Data analysis

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/276378321

Major League Baseball and Twitter Usage: The Economics of Social Media Use

Article  in  Journal of Sport Management · December 2015

DOI: 10.1123/jsm.2014-0229

CITATIONS

35 READS

2,690

3 authors:

Some of the authors of this publication are also working on these related projects:

Economic Discrimination on Sport Social Media View project

Incentives in Sports View project

Nicholas Masafumi Watanabe

University of South Carolina

49 PUBLICATIONS   313 CITATIONS   

SEE PROFILE

Grace Yan

University of South Carolina

30 PUBLICATIONS   432 CITATIONS   

SEE PROFILE

Brian Soebbing

University of Alberta

80 PUBLICATIONS   474 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Nicholas Masafumi Watanabe on 18 December 2015.

The user has requested enhancement of the downloaded file.

619

ARTICLE

Journal of Sport Management, 2015, 29, 619 -632 http://dx.doi.org/10.1123/JSM.2014-0229 © 2015 Human Kinetics, Inc.

Nicholas Watanabe and Grace Yan are with the Department of Parks, Recreation and Tourism, University of Missouri, Columbia, Missouri. Brian P. Soebbing is with the School of Tourism and Hospitality Management, Temple University, Philadelphia, Pennsylvania. Address author correspondence to Nicholas Watanabe at [email protected].

Major League Baseball and Twitter Usage: The Economics of Social Media Use

Nicholas Watanabe and Grace Yan University of Missouri

Brian P. Soebbing Temple University

From the perspective of economic demand theory, this study examines the factors that determine daily changes in Twitter following of Major League Baseball teams as a form of derived demand for a sport product. Spe- cifically, a linear regression model is constructed by taking consideration of factors relevant to fan interest: team performance, market characteristics, scheduling, and so on. The results reveal specific determinants that have significant relationship with Twitter following. From a team management perspective, factors such as the content of social media messages, certain calendar events, and postseason appearances can be used to enhance fan interest on social media. In so doing, it brings together communication inquiries and economic literature by delineating a comprehensive and nuanced account of interpreting sport social media from a consumer demand perspective.

The growing role of social media in sport has been an increasingly researched topic (Clavio & Walsh, 2014; Hambrick, Simmons, Greenhalgh, & Greenwell, 2010; Pedersen, 2012; Pegoraro, 2010; Perez, 2013; Sanderson, 2014; Stavros, Meng, Westberg, & Farrelly, 2014). The surge of research interests has, in response, raised the awareness of the pivotal role of social media platforms in the sport industry, applauding the participatory culture (van Dijck, 2009) that enables fans and participants to seek creative self-expressions in digital spaces (Hutchins, 2014; Rowe, 2014). In so doing, the inquiries have certainly made significant contributions to the field, delineating the communicative patterns of social media used by athletes (e.g., Hambrick, Simmons, Greenhalgh, & Greenwell, 2010), the parasocial relationship between audience and athlete (e.g., Sanderson, 2011), sport organi- zation engagement on social media in fostering marketing effects (Lovejoy, Waters, & Saxton, 2012), and so forth.

Although many individuals have approached Twitter as a communicative tool that forces new ways of thinking about the interaction between sport and digital media (Hutchins, 2011), there is also a critical awareness that these early efforts may not have fully captured the wide

array of possible intersections between social media and sport (e.g., Hutchins, 2014; Pedersen, 2014; Wenner, 2014). It is argued that a limited focus on thematic analy- sis of sport social media content still plays an underpin- ning role in the current research agenda (Billings, 2014; Leonard, 2009; Hardin, 2014; Hutchins, 2014; Wenner, 2014). That is, very little attention has been given to frame the communicative usage of sport social media as a consumption behavior, the demand and meanings of which are engaged in a variety of socioeconomic ramifi- cations (Pedersen, 2012). With this in mind, this research seeks to introduce economic demand theory to examine the usage of social media in relation to Major League Baseball (MLB) franchise performance, scheduling, and other factors, aiming to bridge sport economics research and communication inquiries.

Previously, economic demand theory has been widely used to study fan interest in sport products (Bor- land & Macdonald, 2003). The empirical examination of the demand for sport is often traced to the modeling done by Bird (1982) with a demand function to analyze attendance by fans at sporting contests. Based on this approach, the following studies have further developed and refined models to understand determinants of fan interest for sport products in the realm of sport economics (Borland & Macdonald, 2003; Jewell & Molina, 2005). In recent years, with evolving theoretical discussion as well as emergence of new technologies, the methodologi- cal inquiry of attendance models has been extended into television (Tainsky & McEvoy, 2012), pay-per-view

620 Watanabe, Yan, and Soebbing

JSM Vol. 29, No. 6, 2015

(Watanabe, 2012), and other digital channels of sport product (Budzinski & Satzer, 2011). That is, the views on sport consumption and demand have been expanded from the literal form of attending sport events to a wider array of consumption in contexts that are mediated by various technological platforms and socioeconomic forces (Borland & Macdonald, 2003; Dwyer & Drayer, 2010; Shoham & Kahle, 1996; Tainsky & McEvoy, 2012).

As such, it is recognized that compared with sport attendance, the new forms of sport consumption may be associated with varied levels of efforts and costs, while a commonly identified domain is their serving together as active locations of fan interests and desires to be involved in the sporting scene (Dwyer & Drayer, 2010; Seo & Green, 2008). In particular, social media provides a platform that is convenient to access, and constantly ongoing, as well as with more freedom in choosing degrees of interactivity and personalized involvement (Hutchins, 2014; Wenner, 2014). To some, the experience of engaging sport through social media can be minimal, whereas to others, the convenience of access to social media accounts can be developed into habitual, long- term, and intense involvement (Pegoraro, 2010). Thus, with functions of Following and other communicative features as indicators of fans’ engagement (Jensen, Ervin, & Dittmore, 2014), the usage of sport social media also needs to be incorporated as a form of derived demand for sport products (Perez, 2013), where the employment of an attendance model provides a fitting theoretical framework of examination.

With this in mind, situated within the fan attendance literature and methodology, this study seeks to develop a model via regression analysis to analyze determinants of demand that are critical to fan interest in MLB team social media accounts. In so doing, a large panel dataset is employed, aiming to capture daily changes in Twitter followers and usage for every MLB team over a 1-year period. Specifically, the change of Twitter followers in a 24-hr period serves as the dependent variable within the model constructed in this research. The explanatory variables include measures of on-field success for MLB franchises, timing of MLB games, important dates on the MLB calendar, and other crucial variables to provide a comprehensive analysis of consumer interest in MLB Twitter accounts.

The findings attempt to contribute to the literature from three aspects. First, by extending the lens of inquiry into the usage of social media as consumption behaviors in relation to a variety of factors of team performance and game scheduling, the current study seeks to bridge the communication and sport economic research. That is, considering the call for communicative investigations of sport social media to be grounded in stronger theoretical agendas with elaborated explorations of methodological approaches (Hardin, 2014; Hutchins, 2014; Rowe, 2011; 2014; Wenner, 2014; Yoo, Smith & Kim, 2013), this study aims to address the concerns through incorporating eco-

nomic demand theory and econometric modeling into the inquiry of sport social media. Second, despite the fact that the usage of sport social media has become one dominant way for fans to engage in sport product (Clavio & Walsh, 2014), very little examination has been done to analyze its consumption as a derived form of sport demand from an economic perspective (Perez, 2013). Thus, while this study seeks to enrich the diversified profile of sport attendance studies, the results also shed light on moving toward a further conceptualization of fan engagement with sport organizations through social media. Finally, from the management perspective, this study provides ways to assist sport teams to obtain insights in creating social media management strategies through a better understanding of fan interest. The results pertain to not only team performance, but also scheduling of league events, inclusion of teams in postseason playoffs, as well as a team’s presence and use of Twitter.

Literature Review

Social Media, Economics, and Sport

Social media, or social network sites, operate as virtual communities supported by interactive applications allow- ing users to participate in designing, publishing, editing, and sharing in a dynamic environment where content is primarily equal among users (e.g., Boyd & Ellison, 2007; Kasavana, Nusair, & Teodosic, 2010; Starvos et al., 2014; Williams & Chinn, 2010). Among all the different types of social media websites—including blogs, content com- munities, forums and bulletin boards, and content aggre- gators (Constantinides & Fountain, 2008)—Twitter has been particularly regarded as an intervention to traditional sport broadcasting and communication (Pegoraro, 2010; 2013; Rowe, 2011). With increasingly simple-to-use technological advantages that enable users to self-define the utilization of the platform, it “produces stories about sports, intensifying and proliferating media sports content and information available in the public sphere” (Hutchins, 2011, p. 239). In various ways, Twitter has expanded the sport consumers’ participation in communication by breaking away from the traditional one-to-many, single-medium framework offered by television to the many-to-many possibilities, allowing sport consumers to interact with teams, organizations, and athletes in ways that were previously impossible (Anderson, 2008; Fisher, 2008; Pegoraro, 2013; Rowe, 2011).

As framed by Clavio (2011), the extant literature has primarily approached the investigation of sport social media following two lines of inquiry: content-based inquiry and audience-based inquiry, with the exception of a few studies that integrated both (see p. 310–312). Within content-based inquiry, studies have heavily focused on thematic analysis (Sanderson, 2014), aiming to reveal patterns of communication by various sport stakeholders via social media. For example, Hambrick et al. (2010)

Major League Baseball and Twitter Usage 621

JSM Vol. 29, No. 6, 2015

examined how athletes used Twitter accounts to connect with fans, identifying a range of major communicative categories: interactivity, sport information, promotion, diversion, fanship, and content. From the audience-based perspective, studies have sought to understand the char- acteristics, demographics, uses, and gratifications that influence fans’ interaction with social media (Clavio, 2011). Importantly, the underlying assumption for these studies is that the individuals’ activities on social media, such as following a team’s or an athletic department’s account, are considered as expressing consumer interest. Following this common thread, there includes multiple studies on the utilization of social media by college sport fans (Clavio, 2011; Clavio & Walsh, 2014), and Clavio and Kian’s (2010) study that investigates followers’ use of the feed and the affinity with an athlete, as well as Kassing and Sanderson’s (2010) examination on athletes’ uses of tweets cultivate insider perspectives for fans. In addition, a qualitative study on fan motivation for interacting with each other on social media has been conducted by Stavros et al. (2013), revealing four major motivational factors: passion, hope, esteem, and camaraderie.

There are also a few studies that have attempted to approach social media from an economic perspective (e.g., Feddersen, Humphreys, & Soebbing, 2013; Jensen et al., 2014; Perez, 2013). These studies have focused on the analytical modeling of consumer behavior, with social media use serving as a proxy of fan interest in sport products. That is, Twitter or Facebook Followers/ Likes has been commonly used as an indicator to consider the popularity and consumer interest of a sport figure or organization (Feddersen, et al., 2013; Jensen et al., 2014; Perez, 2013). The study by Feddersen et al. (2013), for example, uses Facebook Likes as a proxy for participants with investor sentiment and analyzes evidence of senti- ment bias in the sport betting market. In another study conducted by Jensen et al. (2014), focus is placed on exploring factors that are significant predictors of Football Bowl Subdivision head football coaches’ popularity on Twitter. Specifically, it assumes the coach’s number of followers on Twitter as an indicator of popularity, where a factor analysis as well as regression model were per- formed by taking the number of the coach’s followers as a dependent variable. Moreover, of particular relevance to this current study is Perez’s (2013) research, which modeled factors in delineating the economic demand for Spanish professional football (soccer) teams on Twitter. That is, by examining the number of followers of teams’ Twitter accounts, the analysis reveals the interconnections between the demand of sport consumption via social media and factors of team performance in addition to other market characteristics. As such, it establishes a framework for social media to be used as a legitimate means in measuring fan interest through the approach of an economic demand study.

With this in mind, this current study needs to be considered as carrying inherent connections with the prior

researches at various levels. From a theoretical perspec- tive, it shares the foundation in economic demand theory with Perez’s (2013) study. Furthermore, in terms of the methodological approach to analyze social media data, it employs a linear regression model by regarding social media use as a proxy of fan interest in sport products, similar to the previously mentioned economic studies (Feddersen, et al., 2013; Jensen et al., 2014; Perez, 2013). This study must also be considered as a further development in creating a more robust model. It takes into account of a number of multifaceted variables that are important and relevant in engaging fans—market characteristics, team performance and scheduling—in delineating a more comprehensive view of the consump- tion and demand patterns of MLB social media usage. That is, considering the capturing of social media data, the previous studies are conducted with a single value to measure the number of Likes or Followers on social media in a season (Feddersen et al., 2013; Jensen et al., 2014). However, social media use is an ongoing dynamic process, meaning that interests and numbers of followers are constantly submitted to changes (Sanderson, 2011). Thus, the employment of a single metric to capture lengthy intervals of social media usage can be problem- atic, as it provides less control of numeric changes in the metric over time. Further, the size of the dataset is also relatively limited in these studies. The study by Perez (2013), for instance, has examined a sample of 14 teams in top-flight Spanish soccer for 23 weeks, producing a total of 658 observations. Considering the nature of speed and ephemerality of sport social media, the purpose to produce more developed and precise models naturally requests studies to incorporate enlarged dataset as well as more control variables. With this in mind, this study seeks to enhance the prior models by analyzing daily changes in Twitter use for a 13-month period by employing team- day observations for every MLB franchise.

Demand Theory

The demand for sport products has been one major research inquiry in the field of sports economics, used to theoretically (Neale, 1964; Rottenberg, 1956) as well as empirically (Bird, 1982) control and model factors as to why individuals decide to consume sport. Eco- nomic demand theory has been used as a framework to investigate factors reflecting economic, organizational, demographic, and market factors that lead to increases or decreases in fan interest (Soebbing, 2008). Specifically, the demand for sport comes in a variety of forms, includ- ing live attendance (Borland & MacDonald, 2003), televi- sion viewership (Bruggink & Eaton, 1996; Carmichael, Millington, & Simmons, 1999; Garcia & Rodriguez, 2002; Feddersen & Rott, 2011; Tainsky & McEvoy, 2012), pay-per-view (Tainsky, Salaga, & Santos, 2013; Watanabe, 2012), and so forth. In reviewing the literature of sport demand studies, Borland and MacDonald (2003)

622 Watanabe, Yan, and Soebbing

JSM Vol. 29, No. 6, 2015

has noted five major categories in driving sport demand: consumer preference, economic variables, quality of viewing, sporting contest and supply capacity. Moreover, with the revenue shift from traditional gate attendance to broadcast revenues, the need to understand the digital demand for sport has also grown tremendously (Budz- inski & Satzer, 2011). That is, through the formation of online ticket sites and social media, the digital realm has increasingly played a significant role in influencing prices as well as demand for sport products (Shapiro & Drayer, 2012; Watanabe, Soebbing, & Wicker, 2013). Thus, for sport organizations to obtain maximized profits, it is important to understand the interrelationships between different segments of their product, which in turn provides insights as how to manage revenue streams for optimal financial and organizational outcomes (Budzinski & Satzer, 2011; Buraimo & Simmons, 2009; Mongeon & Winfree, 2012). As such, extending the investigation of sport product demand into the realm of social media may be considered as the next logical step for research.

Specifically, the modeling of sport demand can be traced back to seminal work by Bird (1982) who noted that demand for attendance at sporting events can be set out Equation 1:

Ait = [Pit, Qit, Qit, Mit] (1)

In Equation 1, A is the average attendance for the home team in a single season. The subscript t denotes year and i denotes franchise. The ticket price is P, the first Q is home team quality, the second Q is the quality of the visiting team, and M is the market potential of the home team to attract consumers. The term M can be a wide variety of market variables, such as racial demographics, income, population density and other franchises in the same area (Bird, 1982). As such, Bird’s modeling attempts to control for various factors to understand variables that affect attendance. Following a similar modeling method, researchers have also produced a reduced form equation where the ticket price variable is omitted; often specifically denoted as sport attendance studies, due to the absence of price variable (Jewell & Molina, 2005; Soebbing, 2008). Practices of sport attendance study based on economic demand theory have been applied in a number of sporting contexts, including professional soccer (Jewell & Molina, 2005), minor (Gitter & Rhoads, 2010) and major (Soebbing, 2008) league baseball, professional hockey (Coates & Humphreys, 2012), and professional football (Coates & Humphreys, 2010).

In consideration of this current study, since there is technically no price variable associated with the usage of Twitter, this investigation should also be positioned as following the approach of an attendance study. Thus, situated within the attendance study literature based on economic demand theory, this research seeks to follow the modeling methodology of Bird (1982) and Jewell and

Molina (2005), with the employment of theoretical factors posited by Borland and Macdonald (2003). Specifically, two research questions are formulated:

RQ1: From the perspective of economic demand theory, which factors have a significant relationship with changes in Twitter following for MLB teams in a 24-hour period?

RQ2: How can results from the model estimated in this research be applied to the management of sport team social media accounts?

Methods To provide effective control for the instantaneous changes in Twitter, the NodeXL social media analytical software was employed to obtain reliable data measuring Twitter use. NodeXL is described as a Social Network Analysis tool, which has been employed in a large array of research studies examining content, use and the importance of various forms of social media (Hansen, Shneiderman, & Smith, 2010; Sharma, Khurana, Shneiderman, Schar- renbroich, Locke, 2011). This software allows users to enter commands that interface the program with social media platforms. In the next procedural step, the data can be exported to an Excel sheet, providing temporal consistency in collection.

MLB Twitter data were collected every morning at 10 a.m. from July 6, 2013, to July 27, 2014. That is, as noted by Edelman (2012), data scraping on a daily basis is vital in collecting data from large scale samples over time, which carries significance in economic research examining consumer behaviors in relation to daily changes in price and other factors (Cavallo, 2011; Ellison & Ellison, 2009). Afterward, the program of NodeXL was operated to facilitate the data collection process during this 1-year period. Data for the number of Followers, individual followed (Followed), Tweets, and Favorites was collected for each MLB team’s Twitter account. The data were then transferred from the NodeXL spreadsheets to a single comprehensive panel dataset, which was later imported into the STATA12 statistical software package for further analysis.

In a detailed account, the unit of observation is a team-day. During the sample collection period, there are 11,259 team-day observations. This sample accounts for a 13-month time period, over which approximately five of the months have no baseball games being played. Although the period does not reflect exactly one perfect calendar year or one complete season, the 13-month period is chosen to include as many observations as pos- sible to better capture the daily changes in social media use over time. Thus, the larger dataset is employed in this model to present results without omitting any data to prevent the potential for bias (Gujarati, 2003). Moreover, following the study by Perez (2013), where the depen-

Major League Baseball and Twitter Usage 623

JSM Vol. 29, No. 6, 2015

Table 1 MLB Twitter Follower Change

Team July 6, 2013 Followers

July 27, 2014 Followers

Net Change

Percentage Change, %

Angels 118,881 190,718 71,837 60

Astros 90,600 152,825 62,225 69

Athletics 121,336 191,042 69,706 57

Blue Jays 297,338 460,188 162,850 55

Braves 357,413 513,734 156,321 44

Brewers 144,701 203,210 58,509 40

Cardinals 308,072 475,063 166,991 54

Cubs 255,892 365,142 109,250 43

Diamondbacks 89,964 140,500 50,536 56

Dodgers 315,440 537,185 221,745 70

Giants 441,137 617,034 175,897 40

Indians 136,266 221,791 85,525 63

Mariners 121,177 203,402 82,225 68

Marlins 88,711 126,570 37,859 43

Mets 192,490 271,599 79,109 41

Nationals 131,795 192,691 60,896 46

Orioles 154,376 229,750 75,374 49

Padres 83,992 127,092 43,100 51

Phillies 766,705 834,755 68,050 9

Pirates 140,443 259,729 119,286 8

Rangers 295,262 401,170 105,908 36

Rays 117,853 176,413 58,560 50

Reds 217,502 313,419 95,917 44

Red Sox 495,768 817,873 322,105 65

Rockies 93,462 141,105 47,643 51

Royals 120,480 193,567 73,087 61

Tigers 276,271 447,433 171,162 62

Twins 161,311 223,188 61,877 38

White Sox 145,965 205,542 59,577 41

Yankees 926,323 1,176,926 250,603 27

dent variable was based on the change in the number of Twitter followers from the previous day, the current study uses the change in the number of Twitter followers from the previous collection period (24 hr in length) as the dependent variable (ΔFollowers). As such, this change measures the new individuals who follow the official account of a MLB franchise. Table 1 presents the change in followers from the beginning of the sample period to the end of the sample period. From Table 1, one notices a large variation in the increase in twitter followers ranging from a 9% increase to an 85% increase.

Explanatory Variables

To examine the factors affecting fans’ engagement with MLB social media accounts, a variety of control variables are included in the modeling. These factors are based on the theoretical underpinning of determinants of demand presented by Borland and Macdonald (2003), but also include more recent empirical consideration in regards to social media (Perez, 2013) postseason appearance (Tainsky, Xu, & Zhou, 2014), as well as scheduling and timing of events (Tainsky, Salaga, & Santos, 2012, 2013;

624 Watanabe, Yan, and Soebbing

JSM Vol. 29, No. 6, 2015

Watanabe, 2012). The model in this research is thus constructed upon variables that have been presented in prior research as being important to take into account when controlling for a variety of factors in the demand for sport products.

The first set of explanatory variables look at the teams’ social media use, considering the amount of interaction and interest that these accounts generated. Specifically, Tweets measures the total number of tweets the account made since the past collection period, and represents how active the account of the franchise is in communicating through social media. Favorites denotes instances when the official Twitter account of an MLB team favorites the tweet of an account (either their own tweets or tweets by other accounts) over the past 24 hr. Followed indicates the number of other accounts an MLB franchise followed on Twitter during the past 24 hr.

TwitterDays is a variable that controls for the number of days the observed team’s Twitter account has been active on the observed day. In reviewing MLB teams’ initial approach to Twitter, all 30 teams joined Twitter through making official accounts over an 18-month period, indicating that initially there was probably not a general directive from the league for all teams to join the platform. Over time, more strategic governance policies have been formulated, as it is now the case that both the MLB front office and all teams have social media direc- tors/coordinators whose sole job duty is to manage Twit- ter and other social media accounts, including posting content and interacting with other users/fans (Hutchins & Rowe, 2012). As such, the creation date of the team’s Twitter account was captured by NodeXL and included in the model to control for number of days that teams are on Twitter. Finally, this model seeks to control for the total number of followers for a team’s Twitter account in the previous collection period (Followers). This vari- able is used to take into account the preexisting number of Twitter followers for a team.

In addition to these specific MLB franchise social media variables, team performance variables as well as performance events with potential to have changed the number of Twitter followers are also taken into con- sideration. Specifically, the team performance data are collected from the Baseball-Reference website, which lists daily and aggregate statistics for MLB team perfor- mance. The first variable attempts to capture if the previ- ous day is either a regular season or a postseason game for the observed team (Gameday). The second variable considers if the observed team is currently participating in the Wild Card, Division, or League Championship playoff series on the observed date (PlayoffPart). It is a dummy variable equal to 1 for participating in a playoff game on the observed date, 0 otherwise. In addition to participating in one of the first three rounds of the MLB playoffs, this model also includes a variable equal to 1 if the observed team is playing in a World Series game on the observed date, 0 otherwise (WSPart). It is anticipated

that both the PlayoffPart and WSPart variables will be positive and significant, reflecting an increase in follow- ers, as prior research by Tainsky et al. (2014) displays the importance of playoffs in improving television rat- ings for professional football. The third variable is the difference in divisional ranking between the observed day and previous day for the observed team (DiffRank). It is expected that as the team moves up in the divisional rankings (e.g., going to 3rd place from 1st place), it leads to an increase in the number of Twitter followers. Finally, this model includes four variables factoring for the impact of winning and losing streaks on the number of Twitter followers. Following a similar approach to Brown and Sauer’s (1993) research examining the impact that winning and losing streaks have on point spreads in the National Basketball Association, this present study includes dummy variables equal to 1 if the observed team has won (loss) two or three games in a row on the observed day. In addition, it also seeks to employ dummy variables equal to 1 if the observed team won (loss) four or more games in a row. As such, it is anticipated these variables have a positive impact on the number of Twitter followers in the previous 24 hr for winning streaks, and negative for losing streaks.

The next set of variables considers important scheduling events on the MLB baseball calendar. These in-season events contain the amateur draft (Draft), the All Star Game (AllStarGame), the July 31 trading dead- line (TradeDeadline), and the expansion of the rosters on September 1 of each season (RosterExp). For each of these variables, a dummy variable is used with 1 indicating that one of the in-season events occurred on the observed day. In addition, the model also includes dummy variables for offseason events such as the Winter Meetings (WinterMeetings) and the first day of free agency (FreeAgency). It is predicted that all these events would have a positive and significant increase in Twit- ter followers. Dummy variables are also constructed to capture the effect of nationally televised games (NatlTV), as well as to classify the type of market a team played in (SmallMkt). The variable for televised games measures the presence of both home and away teams in games shown on the following channels: Fox, Fox1, ESPN, ESPN2, and TBS. The SmallMkt variable measures a 1 for teams that were designated as small market franchises by MLB, and is a dichotomous variable that represents the 15 smallest markets in the league. Furthermore, the final set of variables comprises measures for each day of the week (DOW), month of the year (Month), and a separate dummy variable for federal holidays throughout the calendar year (Holiday).1 Scheduling variables are shown to be of importance in the examination of demand for sport products, including analysis of variables captur- ing month (Tainsky et al., 2012, 2013) as well as days of the week and crucial calendar dates (Watanabe, 2012). A list and description of all the variables included in the model are listed in Table 2.

Major League Baseball and Twitter Usage 625

JSM Vol. 29, No. 6, 2015

Table 2 Variable List and Description

Variable Description

ΔFollowers Change in Followers in a 24-hr period

Tweets Number of Tweets in a 24-hr period

Favorites Number of Favorites in a 24-hr period

Followed Number of accounts Followed in a 24-hr period

TwitterDays Number of days on Twitter

Followers(t–1) Number of Followers on the previous day

Gameday(t–1) Team played a regular season or postseason game the previous day (1 = Yes)

PlayoffPart Team participating in the playoffs (1 = Yes)

WSPart Team participating in the World Series (1 = Yes)

DiffRank Difference in the teams Divisional Rank

WinStrk2 Winning streak of 2 or 3 games (1 = Yes)

WinStrk4 Winning streak of 4 games or more (1 = Yes)

LossStrk2 Losing streak of 2 or 3 games (1 = Yes)

LossStrk4 Losing streak of 4 games or more (1 = Yes)

Draft Major League Baseball draft (1 = Yes)

AllStarGame All Star Game (1 = Yes)

TradeDeadline July 31 trade deadline (1 = Yes)

RosterExp September 1 roster expansion (1 = Yes)

Holiday U.S. Federal holiday (1 = Yes)

WinterMeet Winter meetings (1 = Yes)

FreeAgency First day of free agency (1 = Yes)

NatlTV Teams participating in nationally televised games (1 = Yes)

SmallMkt Teams designated as being in a smaller market by MLB (1 = Yes)

Sunday . . . Sunday through Saturday

January . . . January through December

Note. Dummy variables are also included to measure each day of the week and month of the year.

Model Equation 2 represents the linear model employed to estimate results from the dataset.

DFollowersit = b0 + b1Tweetsit + b2Favouritesit + b3Followedit + b4TwitterDaysit +b5Followersit(t−1) + b6Gamedayi(t−1) + b7PlayoffPartit + b8WSPartit +b9DiffRankit + b10WinStrk2it + b11WinStrk4it + b12LossStrk2it +b13LossStrk4it + b14Draftit + b15AllStarGameit + b16TradeDeadlineit +b17RosterExpit + b18Holidayit + b19WinterMeetit + b20FreeAgencyit +b21NatlTVit + b22SmallMktit + b23→28DOWit + b29→39Monthit + ´it

(2)

where i indexes teams, t indexes days, β0 is the constant term, and ε is the equation error term.

626 Watanabe, Yan, and Soebbing

JSM Vol. 29, No. 6, 2015

Estimation Issues

The present data are a perfectly balanced panel, meaning that there is the same number of observations for each team (Gujarati, 2003). However, it is argued by research- ers that a number of potential estimation issues can affect the estimated results in panel data analysis (Asteriou & Hall, 2011). The first issue considered is multicollinearity. Examining the correlation coefficients of the variables in Equation 2, all correlation coefficients are below 0.70 (Tabachnick & Fidell, 2007).2 From these coefficients, there is no need to omit variables in the current model to correct for multicollinearity. The second estimation issue is to determine if a fixed or random effect estimator is appropriate. A Hausman test is conducted to determine the appropriate estimator, the result of which indicates the fixed effect estimator is appropriate. The third estimation issue deals with cross-sectional dependence. It is noted by Baltagi (2008) that this problem may be an issue when dealing with a long time data series such as the one in this research. Taking this into consideration, this study conducts a Breusch–Pagan statistic for cross-sectional independence in the residuals of a fixed effect regression model. The results of the test indicate that cross-sectional independence is an issue. Thus, the Driscoll and Kraay (1998) standard error correction is also included in this study to correct for cross-sectional independence, the method of which is outlined by Hoechle (2007) and used by Perez (2013) in his analysis of social media and the Spanish soccer league.

Results and Discussion Table 3 presents the summary statistics from the current study. Table 3 displays that the final sample size contains 11,223 team-day observations. There is a reduction of 29 observations from the total potential team-day observa- tions due to missing values for the number of tweets in the previous 24 hr. As Table 3 reveals, it is observed that on average there is an increase of 272 Twitter followers a day for each MLB team. Furthermore, the examination of the dependent variable also discloses a reduction in Twitter followers at certain points in time. Hence, teams may not seek growth of the number of Twitter followers on a daily basis; rather, they may seek long-term gains in Twitter followers. Thus, MLB teams average 21 Tweets over the course of the previous 24 hr, whereas the teams have been on Twitter for an average of 1,695 days (4.6 years). In summary, the data reveals that there is a game on the day before in 47% of the observations. In addi- tion, it also reveals that 10% of observations comprise a winning streak of 2 or 3 games in length, and that 3% are of a winning streak of 4 or more games. Finally, 10% of observations comprise a losing streak of 2 or 3 games, and 4% are of a losing streak of 4 or more games. Con- sidering other Twitter-based interactions pertaining to MLB, teams average 6 Favorites a day and Follow only 1 account every 2 days. While accounts have some large

Table 3 Summary Statistics (n = 11,223)

Variable M SD Min Max ΔFollowers 271.653 441.460 –854 20,498

Tweets 20.850 23.977 –4 385

Favorites 5.713 23.006 –238 844

Followed 0.547 5.840 –165 347

TwitterDays 1,695 140 1,109 2,007

Followers(t–1) 291,561 227,243 83,992 1,176,128

Gameday(t–1) 0.468 0.499 0 1

PlayoffPart 0.008 0.091 0 1

WSPart 0.001 0.038 0 1

DiffRank 0.044 0.537 –5 5

WinStrk2 0.100 0.301 0 1

WinStrk4 0.031 0.174 0 1

LossStrk2 0.096 0.295 0 1

LossStrk4 0.035 0.183 0 1

Draft 0.008 0.088 0 1

AllStarGame 0.005 0.072 0 1

TradeDeadline 0.003 0.051 0 1

RosterExp 0.003 0.051 0 1

Holiday 0.026 0.159 0 1

WinterMeet 0.010 0.101 0 1

FreeAgency 0.003 0.051 0 1

NatlTV 0.026 0.159 0 1

SmallMkt 0.395 0.489 0 1

Sunday 0.145 0.352 0 1

Monday 0.142 0.350 0 1

Tuesday 0.142 0.350 0 1

Wednesday 0.142 0.350 0 1

Thursday 0.142 0.350 0 1

Friday 0.142 0.350 0 1

Saturday 0.142 0.350 0 1

January 0.080 0.272 0 1

February 0.073 0.259 0 1

March 0.080 0.272 0 1

April 0.078 0.268 0 1

May 0.080 0.272 0 1

June 0.078 0.268 0 1

July 0.135 0.341 0 1

August 0.080 0.272 0 1

September 0.078 0.268 0 1

October 0.080 0.272 0 1

November 0.078 0.268 0 1

December 0.080 0.272 0 1

Major League Baseball and Twitter Usage 627

JSM Vol. 29, No. 6, 2015

spikes in the number of Tweets they favorite or follow, generally there is a relatively low amount of activity over a 24-hr period examination.

Table 4 presents the regression results for Equation 2. It is found that the number of Tweets in the past 24 hr has a positive effect on the number of followers. Spe- cifically, the estimated coefficient for Tweets indicates that for every time a team produces a Tweet, they have an increase of about five followers in a 1-day period. As such, this result echoes the previous finding by Jensen et al. (2014), which reveals a positive connection between Twitter interest and the frequency of Tweets from a sport entity (Jensen et al., 2014). However, the number of favorites has no impact on the gain of followers. Further- more, the days on Twitter does not impact the number of people who follow the team. Importantly, the number of total team followers on the previous day has a negative and significant effect on the number of followers in the past 24 hr. In regards to Twitter interactions captured in this model, the indication is that small interactions in the form of favoriting a Tweet or Following other users does not draw people to an account; rather, it is the creation of content in the form of posting Tweets that is of interest to individuals. Thus, as MLB teams try to draw more followers to social media, it is important for them to consider the ramifications of managing social media accounts. That is, the results reveal that a certain type of interactions, such as producing more Tweets, is beneficial in regards to gaining more followers. It is from this perspective that the estimated results presented in Table 4 shall be considered as providing guidance toward strategic management of MLB team accounts to obtain more attention and exposure in the realm of social media.

In examining the team performance variables, it is revealed that the status of participating in the playoffs or World Series has a positive increase in the number of followers over the past 24 hr. According to the model, for each game a team plays in the playoffs, an average increase of 433 Twitter followers is received. Meanwhile, it witnesses a greater jump of almost 2,000 followers each day the team plays in the World Series. These results supported the findings presented by Perez (2013), where the connection between team performance and Twitter followers is emphasized. In addition, in a comprehen- sive view of economic demand, this result also supports Soebbing’s (2008) MLB attendance study, which reveals an increase of fan attendance during the playoffs (Soeb- bing, 2008).

Furthermore, the results reveal that moving up or down in the divisional standings as well as day-to-day performance does not impact the number of Twitter followers. However, a team with a losing streak of four or more games has a significant decline in Twitter fol- lowers over the past 24 hr, and a team with a winning streak of four or more games had significant increases. As such, it echoes the previous literature that fans are sensitive to longer winning and losing streaks (Fort & Rosenman, 1999). From a practical standpoint, the results

indicate that individual games do not have relation with the number of Twitter followers an MLB team account gained; rather, it is larger factors such as winning/losing streaks, playoffs, and being in the World Series that

Table 4 Regression Results

Variable Coefficient SE p-value

Tweets 4.590 1.036 <0.001

Favorites 0.111 0.309 0.723

Followed –1.211 0.605 0.055

TwitterDays 0.085 0.152 0.581

Followers(t–1) –0.001 0.000 0.006

Gameday(t–1) 24.243 32.482 0.462

PlayoffPart 433.870 176.276 0.020

WSPart 1,996.647 308.039 <0.001

DiffRank 21.617 14.810 0.156

WinStrk2 26.692 30.480 0.389

WinStrk4 65.062 35.295 0.076

LossStrk2 –32.862 19.537 0.104

LossStrk4 –47.511 20.131 0.025

Draft 7.149 24.432 0.772

AllStarGame 41.626 127.787 0.747

TradeDeadline 253.522 33.422 <0.001

RosterExp 18.750 47.932 0.699

Holiday –17.491 19.510 0.378

WinterMeet 0.301 18.442 0.987

FreeAgency 93.525 15.139 <0.001

NatlTV 349.551 123.850 0.009

SmallMkt 39.099 7.745 <0.001

Monday 78.004 24.038 0.003

Tuesday 95.057 41.518 0.030

Wednesday 49.439 18.323 0.012

Thursday 91.689 25.701 0.001

Friday 65.944 19.618 0.002

Saturday 35.609 15.684 0.031

February 64.298 10.835 <0.001

March 79.684 23.982 0.002

April 161.371 86.368 0.072

May –60.152 47.511 0.216

June –63.774 53.077 0.240

July –22.621 51.230 0.662

August 5.760 50.989 0.911

September 16.771 63.908 0.795

October 20.922 29.345 0.482

November –42.572 14.375 0.006

December –31.653 15.266 0.047

Β0 322.513 210.685 0.137

628 Watanabe, Yan, and Soebbing

JSM Vol. 29, No. 6, 2015

cause changes. While these are not issues that an MLB team can necessarily control, it is important for teams to consider the role that team performance plays in drawing fan interest on Twitter.

Baimbridge (1997) notes that certain events of sport team/league management and their associated time periods are connected to an increased demand for sport products. In this study, the factors that are distinguished as significant in gaining MLB Twitter followers are the trade deadline and the first day of free agency. By con- trast, the factors All Star game, the expansion of rosters on September 1st, and the winter meetings, as well as federal holidays are not revealed as producing significant impact on MLB social media demand. In an overall view of the patterns of demand across the year, it is witnessed that all days of the week excluding the reference day (Sunday) have a positive relation with the number of Twit- ter follows. In addition, possibly due to spring training, there is a positive increase in February, March, and April compared with the reference month (January). A negative and significant effect is found for the months of Novem- ber and December compared with the reference month. This might be due to baseball being in the offseason with the emergence of rival sport competitions, including the NFL playoffs, college football coming to the close of its regular season, and the beginning of men’s college bas- ketball. Importantly, in the sports economics literature, variables capturing time frames have mixed results, due to the different nature of various sport products (Borland & Macdonald, 2003; Buraimo, 2008). Thus, the results presented above reflect the standard expectations of the connection between time frame and sport demand.

In regards to nationally televised games, the model estimates a positive and significant relationship with an increase in the number of Twitter followers. That is, this coefficient reveals that the broadcast of a game on specific channels increases Twitter following by around 350 users in a day. In other words, when teams are featured across the country, they receive higher fan interest on Twitter, which is similar to what Buraimo (2008) finds in examin- ing demand for sport television viewership.

Furthermore, the small market variable is also positive and significant, indicating that teams playing in regions with smaller populations have greater increases in Twitter followers. Generally, the literature supports that larger market teams tend to have higher attendance because of population size (Jewell & Molina, 2005; Soe- bbing, 2008). However, it is also noted that employing different regional definitions in a model can potentially change the results (Watanabe, 2012). Thus, the findings of the present research can be attributed to the fact that small market teams may do a better job in attracting fans to their product, or that MLB social media users exhibit different behaviors than those who attend games in person. In addition, because there is no price charged to use Twitter, it may be an indication that small market teams have identified Twitter and other social media platforms as cheaper ways to market to fans. Thus, the relatively lower cost of running a social media campaign

may be of more importance to smaller teams with lower budgets to entice fans.

Conclusion, Limitations, and Future Directions

According to the New York Times (Shpigel, 2014), sport franchises are “widening their embrace of social media, employing people to find creative ways to inform and connect with their fan bases on platforms like Facebook, Instagram and Twitter” (Shpigel, 2014, para. 5). It has thus become vital for teams/leagues to further develop methods to manage media messages and employees work- ing with these platforms. In the meantime, even though the benefits of well-managed social media accounts are evident, franchises still grapple with difficulties in deal- ing with variations and behaviors when trying to obtain fan interest. As such, the findings from this study enable sport franchises to accomplish better understanding and strategic focus of managing social media.

Specifically, the results lead to four major folds of managerial implications. To begin with, the model indi- cates that it is through the means of developing media messages on the platform of Twitter, instead of taking simple actions such as Retweeting or Favoriting, where sport franchises foster fan interest. As such, it sets priori- ties for teams to focus on developing the digital content of social media by looking further into designing and tailor- ing messages. Secondly, teams that appear in the playoffs, World Series, or have long winning/losing streaks witness a drastic change in the number of followers in a 24-hr period. In this light, teams should strategically prepare social media content to take advantage the time period when there is good performance. In facing difficult times of losing streaks, the findings signify that managers need to take remedial actions, providing strategies in develop- ing social media messages for fans when they are most likely to leave an account. Furthermore, the fact that teams that make the playoffs or World Series witness a large increase in Twitter Followers each day they are playing in these series presents an elongated period for franchises to attract more fans to their digital content. It is thus vital to prepare social media campaigns for these times in order to maximize the benefits from teams making the postseason. In addition, further expansion of the playoffs, for example, may be considered by the entire league to garner more fan interest for a larger group of teams across all their products and platforms each year.

Thirdly, the importance of scheduling events in drawing fans to MLB Twitter accounts must also be highlighted. In particular, the appearance of teams in nationally televised games and the association with a large gain in Followers points at how crucial time man- agement is for social media accounts for MLB teams. It is possible for team managers to take advantage of the timing of broadcasts and integrate strategic messages to enhance the overall interest from fans. However, consid- ering MLB’s recent development of its own exclusive

Major League Baseball and Twitter Usage 629

JSM Vol. 29, No. 6, 2015

media channels, it is advised that MLB take measures to reevaluate potential consequences of reducing the number of games on national channels in relation to fan demand (Budzinski & Satzer, 2011). Furthermore, in viewing the timing of certain calendar events which play a signifi- cant role in creating attention on Twitter—including the beginning of Free Agency and the Trade Deadline—MLB teams should strategically focus on these events and time frames in developing social media messages. Fourthly, the model delineates that smaller market teams are able to draw more fans to Twitter accounts, which indicates that social media may be a cost-effective way for low- revenue franchises to assist in managing their overall marketing campaigns in various media channels. Still, it is possible that larger market teams may not have used social media to its full potential yet. Thus, league-wide policies to enhance social media use would benefit all teams and fulfill multiple organizational goals (Hutchins & Rowe, 2012).

In discussing this article’s contribution to the aca- demic literature, the incorporation of economic demand theory as well as econometric modeling provides a new venue toward a more comprehensive understanding of sport social media communications. In so doing, it raises awareness for future research to go beyond the current dominant frameworks, and to instead interpret and analyze sport social media from broader and more comprehensive approaches (Hutchins, 2014; Rowe, 2011). Furthermore, this study also aims to contribute to the sports economics literature from multiple perspec- tives. Firstly, the inclusion of social media as a critical form of derived sport product provides a new context in examining the economic demand for sport. Consider- ing the volume and growth of digital media in the sport marketplace (Budzinski & Satzer, 2011), it naturally speaks of the importance in exploring various variables and understanding this emerging form of sport demand. More importantly, the examination of sport social media allows for analysis as to whether determinants are con- gruent across different forms of sport products, which contributes to providing an overall enhanced view of economic demand theory. Secondly, the size and scope of data collection in this study has also contributed to the economic modeling of social media research. Com- pared with the previous studies that have focused only on social media data collection at a single point in time (Jensen et al., 2014) or at weekly intervals over a several month period (Perez, 2013), this study has developed a more nuanced account by collecting daily changes in the Twitter followers of all MLB teams. As such, it reveals changes and usage of social media at a relatively micro and dynamic level. That is, as noted by van Dijck (2009) in proposing future research of social media, “such a multifaceted concept needs to be met with proposals for multi-leveled methodologies . . . with analysis charting techno-economic aspects of media use” (p. 55).

There are potential limitations that must also be acknowledged in this research. Considering that the results have pointed out how important it is for teams

develop social media message to engage fans, it encour- ages future research to incorporate actual social media messages into examination. That is, by delineating the narrative construction as well as rhetorical positioning of messages created by teams from perspectives of con- sumption and demand, it enables examination of complex social interactions in the format of multiple methods of inquiry (Sale, Lohfeld, & Brazil, 2002) as part of a con- tinuum in understanding behaviors (Casebeer & Verhoef, 1997). And from an economic inquiry perspective, one limitation that arises in dealing with official social media accounts for teams is the ability to delineate the difference of an individual who is following the account because of their interest in a team, and the interest in a single or sev- eral players on the team. It is thus meaningful for future studies to further differentiate how fans follow either a team, an athlete, or both. Another potential limitation is the conception of whether a significant portion of fan interest is being captured by examining the followers of the official social media account of a team. In specific, the competing social media accounts, such as popular fan-generated accounts that also draw interest from fans, may be mistaken by fans as being official accounts associ- ated with the franchise. Therefore, it is possible that the followers of an official account may not capture all of the potential individuals who are on social media at least partly to interact with the team. As such, this provides another direction into which future research can look.

Notes 1. The U.S. Federal holidays coded in the sample are New Year’s Day, Martin Luther King, Jr. Day, President’s Day, Memorial Day, Independence Day, Labor Day, Columbus Day, Veterans Day, Thanksgiving Day, Christmas Day.

2. The study attempted to include separate dummy variables for spring training, playoffs, and regular season. However, these variables were all highly correlated with some of the other explanatory variables. Thus, the aforementioned variables were removed from Equation 2.

References About Twitter. (2014, September 21). Retrieved from https://

twitter.com/about Anderson, C. (2008). The long tail. New York: Hyperion. Asteriou, D., & Hall, S.G. (2011). Applied econometrics. New

York: Palgrave Macmillan. Baimbridge, M. (1997). Match attendance at Euro 96: was the

crowd waving or drowning? Applied Economics Letters, 4(9), 555–558. doi:10.1080/135048597355014

Baltagi, W. (2008). Econometric analysis of panel data. Chich- ester, UK: Wiley.

Billings, A. (2014). Power in the reverberation why Twitter matters, but not the way most believe. Communication & Sport, 2, 107–112. doi:10.1177/2167479514527427

Bird, P.J. (1982). The demand for league football. Applied Eco- nomics, 14(6), 637–649. doi:10.1080/00036848200000038

630 Watanabe, Yan, and Soebbing

JSM Vol. 29, No. 6, 2015

Brown, W.O., & Sauer, R.D. (1993). Does the basketball market believe in the hot hand? Comment. The American Economic Review, 83, 1377–1386.

Borland, J., & MacDonald, R. (2003). Demand for sport. Oxford Review of Economic Policy, 19, 478–502. doi:10.1093/ oxrep/19.4.478

Boyd, D., & Ellison, N.B. (2007). Social network sites: Defini- tion, history and scholarship. Journal of Computer-Medi- ated Communication, 13, 210–230. doi:10.1111/j.1083- 6101.2007.00393.x

Bruggink, T.H., & Eaton, J.W. (1996). Rebuilding attendance in Major League Baseball: The demand for individual games. In J. Fizel, E. Gustafsen, & L. Hadley (Eds.), Baseball economics: Current research (pp. 9–31). Westport, CT: Praeger.

Budzinski, O., & Satzer, J. (2011). Sports business and mul- tisided markets: Towards a new analytical framework? Sport, Business, and Management: An International Jour- nal (Toronto, Ont.), 1, 124–137. http://dx.doi.org/10.2139/ ssrn.1237014

Buraimo, B. (2008). Stadium attendance and television audience demand in English league football. Managerial and Deci- sion Economics, 29(6), 513–523. doi:10.1002/mde.1421

Buraimo, B., & Simmons, R. (2009). A tale of two audiences: Spectators, television viewers and outcome uncertainty in Spanish football. Journal of Economics and Business, 61, 326–338. doi:10.1016/j.jeconbus.2008.10.002

Carmichael, F., Millington, J., & Simmons, R. (1999). Elasticity of demand for Rugby League attendance and the impact of BSkyB. Applied Economics, 6, 797–800.

Casebeer, A.L., & Verhoef, M.J. (1997). Combining qualita- tive and quantitative research methods: Considering the possibilities for enhancing the study of chronic diseases. Chronic Diseases in Canada, 18(3), 130–135.

Cavallo, A. (2011, January 27). Scraped data and sticky prices. doi:10.7910/DVN/IAH6Z6

Clavio, G. (2011). Social media and the college football audi- ence. Journal of Issues in Intercollegiate Athletics, 4, 309–325.

Clavio, G., & Kian, T.M. (2010). Uses and gratifications of a retired female athlete’s Twitter followers. International Journal of Sport Communication, 3, 485–500.

Clavio, G., & Walsh, P. (2014). Dimensions of social media utilization among college sport fans. Communication & Sport, 2, 261–281. doi:10.1177/2167479513480355

Coates, D., & Humphreys, B.R. (2010). Week to week atten- dance and competitive balance in the National Football League. International Journal of Sport Finance, 5, 239–252.

Coates, D., & Humphreys, B.R. (2012). Game attendance and outcome uncertainty in the National Hockey League. Journal of Sports Economics, 13, 364–377. doi:10.1177/1527002512450260

Constantinides, E., & Fountain, S.J. (2008). Web 2.0: Con- ceptual foundations and marketing issues. Journal of Direct. Data and Digital Marketing Practice, 9, 231–244. doi:10.1057/palgrave.dddmp.4350098

Driscoll, J., & Kraay, A.C. (1998). Consistent covariance matrix estimation with spatially dependent data. The

Review of Economics and Statistics, 80, 549–560. doi:10.1162/003465398557825

Dwyer, B., & Drayer, J. (2010). Fantasy sport consumer seg- mentation: An investigation into the differing consumption modes of fantasy football participants. Sport Marketing Quarterly 19, 207–216.

Edelman, B. (2012). Using Internet data for economic research. The Journal of Economic Perspectives, 26, 189–206. doi:10.1257/jep.26.2.189

Ellison, G., & Ellison, S.F. (2009). Tax sensitivity and home state preferences in internet purchasing. American Eco- nomic Journal. Economic Policy, 1, 53–71. doi:10.1257/ pol.1.2.53

Feddersen, A., Humphreys, B.R., & Soebbing, B.P. (2013). Sentiment bias and asset prices: Evidence from sports betting markets and social media. Available at http://www. be.wvu.edu/phd_economics/pdf/13-07.pdf

Feddersen, A., & Rott, A. (2011). Determinants of demand for televised live football: Features of the German national football team. Journal of Sports Economics, 12, 352–369. doi:10.1177/1527002511404783

Fisher, E. (2008). Social perspective. Sport Business Journal, 11(29), 15–23.

Fort, R., & Rosenman, R. (1999). Streak management. In J. Fizel, E. Gustafson, & L. Hadley (Eds.), Sports eco- nomics: Current research (pp. 119–134). Westport, CT: Praeger.

Garcia, J., & Rodriguez, P. (2002). The determinants of football match attendance revisited: Empirical evidence from the Spanish football league. Journal of Sports Economics, 3, 18–38. doi:10.1177/152700250200300103

Gitter, S.R., & Rhoads, T.A. (2010). Determinants of minor league baseball attendance. Journal of Sports Economics, 11, 614–628. doi:10.1177/1527002509359758

Gujarati, D.N. (2003). Basic econometrics (4th ed.). New York, NY: McGraw-Hill/Irwin.

Hambrick, M.E., Simmons, J.M., Greenhalgh, G.P., & Green- well, T.C. (2010). Understanding professional athletes’ use of Twitter: A content analysis of athlete tweets. Inter- national Journal of Sport Communication, 3, 454–471.

Hansen, D., Shneiderman, B., & Smith, M.A. (2010). Analyz- ing social media networks with NodeXL: Insights from a connected world. Morgan Kaufmann.

Hardin, M. (2014). Moving beyond description: Putting Twit- ter in (theoretical) context. Communication & Sport, 2, 113–116. doi:10.1177/2167479514527425

Hoechle, D. (2007). Robust standard errors for panel regres- sions with cross-sectional dependence. The Stata Journal, 7, 281–312.

Hutchins, B. (2011). The acceleration of media sport culture: Twitter, telepresence and online messaging. Information Communication and Society, 14, 237–257. doi:10.1080/1 369118X.2010.508534

Hutchins, B. (2014). Twitter: Follow the money and look beyond sports. Communication & Sport, 2, 122–126. doi:10.1177/2167479514527430

Hutchins, B., & Rowe, D. (Eds.). (2012). Sport beyond televi- sion: The Internet, digital media and the rise of networked media sport (Vol. 40). Routledge.

Major League Baseball and Twitter Usage 631

JSM Vol. 29, No. 6, 2015

Jensen, J.A., Ervin, S.M., & Dittmore, S.W. (2014). Exploring the factors affecting popularity in social media: A case study of Football Bowl Subdivision head coaches. Inter- national Journal of Sport Communication, 7, 261–278. doi:10.1123/IJSC.2014-0008

Jewell, R.T., & Molina, D.J. (2005). An evaluation of the relationship between Hispanics and Major League Soccer. Journal of Sports Economics, 6, 160–177. doi:10.1177/1527002504263400

Kasavana, M.L., Nusair, K., & Teodosic, K. (2010). Online social networking: redefining the human web. Jour- nal of Hospitality and Tourism Technology, 1, 68–82. doi:10.1108/17579881011023025

Kassing, J.W., & Sanderson, J. (2010). Fan-athlete interaction and Twitter tweeting through the Giro: A case study. Inter- national Journal of Sport Communication, 3, 113–128.

Leonard, D.J. (2009). New media and global sporting cultures: Moving beyond the clichés and binaries Editorial, Special Issue. Sociology of Sport Journal, 3, 113–128.

Lovejoy, K., Waters, R.D., & Saxton, G.D. (2012). Engaging stakeholders through Twitter: How nonprofit organiza- tions are getting more out of 140 characters or less. Public Relations Review, 38, 313–318. doi:10.1016/j. pubrev.2012.01.005

Mongeon, K., & Winfree, J. (2012). Comparison of television and gate demand in the National Basketball Association. Sport Management Review, 15, 72–79. doi:10.1016/j. smr.2011.09.001

Neale, W.C. (1964). The peculiar economics of professional sports: A contribution to the theory of the firm in sporting competition and in market competition. The Quarterly Journal of Economics, 78, 1–14. doi:10.2307/1880543

Pedersen, P.M. (2012). Reflections on communication and sport: On strategic communication and management. Communi- cation & Sport, 1, 55–67. doi:10.1177/2167479512466655

Pedersen, P.M. (2014). A commentary on social media research from the perspective of a sport communication journal editor. Communication & Sport, 2, 138–142. doi:10.1177/2167479514527428

Pegoraro, A. (2010). Look who’s talking—Athletes on Twitter: A case study. International Journal of Sport Communica- tion, 3, 501–514.

Pegoraro, A. (2013). Sport fandom in the digital world. In P.M. Pedersen (Ed.), Routledge handbook of sport communica- tion (pp. 248–258). New York, NY: Routledge.

Perez, L. (2013). What drives the number of new Twitter fol- lowers? An economic note and a case study of professional soccer teams. Economic Bulletin, 33, 1941–1947.

Rottenberg, S. (1956). The baseball players’ labor market. Journal of Political Economy, 64(3), 242–258. doi:10.1086/257790

Rowe, D. (2011). Global media sport: Flows, forms and futures. New York, NY: Bloomsbury Academic.

Rowe, D. (2014). Following the followers sport researchers’ labour lost in the Twittersphere? Communication & Sport, 2, 117–121. doi:10.1177/2167479514527431

Sale, J.E., Lohfeld, L.H., & Brazil, K. (2002). Revisiting the quantitative-qualitative debate: Implications for mixed- methods research. Quality & Quantity, 36(1), 43–53. doi:10.1023/A:1014301607592

Sanderson, J. (2011). To tweet or not to tweet: Exploring Divi- sion I athletic departments’ social-media policies. Inter- national Journal of Sport Communication, 4, 492–513.

Sanderson, J. (2014). What do we do with Twitter? Communica- tion & Sport, 2, 127–131. doi:10.1177/2167479514527429

Seo, W.J., & Green, B.C. (2008). Development of the motiva- tion scale for sport online consumption. Journal of Sport Management, 22(1), 82–109.

Shapiro, S.L., & Drayer, J. (2012). A new age of demand- based pricing: An examination of dynamic ticket pricing and secondary market prices in Major League Baseball. Journal of Sport Management, 26, 532–546.

Sharma, P., Khurana, U., Shneiderman, B., Scharrenbroich, M., & Locke, J. (2011). Speeding up network layout and centrality measures for social computing goals. In Social computing, behavioral-cultural modeling and prediction (pp. 244–251). Springer Berlin Heidelberg. doi:10.1007/978-3-642-19656-0_35

Shpigel, B. (2014, November 14). Win or lose on Sunday, Social media is a draw. The New York Times. Retrieved from http://www.nytimes.com/2014/11/16/sports/as-nfl- teams-exploit-social-media-fans-respond-not-always-in- kind.html

Shoham, A., & Kahle, L.R. (1996). Spectators, viewers, readers: Communication and consumption communities in sport marketing. Sport Marketing Quarterly, 5, 11–20.

Soebbing, B.P. (2008). Competitive balance and attendance in Major League Baseball: An empirical test of the uncer- tainty of outcome hypothesis. International Journal of Sport Finance, 3, 119–126.

Stavros, C., Meng, M.D., Westberg, K., & Farrelly, F. (2014).Understanding fan motivation for interacting on social media. Sport Management Review, 17, 455–469. doi:10.1016/j.smr.2013.11.004

Tabachnick, B.G., & Fidell, L.S. (2007). Using multivariate statistics (5th ed.). Boston: Pearson Education Inc.

Tainsky, S., & McEvoy, C.D. (2012). Television broadcast demand in markets without local teams. Journal of Sports Economics, 13, 250–265. doi:10.1177/1527002511406129

Tainsky, S., Salaga, S., & Santos, C.A. (2012). Estimating attendance for the Ultimate Fighting Championship: a demand theory approach. International Journal of Sport Management and Marketing, 11, 206–224. doi:10.1504/ IJSMM.2012.047130

Tainsky, S., Salaga, S., & Santos, C.A. (2013). Determinants of pay-per-view broadcast viewership in sports: The case of the Ultimate Fighting Championship. Journal of Sport Management, 27, 43–58.

Tainsky, S., Xu, J., & Zhou, Y. (2014). Qualifying the game uncertainty effect a game-level analysis of NFL postsea- son broadcast ratings. Journal of Sports Economics, 15, 219–236. doi:10.1177/1527002512457946

Van Dijck, J. (2009). Users like you? Theorizing agency in user-generated content. Media Culture & Society, 31(1), 41–58. doi:10.1177/0163443708098245

Watanabe, N.M. (2012). Demand for pay-per-view consump- tion of Ultimate Fighting Championship events. Interna- tional Journal of Sport Management and Marketing, 11, 225–238. doi:10.1504/IJSMM.2012.047128

632 Watanabe, Yan, and Soebbing

JSM Vol. 29, No. 6, 2015

Watanabe, N.M., Soebbing, B.P., & Wicker, P. (2013). Exam- ining the impact of the StubHub agreement on price dispersion in Major League Baseball. Sport Marketing Quarterly, 22, 129–137.

Wenner, L.A. (2014). Much Ado (or not) about Twitter? Assessing an emergent communication and sport research agenda. Communication & Sport, 2, 103–106. doi:10.1177/2167479514527426

Williams, J., & Chinn, S.J. (2010). Meeting relationship- marketing goals through social media: a conceptual model for sport marketers. International Journal of Sport Com- munication, 3, 422–437.

Yoo, S.K., Smith, L.R., & Kim, D. (2013). Communication theories and sport studies. In P.D. Pedersen (Ed.), The Routledge handbook of sport communication (pp. 8–19). New York, NY: Routledge.

Copyright of Journal of Sport Management is the property of Human Kinetics Publishers, Inc. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

View publication statsView publication stats