as reps in periods of conflict. While position players who serve as team reps during cooperation still experience positive outcomes in this stage (i.e., increased HOF votes), position players who are reps during strikes and lockouts receive significantly fewer HOF votes from baseball writers than their non-activist counterparts. Specifically, position players who were reps during conflict receive 46.7 percent to 49.8 percent fewer votes than non-activists. Pitchers, however, were not affected in the HOF voting stage when engaging in union activism during work stoppages. Even without direct ownership interests, the media is a key stakeholder in labor-management relationships in professional sports via extensive financial partnerships with team ownership, primarily through broadcast rights (Lewis and Proffitt 2013). Work stoppages can exert significant collateral damage on sports journalists and media, as Wanta and Kunz (1997) showed regarding the 1994−1995 MLB strike. These financial impacts may lead the media to view organized labor less favorably during more contentious periods, as sports media entities have a vested financial interest in games continuing uninterrupted. Our results, particularly for position players, support the notion that union activists were viewed less favorably by journalists responsible for HOF voting during strikes and lockouts, leading to fewer HOF votes. In contrast, when there is relative peace and cooperation between organized labor and management, the media directly benefits, because there are fewer work stoppages preventing the continuation of sports games essential to their bottom line. This may explain why union activists in years with no strikes or lockouts were not penalized with fewer HOF votes.
This leads to an unresolved question: why did position players who were union activists during conflict experience HOF voting penalties, while pitchers did not? The answer may lie in their differing visibilities. As Werner and Mero (1999) note, position players play more frequently while pitchers play less often. Thus, position players (especially those eligible for the HOF ballot) are required to face the media more often than pitchers, as media availability on a given day is most often determined by who plays in that day's game. It may be the case that the added exposure and visibility received from union activism combined with the heightened media availability requirements of playing nearly every game makes position players more susceptible to negative media bias in HOF voting. Conversely, pitchers may be comparably shielded from these effects, because while they may be just as active and vocal in union activities as position players, their lower playing frequency requires less media availability than that of position players.
This study is not without limitations. As noted earlier, we cannot account for unobservable individual characteristics potentially associated with union activism. While prior work has shown that union instrumentality, pro-union attitudes, organizational commitment and job satisfaction are antecedents of union commitment (Bamberger, Kluger and Suchard 1999) and that decreases in real wages and increases in unemployment increase the probability of union activism (Angrave, Charlwood and Greenwood 2017), no prior work has examined the relationship between the tenor of labor-management relations and the types of workers drawn into union engagement. Periods of labor-management conflict or labor-management cooperation may draw different employees into union engagement; as mentioned in the results, we are not able to disentangle the role of being a union rep from the person who takes on that role. During periods of conflict, players who are more conflict-averse may be more likely to turn down the union representative role, while players who are more assertive may be more likely to accept the role. Conversely, during periods of cooperation, players who are more conflict-averse may find the environment more suitable for their personality and, thus, may be more willing to accept the role. In sum, the personality makeup of union representatives may differ by period, based on the existing labor-management relational environment; this can lead to varying behavioral patterns. While anyone accepting the role of union representative likely has leadership qualities, we cannot control for other characteristics of these individuals or account for variation in their personality or union representative style. This may affect media voting behavior, depending on how the media reacts to different union representative behaviors. Future research is encouraged to explore how the tone and context of labor-management relations affects the personality mix of those drawn into union engagement.
A second limitation of our analysis is that it cannot speak to whether union representatives who are activists are negatively impacted with respect to HOF induction. Our sample of former players who were inducted was low. Even so, analyzing effects on proportions of HOF votes received is more informative in the context of this study. As HOF voting is a binary choice (yes or no) for each BBWAA voter, focusing on the total number of “yes” votes received for each player is a better method for analyzing the existence of media bias against union activists, as it examines groups of individual BBWAA votes rather than aggregate outcomes (e.g., HOF induction).
Finally, the generalizability of our results beyond the sports and entertainment context are less clear. There are few contexts where the media can have a direct systemic economic impact on a worker post career—as occurs in the context of the baseball HOF selection process. 20 Former players who are included on the ballot, then remaining on the ballot for multiple years, and then eventually selected as members of the HOF, are likely to garner financial benefits from the exposure provided by the HOF selection process. The media also plays a significant role in the selection of players to the Pro Football HOF. Thus, an examination of that process could be informative. Still, few industries have historical reward systems, where the media has a powerful gatekeeping role. Thus, the major contribution of this work is its addition to the broader literature examining media bias and organized labor.
Nonetheless, we encourage future research to explore the effects of media bias on union activists and other individual actors in other contexts, both within and outside the sports domain. For example, longitudinal studies that follow media coverage of union leaders (e.g., Presidents of National or Local Unions) through periods of labor-management cooperation and conflict would be warranted.
Footnotes
Acknowledgments
The authors want to thank Dale Belman, Matthew Bodah, Paul Swiercz, Paul Wolfson, David Findlay, Mark Hyman, Michael Weiner, and all reviewers for their helpful comments and suggestions. We also thank Eric Bohn, Kevin Johnson, Stephen Souzzi and Christie Witzig who provided valuable assistance in data collection. Finally, the authors thank the George Washington University International Institute of Tourism Studies for financial support. All remaining errors are ours.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the George Washington University International Institute of Tourism Studies.
