Does race matter? Assessing consumer discrimination in the secondary basketball card market

Abstract

Sociologists continue to observe the ways race permeates America's social institutions, the institution of sport being no exception. Although researchers have explored customer racial discrimination via examinations of the secondary sports card market, only three studies have explored the phenomenon in the context of basketball, a sporting context with a higher proportion of non-White players than the baseball and football leagues that have been the primary focus to date. We explore the unique way race matters on the hardwood by employing a methodological approach that previously has been used to study card collecting in other contexts. Data were obtained for 215 retired players and their rookie cards. Controlling for other factors, to include career performance, position, and card scarcity, the results reveal no direct effect of race on card values, but there is an interaction effect between race and Hall of Fame status that impacts card prices. The potential source and implications of this interaction are discussed as well as suggestions for future research.

Notes

1 In a similar—but not sports-focused—inquiry Eberhardt, Davies, Purdie-Vaughns, and Johnson (2006, p. 383) examined “whether the likelihood of being sentenced to death is influenced by the degree to which a Black defendant is perceived to have a stereotypically Black appearance” finding “that in cases involving a White victim, the more stereotypically Black a defendant is perceived to be, the more likely that person is to be sentenced to death.”

2 The 2009 Racial and Gender Report Card, published by The Institute for Diversity and Ethics in Sport, compiles 20 years of results evaluating player race in various professional sport leagues (Lapchick, 2009); however, data were not collected for every sport in each year of interest. For the NBA, no data were available for the 2002–2003 season, for MLB data were not available for the 1989 season, and data were not available for the NFL for the 2000, 2001, 2002 and 2004 seasons.

3 Our focus on race and card prices necessitated certain limitations on the sample that would allow us to control for as many potential factors that could relate to card prices. By selecting cards produced during a specific time period and in a certain condition grade, we are holding constant the potential effect of these variables on card prices. By including only players who are proven high-producers we are able to avoid a number of the theoretical and methodological problems associated with “common cards” (see Regoli et al., 2010). If we were to include the entire population of cards, then we would introduce a myriad of variables in addition to a number of new methodological and statistical concerns into the analysis. Thus, by limiting the sample in the manner we have, we allow ourselves a fairer opportunity to assess whether race influences card prices. At the same time, we do recognize that these sampling restrictions may limit the generalizability of our results and as such, we encourage further research along similar lines of inquiry employing different sampling strategies.

4 In total, 229 players met at least one of the three criteria; however, 14 of those players were not included in the analysis. Two players never appeared on a regular issue rookie card. Three players never appeared in a nationally distributed card set. Four players’ rookie cards were produced by Star Company, a short-lived basketball card producer whose cards were regionally distributed in team sets through hobby dealers, an unusual practice for distributing a card set that has resulted in these cards being labeled “extended rookie cards” as opposed to “regular rookie cards” (Beckett & Hower, 2002). Five players’ rookie cards appeared in the 1980–1981 Topps “Triple Panel” set in which three individual players appeared on each standard card. We determined that the inclusion of cards from this set would be problematic because it is impossible to separate out the value assigned to a specific player on a card that features three players. The 2003 Beckett Basketball Card Price Guide notes that “cataloging the 1980–81 set … is challenging, to say the least” (Beckett & Hower, 2002, p. 30).

5 A list of all cases, to include each player's name, race, rookie year, playing years, and hall of fame status will be made available upon request from the first author.

6 There were nine cases in which a photograph was not available on either Beckett's Sports Cards website or NBA.com. For those cases, an additional Internet search was conducted to find pictures of the player in question. Sources consulted include ebay.com and Google images for pictures of basketball cards featuring the player. In addition, pictures provided on hoopedia.nba.com (“the basketball wiki”) were consulted in instances in which race was not easily discernable from the pictures available on the primary websites consulted in this project.

7 The All Time Pro Sports Statistics Database for the Internet was selected for three reasons: (1) it is an extensive and trustworthy archive of NBA statistics, (2) it is publicly accessible on the website of Bob Bellotti, a long-time NBA team statistical consultant and editor of Total Basketball (Shouler et al., 2003), and (3) a random selection of data from the database revealed that the statistics maintained on-line matched those provided in Total Basketball, the source consulted in this project to provide individual player statistics (see note 9).

8 To calculate an individual player's PCM, the Value of Ball Possession (VBP) (or points per possession) for the entire league for the season of interest must first be calculated using the following formula: VBP = Points/(Rebounds + Field Goals Made + Turnovers + (Free Throws Made/2) + (Blocked Shots/2)). The individual player's PCM can then be calculated by inputting his statistics in the following “points created complex formula”: PCM = ((Points) + ((Rebounds + Steals + Blocked Shots) × VBP) + (Assists × (2 − VBP)) − ((Missed Field Goals + Turnovers) × VBP) − ((Missed Free Throws/2) × VBP) − (Personal Fouls × VBP/2)). See Shouler et al. (2003) for a detailed description of the PCM formula.

9 Two concerns emerged from the use of the PCM statistic that warranted specific consideration. First, Total Basketball (Shouler et al., 2003) only provides data through the 2002–2003 season. Seven players in the dataset, however, played beyond the 2002–2003 season with careers extending from one to three additional seasons. Relying on data from The All Time Pro Sports Statistics Database for the Internet, we randomly selected 12 players, two each from six different seasons, for which Total Basketball maintained statistics. Utilizing the complete PCM formula provided in Total Basketball we calculated each player's PCM and confirmed first, that the statistics maintained in the on-line database matched those in Total Basketball, and second, that we were correctly calculating PCM as described in Total Basketball. Following the confirmation of these concerns, we proceeded to calculate the career-spanning PCM for each of the seven players that played beyond the 2002–2003 season leaving no career data incomplete. Second, a number of players played for multiple teams in a single season. To calculate the regular season PCM in such instances, we developed a formula that enabled a relative assessment of performance based on minutes played per team. For each team played for, the number of minutes played was divided by the total number of minutes played in the entire season and multiplied by the PCM reported for that team. The calculated values for each team were then added to create a season total PCM. For example, Sam Lacey played 20 min for one team (Kansas City Kings) with a PCM of .562, and 650 min with a PCM of .287 for a second team (New Jersey Nets) in the 1981–1982 season. With a total of 670 min of playing time, his PCM for that season was calculated as follows: ((20/670) × .562) + ((650/670) × .287) = .295.

10 Previous research has found vintage, the age of a sports card, to be a significant factor influencing its value. An assessment of our data revealed collinearity between vintage and availability (r = .83). In a model that included both vintage and availability, the overall variance explained was identical to that reported in a model that included availability only (see Model 2) and the vintage variable was not significant (t = −.25, p = .81). In addition, we estimated a model that substituted vintage for availability finding that the variance explained dropped 10 percent, yet the size and direction of the other variables were not meaningfully affected. As such, we retain our focus on availability in the analyses herein.

11 These dollar figures were calculated by taking the exponential function of the predicted y values ($e^{\hat{y}}$) using the coefficients from Model 5.

12 Though Bloom's (1997) ethnographic analysis of sports card collecting and popular culture has become a primary source suggesting that card collectors are predominantly White middle-class men, research in this area is sorely needed. Without a substantial body of research to draw from, an anonymous reviewer correctly noted that (1) the collectors of basketball cards today may have a racial composition that does not reflect that found by Bloom, or (2) perhaps there is a highly select subset of Black basketball card collectors driving up the prices of Black BHOF member cards. Without evidence to lend support for these possibilities, we rely here on Bloom's research but note the need for future efforts to further explore this area of card collecting.