Content uploaded by Jeffrey Cisyk
Author content
All content in this area was uploaded by Jeffrey Cisyk on May 01, 2023
Content may be subject to copyright.
Coolerbobbleheadsprevail:bobbleheadcharactertypeandimpacton
attendanceatMajorLeagueBaseballgames
Giveaways such as jerseys, caps, or bobbleheads are part of a host of promotions used to entice fans to
purchase tickets and attend live events, particularly in sports. Bobbleheads have often been viewed as the
bellwether giveaway in demand-side analytical studies as they are common across sport, league, and team, and
are viewed as highly prized collectable items. Empiricists typically code promotions as binary variables to
measure the impact of these additional perks have on attendance, yet each bobblehead event features a distinct
figurine distinguished by several previously unexplored dimensions. Due to the nature of bobbleheads typically
resembling real-life individuals, consumer sentiment towards the bobblehead’s namesake may drive or deter
the demand for the giveaway. Using a novel dataset of descriptions of bobblehead giveaways, this study finds
that demand for Major League Baseball games varies by a bobblehead’s character type, where the impact on
attendance is greatest for bobbleheads featuring players and sports-casters, lowest for managers, and middling
for mascots and other types. Evidence suggests that bobbleheads depicted in costumes or as characters of
licensed brands lead to greater attendance than brand-less bobbleheads, yet the gains observed from the
average branded bobblehead game may not off-set the additional costs associated with licensing agreements.
1
Introduction
On the day of the 2009 Major League Baseball (MLB) trade deadline, the Cleveland baseball team sent
one of its top players, Victor Martinez, to the Boston Red Sox in exchange for younger prospects and players.
While this was seen as a prudent business decision for a Cleveland team that had little hope of winning a World
Series that season, it made for an embarrassing situation when its fans were given Victor Martinez bobbleheads
as a souvenir for attending the following evening’s game (the embarrassment only compounded itself when the
team featured a second Martinez-themed giveaway a week later). While fans largely still attended the games
and still collected the giveaway, it is not hard to imagine that these promotions did not have the desired effect
as planned when Victor Martinez had still been a member of the team.
Basic economic rationale can be used to explain why teams use stadium giveaways: enticing consumers
to make ticket purchases for events they would otherwise forgo. Much time and effort are afforded by
marketing practitioners deciding upon the particulars of a giveaway. Past studies have explored when teams
should schedule promotional-giveaway games (e.g., Kappe et al, 2014), what should be given away (e.g., Barilla
et al., 2008), and howmany items should be made available (Cisyk & Courty, 2021). Results suggest that teams
trying to maximise ticket purchases should offer bobblehead giveaways (Boyd & Krehbiel, 2006; Barilla et al.,
2008; Paul et al., 2019); on weekdays (Barilla et al., 2008); during night-, not day-, games (Boyd & Krehbiel,
2006); and restrict the number of items distributed to approximately 40 percent of the stadium (Cisyk & Courty,
2021). Yet, if a bobblehead is scheduled as the promotional item, one key decision made by the team remains
unexplored by prior research: who should the bobblehead resemble?
The purpose of this study is to determine whether there is a heterogeneous demand-response to
stadium giveaway promotions based on differentiable characteristics. This is accomplished by examining the
impact of bobblehead giveaways at MLB games based on the bobblehead’s name-sake (i.e., the persona it is
designed to resemble). If the design of the bobblehead should impact the demand response from consumers,
previous studies treating promotional giveaways as zero-one binary variables are limited to capturing the
average effect and demand estimates may suffer from an omitted variable bias. The findings of this study have
both theoretical and practical implications as future studies and/or applications should consider additional
qualities of a giveaway to produce more statistically efficient estimates of a promotion’s impact and to execute
more efficient promotional schedules. This study, in fact, uncovers a varied demand-response based on the
characteristics of the figurine. By using MLB attendance data as a proxy for demand and parsing through
descriptions of bobblehead games from 2012 to 2019, player and sports-caster bobblehead types are found to
have the greatest impact on attendance, while manager types have the lowest, and mascots and other types fall
in-between. Evidence suggests that some themes—bobbleheads featuring branded characters and/or
costumes—has a positive impact on ticket sales, yet large variation in the estimates may caution risk-averse
decision-makers to avoid the additional costs associated with the licensing fees required to use these themes.
Other demand- and supply-side issues are examined and applicable conclusions for marketing decision-makers
are discussed.
2
BackgroundandMotivation
Promotions for live events are broadly divided into two categories, namely price and non-price
promotions. Price promotions provide discounts on ticket prices to events for all or for select consumers (e.g.,
senior-citizen’s day) whereas non-price promotions offer some additional benefit with purchase of a ticket
beyond the event itself. Non-price promotions include many sub-categories such as pre-game concerts, post-
event fireworks, or giveaways. The use of a non-price promotion is a form of price discrimination:
understanding not all potential buyers value a product identically, a seller may offer a menu of options designed
to have consumers self-select the product (or version thereof) at a price that they find most preferable
(Rascher & Schwarz, 2012). Within this framework, Cisyk and Courty (2021) describe two types of consumers
for live sporting events: unconditional and conditional fans. The former segment will attend a sporting event
regardless of a promotion, while the latter requires some additional benefit to motivate their purchase. The
goal of the promotion is to attract these conditional fans while also avoiding the crowding out that occurs when
from unconditional fans selectively attend games coinciding with promotions. Such an effect, referred to as
temporal substitution or cannibalisation, is of great concern for promotions as first noted by Siegfried &
Eisenberg (1980): if consumers attend a fixed number of events and choose to do so only when a promotion is
available, games with promotions may see a large influx of attendees, yet no additional tickets would be sold
over the course of the entire season.
Many choices present themselves to the empiricist when attempting to measure the impact of
promotions on attendance, perhaps best exemplified by the earliest two studies on the topic: Hill et al. (1982)
explores a promotion’s impact on game-specific attendance at MLB games in 1977 by use of a single binary
variable indicating the presence of any price and/or non-price promotion, while Siegfried & Eisenberg (1980)
explores aggregate season attendance of select Minor League Baseball (MiLB) teams from 1973 to 1977 with
two count variables for the number of price and, separately, non-price promotions used by the team across
entirety of the season. In both studies, promotions are found to increase the number of attendees by significant
amounts, yet numerous dimensions distinguish these studies from one another: the league, the breadth of
teams and time periods considered, the dependent variable, and how to encode promotions.
Few studies exist for other sports leagues—such as Major League Soccer (DeShriver, 2007), the
National Hockey League (Kelley et al., 2014), or the American Hockey League (Paul & Chatt, 2011)—much of
the existing literature tends to focus on either MLB or MiLB as is further detailed below. Both MLB and its
minor-league affiliates are said to be the ideal case studies for promotional impacts on attendance (e.g., Boyd
& Krehbiel, 2006) as these leagues feature long seasons (MLB teams are scheduled to host 81 home games each
season) played across many months of the year (typically April to September) where the demand ebbs and
flows with distinct weather patterns and availability of substitute sporting events and Summer-time activities
(Barilla et al., 2008). Moreover, baseball games rarely sell out (Boyd & Krehbiel, 2006; Paul et al., 2007; Cisyk
& Courty, 2017) allowing the empiricist to observe the full effect of positive demand shocks unabated by
stadium capacity (Rascher & Schwarz, 2012; Cisyk, 2020). Most papers use game-day attendance (or a natural-
3
logarithmic transformation thereof) as its primary dependent variable, however notable exceptions include a
focus on season-aggregate attendance (Hill et al., 1982) or revenues (Cebula, 2009).
Several studies cover all teams within the league (Boyd & Krehbiel, 2006; Paul et al., 2007; Barilla et
al., 2008; Lemke, 2010; Paul & Weinbach, 2013), while others cover multiple seasons (Chupp et al., 2007;
Burnett & Van Scyoc, 2008; Kappe et al., 2014). Several studies consider heterogeneity in the demand for
promotions across MiLB leagues (Anthony et al. 2011; Paul & Weinbach 2011; 2013). Two studies take a
narrow approach and focus on few teams and single seasons (Marcum & Greenstein, 1985 and Paul et al., 2019
each use just two teams and Boyd & Krehbiel, 2003 uses six). Some studies make use of panel datasets (Siegfried
& Eisenberg, 1980; Boyd & Krehbiel, 1999; Cebula, 2013; Cisyk & Courty, 2021).
Much variation exists in the way to encode qualitative promotional information. Some use a single
binary variable to estimate a common impact of price and non-price promotions (Hill et al., 1982; Boyd &
Krehbiel, 2003; 2006). Due to vast changes in ticket pricing strategies rendering price promotions less common
(Cisyk & Courty, 2021), many studies choose to focus on non-price promotions (Chupp et al., 2007; Lemke,
2010; Kappe et al., 2014) or solely on giveaways (Marcum & Greenstein, 1985; McDonald & Rascher, 2000;
Barilla et al., 2008; Cisyk & Courty, 2021). Giveaways can be disaggregated by item (e.g., estimating separate
impacts from caps, shirts, bobbleheads, etc.; Barilla et al., 2008) or by high- versus low-value (Marcum &
Greenstein, 1985; Burggink & Eaton, 1996; Cebula, 2009; 2013; Cebula et al., 2009; 2013; Howell et al., 2015).1
Of the studies that disaggregate giveaways by item, bobbleheads are repeatedly found to have the
greatest impact on attendance (Boyd & Krehbiel, 2006; Barilla et al., 2008; Paul et al., 2019). Bobbleheads are
said to be an excellent bellwether giveaway as they are common across sport, league, and team, and are viewed
as highly prized collectable items and the sole focus of the demand estimation of Cisyk & Courty (2021).
The data considered in this study follows that of Cisyk & Courty (2021) to answer the research
question of whether the observable differences in promotions affect the demand response by using
bobbleheads as the exemplar promotion. As seen through previous studies, bobblehead giveaways consistently
have the largest impact on game-day attendance. Whether concerns of satiation (for a detailed discussion, see,
e.g., Paul & Weinbach, 2013) or budgetary constraints, the average MLB team hosts but five bobblehead games
during its 81-game schedule each season. The powerful influence on attendance combined with the relative
scarcity of these events gives each bobblehead game heightened importance to each team and every detail of
the giveaway should be given much consideration. In the next section, the impacts of each aspect of these
decisions are explored.
1 Marcum & Greenstein (1985) and Burggink & Eaton (1996) use the language of ‘major’ and ‘minor’ promotion with
but a vague description discerning the two. Of the five studies that use the terms high- and low-value, Howell et al. (2015) is the only study
that categorises bobbleheads as high-value. Boyd & Krehbiel (2006) distinguishes between three mutually exclusive giveaways categories:
a perceived value of greater than $5.00, of less than $5.00, and bobbleheads. McDonald & Rascher (2000), Burnett & Scyoc (2008), and
Kappe et al. (2014) use the promotion’s total cost as a continuous variable to proxy the promotions value and estimate a direct return on
investment.
4
DataandMethodology
Understanding the demand for a good or service is essential for its producer in any industry, and live
sporting events are no exception: a sports franchise attempts to maximise its profits as either a direct objective
or as a means to satisfy its other objectives, such as re-investing profits into its human capital (Schreyer &
Ansari, 2022). Demand for sport can be ascertained through different means, each with its own set of
advantages and limitations. For example, surveys can directly ask an individual’s willingness to pay for a
product but can be expensive and is not guaranteed to capture a representative sample of the consumers that
contribute to the revenues of a team (Cisyk & Courty, 2017). Instead, demand can be proxied by expenditures
or other quantifiable actions undertaken by consumers, such as the number of attendees at live games. Due to
its public availability, attendance has indeed become the standard for estimating the demand for sport (see
systematic reviews in, for example, Cairns et al., 1986; Borland & MacDonald, 2003; Garcia & Rodriguez, 2009;
Schreyer & Ansari, 2022). Yet, attendance, too, has its limitations, as it often is measured by paid attendance:
the reported number of tickets sold to a particular game, regardless of a ticket-holders decision to physically
attend. Despite still being represented in the paid attendance, an unused ticket, or ‘no-show,’ leaves the host
team without the ancillary revenue from catering, parking, etc. (Schreyer et al., 2019; Popp et al., 2023). In such
instances, the paid attendance will overstate the true demand for the game. Despite this disadvantage, paid
attendance remains ideal to assess this research question as the objective of a team and its management
remains to sell tickets (as is reflected by an increase in its paid attendance). Thus, information originally
sourced from MLB box scores is collected from Baseball-Reference.com for the 2012 to 2019 regular seasons
for all 30 MLB teams across eight regular seasons.2 This information includes the identity home and away team
of each game and the paid attendance (hereafter referred to simply as attendance).
The variables of interest are collected from three independent sources: Beckett Collectibles, Cardboard
Connection, and Stadium Giveaway Exchange.3 Each source lists the date of the bobblehead game, the number
of bobbleheads given away (availability), the identity of the bobblehead character (type), and, where
applicable, the licensed brand associated with the bobblehead (theme). In rare instances where information
differed between sources, additional research is undertaken, such as reviewing Twitter announcements made
by host team’s official account. Table 1 illustrates an example of a bobblehead schedule for the 2019 New York
Mets. Each of the seven scheduled bobblehead games have an availability of 25,000, corresponding to 60
percent of its home stadium’s capacity.4 Four of the bobblehead games feature then-current players, and one
2 This is a preliminary total of 19,440 games (30 81 8), yet some exceptions apply: the sample excludes (a) games
cancelled due to inclement weather and never re-scheduled (one less game per occurrence; N = 6); (b) games hosted in neutral sites due
to marketing initiatives (one less game per occurrence; N = 22); and (c) games cancelled due to extenuating circumstances (the Baltimore
Orioles closed one game to the public on April 29, 2015 due to protests and civil unrest; the Houston Astros played the Texas Rangers t hree
times in Tampa, Florida due to Hurricane Harvey; the Tampa Bay Rays played the New York Yankees three times in the stadium of the New
York Mets due to Hurricane Irma; N = 7).
3 Beckett Collectibles. www.beckett.com/; The Cardboard Connection. www.cardboardconnection.com/; and Stadium
Giveaway Exchange. www.stadiumgiveawayexchange.com/mlb-bobblehead-history/.
4 The home stadium of the New York Mets, Citi Field, has a capacity listed at 41,922 for the 2019 MLB regular season.
Availability is hereafter expressed solely as a ratio to the home team’s stadium capacity for ease of analysis and comparison across MLB
teams.
5
bobblehead game individually features one of each of the following: Mr. Met (the team’s mascot), Jerry Seinfeld
(local celebrity and known fan), and the Marvel character “Spider-Man.” Three of the bobbleheads use licensed
brands, where two players are depicted in the costumes of characters from Game of Thrones and Star Wars,
while one bobblehead, namely “Spider-Man,” is specifically depicted as the comic book character. As is common
with themed bobbleheads, the Star Wars figurine is marketed with play-on-words, mashing the names of the
branded character “Obi-Wan Kenobi” and the then-current player “Robinson Cano” to form “Obi-Wan Canobi.”
Table1:2019NewYorkMets’bobbleheadpromotionalschedule
All bobblehead games are reviewed and coded using mutually exclusive and collectively exhaustive
categories in an equivalent manner. The counts of type and theme of all 1,030 bobblehead games from the 2012
to 2019 regular seasons are displayed in Table 2. Each bobblehead is assigned a type of either player, manager,
mascot, sports-caster, or other.5 Themes include many licensed brands, typically of then-popular movies or
television series such as Star Wars, Marvel, Hello Kitty, or Game of Thrones. Due to low counts of themed-
bobblehead games, themes with fewer than ten observations are combined into the ‘other’ category for
estimation purposes. These characteristics are coded into mutually exclusive binary variables used in the
analysis described below: the summary statistics of these binary variables, along with the summary statistics
of all variables, are presented in Table A1.
Table2:Countsofbobbleheadgamesbycharactertypeandtheme,2012to2019regularseasons
5 Other includes characters such as local celebrities, fictional characters, generic baseball characters, or players that
were neither former nor then-current players of the home team. Players can be further classified into categories based on primary
defensive position. One convention would be to group players into three mutually exclusive and approximately equally sized categories of
pitchers, infielders, and other (pitchers include both starting and relief pitchers; infielders include non-pitchers that primarily play first
base, second base, third base, or shortstop; others include non-pitchers that primarily play catcher, left field, centre field, right field,
designated hitter, or non-pitchers with no well-defined defensive position—i.e., utility player—in addition to bobbleheads featuring more
than one player from at least two distinct defensive-position categories.). Results from splitting the player type into these three sub-types
are found to be statistically identical to that of the pooled estimate and are not presented in this study.
Date Description Availability Type Theme
04/06/2019 Todd Frazier 25,000 player
04/07/2019 Jacob deGrom 25,000 player
04/27/2019 Noah Syndergaard 25,000 player Game of Thrones
05/25/2019 Robinson Cano 25,000 player Star Wars
07/05/2019 Jerry Seinfeld 25,000 other
07/07/2019 Spider-Man 25,000 other Marvel
07/27/2019 Mr. Met 25,000 mascot
Type Star Wars Marvel Other None Total
Player 19 8 2 848 877
Manager 0 0 0 41 41
Mascot 3 0 0 20 23
Sports-caster 1 0 0 15 16
Other 2 8 15 48 73
Total 25 16 17 972 1,030
Theme
6
Note three instances are identified where a bobblehead promotion featured a player that was traded
away from the team earlier in the season.6 Due to this being a relatively rare phenomenon, demand response
to this such a peculiar situation is difficult to generalise, yet, if left untreated, these observations may introduce
bias in the results. These three games are therefore removed from the sample as is already reflected in Table 2
and elsewhere.
Lastly, note that many teams offer special-ticket packages that include a guarantee for a promotional
giveaway. Although publicly available data is lacking, the price of these special-ticket packages likely include
mark-ups above its counterfactual face value to partially or completely off-set the costs of the bobblehead and
licensing where applicable. These special-ticket packages remain outside the scope of this study as the effects
of these promotions could theoretically be directly observed by a well-informed empiricist (i.e., one that has
access to the purchases of standard and special-ticket packages and the pricing thereof).
DemandAnalysis
The following ordinary least squares regression is estimated:
ln𝐴,𝕀
,,𝛾
𝑎, 𝛾𝑎,
𝛾
𝑎,
𝛽𝑋𝜖 (1)
The dependent variable, ln𝐴,, is the paid attendance for team 𝑖 at time 𝑡 expressed in natural
logarithm, and 𝜖 is a stochastic error term. Because availability, represented by 𝑎,, is shown to have a non-
linear relationship with attendance (Cisyk & Courty, 2021), a cubic function is estimated. Lastly, 𝕀,, represents
a set of binary variables for each bobblehead characteristic 𝐶,, type or theme (estimated separately).
Demand-based co-variates and fixed effects, as represented by 𝑋, are used in the estimation of
Equation 1. These include fixed effects controlling for the demand cycles related to the first game of the season,
the time of day of the first pitch, the day of the week, and the month of the game.7 Information on the probability
of each game’s outcome is supplemented to the box score information to calculate a measure of within-season
team quality, termed the ‘predicted season wins’ (Cisyk, 2020). This represents an estimates number of wins
the team will accumulate across the season by summing a team’s realised wins prior to the game and the
expected wins from future games inferred from betting odds. Team-year fixed effects then capture inter-
temporal changes in team quality in addition to long-term economic conditions affecting the team’s local
economy.
Full results of Equation 1 are presented in Table A2 of the Appendix. In Columns 1 and 2, giveaway
availability is considered to have no impact on attendance (i.e., the condition 𝛾𝛾
𝛾
0 is imposed). The
results of Column 1 suggest that, although each type is found to have a positive and statistically significant
impact on attendance, there is much heterogeneity in estimates: sports-caster types have the largest impact on
attendance with an increase of roughly 12.5 percent, whereas manager types have just a 4.1 percent increase.8
6 The Philadelphia Phillies distributed a Hunter Pence bobblehead on 21 August 2012 despite having traded him on 31
July 2012; the Los Angeles Dodgers distributed a Juan Uribe bobblehead on 11 July 2015 despite having traded him on 27 May 2015; and
the Texas Rangers distributed a Star Wars-themed bobblehead of Jonathan Lucroy on 2 September 2017 despite having traded him on 30
July 2017. In each instance the team ran the promotion as scheduled.
7 Due to few observations, March games are treated as April and October games are treated as September.
8 These figures are calculated as exp𝛽1.
7
Player-type bobbleheads, the largest category, are estimated to increase attendance by 10.4 percent. While
results were estimated by splitting the player-type into the three mutually exclusive sub-types, results for each
individual player sub-types were not statistically different from the pooled estimate.
Note that the model predicts 78 percent of the variation in attendance and each control variable has
the expected sign and significance as is the case with each specification considered. A game against an opponent
of the same division is found to have a 1.3 percent increase in attendance while a game against an opponent of
the opposite league (e.g., an “interleague” match-up) is found to increase attendance by 7.9 percent, and a large
effect is found for a team’s first home game of the season. Each additional win (in expectation) is associated
with a 1.1 percent increase in attendance, meaning, for example, a 95-win team (indicating exceptional skill)
would have a 31.7 percent higher attendance than a 65-win team (indicating low skill), all else equal.
Column 2 repeats the same exercise as Column 1 considering only the bobblehead theme and finds
Star Wars-themed bobbleheads are associated with a 17.8 percent increase in attendance. Marvel-themed and
themeless bobbleheads have 11.2 and 9.7 percent increases to attendance respectively, and other themes are
found to have the lowest impact on attendance of 6.4 percent, yet this estimate is not significantly different
than zero.
Columns 3 and 4 of Table A2 add controls for bobblehead availability (i.e., the condition 𝛾𝛾
𝛾
0 is removed). Because the estimates of each type and theme cannot be easily interpreted in insolation from
availability, Table 3 illustrates the estimated impact of each of the different types and themes at 42 percent
availability (the revenue-maximising level of bobblehead availability).9 For example, a player-type bobblehead
at 42 percent availability would be expected to have a 13.4 percent increase in attendance, all else equal.
Table3:Estimatedimpactofbobbleheadtypeandthemeonattendance
9 This is the global maxima within the availability rang e of 0 𝑎
,100 as calculated using the following root formula
of a cubic function:
𝛾𝛾
3𝛾𝛾42%. For a longer discussion of availability of giveaways and demand maximisation, see
Cisyk and Courty (2021).
Type Count Impact Std. Err p-Value
= Sports-caster
= Player = Other = Mascot
Sports-caster 16 17.0% (0.041) 0.00
Player 877 13. 4% (0.009) 0.00 0.446
Other 73 9.9% (0.020) 0.00 0.158 0.101
Mascot 23 9.9% (0.029) 0.00 0.207 0.279 0.996
Manager 41 7.4% (0.025) 0.00 0.065 0.023 0.433 0.532
Theme Count Impact Std. Err p-Value = Star Wars = Marvel = O ther
Star Wars 25 20.4% (0.021) 0.00
Marvel 16 14.9% (0.045) 0.00 0.338
Other 17 8.6% (0.045) 0.07 0.036 0.373
None 972 12.7% (0.009) 0.00 0.002 0.665 0.415
42% Availability
42% Availability
p-Value of t-Test
p-Value of t-Test
8
Table 3 illustrates that the demand response is greatest for sports-caster types (an increase of 17.0
percent), followed by player (13.4), other (9.9), mascot (9.9), and manager (7.4). Each type has a statistically
significant impact on attendance, yet some point estimates are not found to be statistically different from
others, likely due to low observation count and, thereby, low power of the t-test. Despite this, the manager type
is still found to be statistically different than sports-caster and player types at the 90- and 95-percent
confidence level, respectively.
Regarding theme, the Star Wars theme is found to have a 20.4 percent increase in attendance,
statistically greater than themeless (an increase of 14.9 percent) or other-themed bobblehead games (8.6
percent). A large variance in the point estimate of the other theme leads to an impact that is not statistically
different from zero at the 95-percent confidence level.
Demand- and supply-side rationales compete for the explanation for these relatively small impacts of
manager-type and other-themed bobbleheads. While consumers may simply value these bobbleheads
differently, over-supply may result in satiation among consumers wherein the marginal benefit of a promotion
game diminishes with each additional one. In other words, it is possible that bobblehead games of all types and
themes experience diminishing marginal returns yet only some types and themes are featured as the
incremental bobblehead only after the team has exhausted all other options. A t-test reveals that team-seasons
with at least one manager-type bobblehead had on average 2.3 more bobblehead games than team-seasons
without, suggesting some evidence that satiation could explain the low impact of manager-type bobbleheads.10
Yet, team-seasons featuring at least one sports-caster bobblehead have 3.0 more bobblehead games than those
without and sports-caster bobbleheads are found to elicit a significantly higher demand response than
manager-types, which muddies this explanation.11 To formalise this, Columns 5 and 6 of Table A2 add two
controls following the methodology of Paul & Weinbach (2013): a count variable representing the cumulative
total number of bobblehead games that have already occurred within the team-season is added to Equation 1
as well as the squared value of this measure.12 The results suggest there is some evidence of satiation and that
the effect bobbleheads have on attendance declines (although at a decreasing rate) with each additional game
bobblehead game. The estimated effect implies diminishing effectiveness of bobblehead games from the first
to seventh occurrence, generally consistent with the findings of McDonald & Rascher (2000) and Cisyk & Courty
(2021).13
Cannibalisation is tested in several ways (although, for brevity, only one version is shown in Table A3).
First, games of the same team immediately adjacent to a bobblehead game are assigned a dummy
10 The number of bobblehead games for the 40 team-seasons with at least one manager-type bobblehead game is 6.2
versus 3.9 for the 200 team-seasons without; the difference is statistically different at the 99-percent level.
11 The number of bobblehead games for the 15 team-seasons with at least one sports-caster type bobblehead game is
7.1 versus 4.1 for the 225 team-seasons without; the difference is statistically different at the 99-percent level.
12 This count variable takes the value of one for a team’s first bobblehead game in the season, two for the second, and
so on, and so forth; it takes the value of zero for all non-bobblehead games.
13 The function of bobblehead game count and the count squared has a global minimum at seven (considering integer
values). While the estimated coefficients imply increasing effectiveness of bobblehead games after the seventh game, note that 91 percent
of team-seasons have seven or fewer bobblehead games (218 of 240 team-seasons). Note that Paul & Weinbach (2013), which prescribes
this methodology, does not find any satiation effect with promotions yet focuses solely on the count of post-game fireworks whereas
McDonald & Rascher (2000) and Cisyk & Courty (2021) consider only giveaway promotions.
9
corresponding to the type or theme of the bobblehead available in the preceding or succeeding game. In other
words, two sets of binary variables, 𝕀,, and 𝕀,,, are added to Equation 1. Negative coefficients on any of
these binary variables would suggest evidence of consumers substituting to a bobblehead game from games
immediately preceding or succeeding thereby overstating the attendance increase observed on bobblehead
games. However, none of these binary variables are negative and significant while a few are, in fact, positive
and significant. The same is true when considering other lagging or leading times, such as 𝑡2 or 𝑡7 (the
latter representing a given day of the week that may align best with consumers’ schedules). The weak support
for any cannibalisation is consistent with the findings of other studies (e.g., Kappe et al., 2014; Cisyk and Courty,
2021) and the positive increases are said to be potential spill-over effects associated with promoting the
giveaway day. Collectively, there is no evidence of cannibalisation from bobblehead games, regardless of type
or theme.
A parting concern, having identified a heterogeneous demand based on bobblehead characteristics,
considers whether consumers also value former and current bobblehead types differently. For example, an
exceptional former player may have a greater impact on attendance than a current player, or vice versa if the
average consumer lacks the familiarity with the team’s history to recall the prominence of the former player.
To test for this, a binary variable for whether the bobblehead game features a former player, former manager,
or former sports-caster is added to Columns 7 and 8 of Table A2, but this variable is ultimately found to be
insignificant implying that the distinction between past and present individuals does not matter to the average
consumer.
CorrelationofBobbleheadSupplyandOtherDemandFactors
In order to substantiate the demand-side analysis, it is important to illustrate the supply is not
systemically correlated with other factors that determine demand. The availability of each type and theme of
bobblehead is tested. Cisyk and Courty (2021) explains availability is not correlated with the games’ day of the
week, month, year, or opponent, yet the possibility that type or theme may impact the supply of bobbleheads
remains unexplored. Table 4 displays the regression results of bobblehead availability on these two measures.
Columns 1 and 3 do not employ the use of fixed effects and suggest bobblehead type and, separately, theme
explain very little variation alone in the availability across all teams and seasons (the adjusted coefficient of
correlation is just 0.001 in both columns). Adding in team-season fixed effects, as is done in Columns 2 and 4,
helps to explain nearly 95 percent of the variation. Focusing on type, only “other” is found to have statistically
lower availability than the remaining types (at the 95-percent confidence level). This result is puzzling, yet it is
worth noting that teams offer the same availability for each of the types that are directly associated with its
own brand (i.e., the depicted player, manager, or sports-caster are all current or former employees of the team
and often are portrayed donning team-branded clothing). For themed bobbleheads in Column 4, the Marvel-
and other-themed bobbleheads have statistically lower availability, indicating teams may be less willing to
incur the additional licensing costs of these themes yet do not appear to take issue with incurring the additional
costs associated with the Star Wars branding. Licensing agreements are typically private and the possibility
10
remains that the fees associated with Star Wars are lower than that of the remaining themes observed in the
sample period. Despite these findings, because type and theme unilaterally explanation very little of the
variation in availability, endogeneity is likely of little concern.
Table4:Bobbleheadavailabilityregressionresults
DiscussionandConcludingRemarks
Promotions are an important tool to elicit demand for live sporting events, especially in instances
where gate and concession revenues represent a sizeable portion of income. Specifically, bobbleheads have
been repeatedly shown to have the greatest promotional impact on additional ticket sales (Boyd & Krehbiel,
2006; Barilla et al., 2008; Paul et al., 2019). Given both the budgetary constraints and a satiation effect of
decreasing marginal impact of promotional games, decision-makers within a sports franchise are limited by
the volume of games that can offer bobbleheads and other promotions to ticket-holders. Consider that, from
2012 to 2019, of the 81 scheduled home games for each team, the modal number of bobblehead games just
three per season: this places a heightened importance on correctly identifying the optimal strategy for
bobblehead games. While the timing of promotions (e.g., the day of the week or the opponent) and the
availability (i.e., the fraction of the stadium’s capacity that can receive the promotion) remain important aspects
for marketing decision-makers to consider when scheduling bobblehead promotions, this study illustrates that
the distinct characteristics of the bobblehead’s design affect the impact these promotions have on attendance
at live sporting events. Future studies should consider these additional details, as ignoring this may bias
estimates of the effectiveness of promotions. Decision-makers within sports organisations should also yield
additional consideration, beyond timing and availability, to the consumer’s familiarity with the bobblehead’s
Pla
y
er
(
base
)
-0.052 0.002
(0.032) (0.004)
0.053 -0.002*
(0.064) (0.001)
0.091 -0.004
(0.066) (0.007)
-0.008 -0.027**
(0.039) (0.013)
0.528*** 0.530***
(0.019) (0.001)
Home Team × Year FE No Yes
Observations 1,030 1,030
Ad
j
usted R20.001 0.945
Dependent Variable:
Bobblehead Availabilit
y
(1) (2)
Constant
Manager
Mascot
Sports-caster
Other Type
Themeless
(
base
)
-0.040 0.018
(0.041) (0.029)
-0.120** -0.061* **
(0.053) (0.023)
-0.018 -0.057*
(0.053) (0.031)
0.531*** 0.529***
(0.020) (0.002)
Home Team × Year FE No Yes
Observations 1,030 1,030
Ad
j
usted R20.001 0.947
Constant
Dependent Variable:
Bobblehead Availabilit
y
(3) (4)
Star Wars
Marvel
Other Theme
Notes: * -
p
< 0.1; ** -
p
< 0.05; *** -
p
< 0.01. Standard errors clustered b
y
team-
y
ear in
p
arentheses .
11
namesake (e.g., player versus manager) and, where applicable, the featured brand (e.g., Star Wars versus Hello
Kitty). Manager-type bobbleheads are found to have statistically lower impacts on attendance than player
types: decision-makers may be best-served to exhaust all player options, current or former, before scheduling
a manager bobblehead (although this finding may be partially explained by satiation). Star Wars-themed
bobbleheads are associated with significantly higher attendance yet estimates of the attendance impact of other
themes are not statistically different from that of themeless bobbleheads: risk-averse decision-makers may
want to avoid the additional costs of licensing fees by opting for themeless bobbleheads. Future research may
also want to consider delineating player-types by additional qualities, such as player prominence, although low
observation count may limit the ability to generalise results.
While the stakes for optimising bobblehead timing, availability, and design remain high, cost-saving
measures are available to the promotion decision-makers to alleviate the additional burden of budget
constraints. A promotion sponsorship, wherein a corporate sponsor pays to have its logo featured alongside
the promotion, is a common and effective tool to either partially or completely off-set the cost of the promotion.
This can be used to set availability to revenue-maximising levels without impacting the effectiveness of the
promotion (Burnett & van Scyoc, 2008). Elsewhere, special-event tickets (i.e., a ticket guaranteeing its holder
the giveaway item) deserves future attention as the availability of public data has limited empiricists from
assessing their impacts. However, the methods and results discussed within this study are still applicable to
these promotions that rely on special-event tickets. For example, due to lead times in manufacturing, organisers
must estimate the demand for a given type of bobblehead on a given date slated for a special-event promotion
in order to purchase the appropriate quantities and minimise waste (unlike the typical stadium giveaway
where items are available while supplies last, unsold special-tickets leads to excess special-event bobbleheads).
The findings of this study may also not be limited to bobblehead promotions. For example, while it
warrants further research, it may not be unreasonable to assume the same heterogenous demand response
based on a bobblehead’s name-sake would hold for other personalised promotions, such as t-shirts or replica
jerseys that have the option to bear a surname and number on the back-side (baseball has the added
idiosyncrasy as being the only major sport wherein the on-field coaches and manager wears the same team
uniform as the players and are each assigned unique jersey numbers).
Budget constraints and consumer satiation limit the breadth (i.e., count) and depth (i.e., availability)
of promotions a team can offer each season. Given the magnitude of the demand response to bobblehead games
and the relative scarcity, teams must give each aspect of the giveaway much consideration. This paper adds to
the literature additional findings to help maximise the return on investment of bobblehead promotions.
12
References
Anthony, T., Kahn, T., Madison, B., Paul, R. J., & Weinbach, A. (2014). Similarities in fan preferences for minor-league baseball across the
American Southeast. JournalofEconomicsandFinance, 38(1), 150-163.
Barilla, A. G., Gruben, K., & Levernier, W. (2008). The effect of promotions on attendance at Major League Baseball games. JournalofApplied
BusinessResearch(JABR), 24(3).
Borland, J., & MacDonald, R. (2003). Demand for sport. Oxfordreviewofeconomicpolicy, 19(4), 478-502.
Boyd, T. C., & Krehbiel, T. C. (1999). The effect of promotion timing on Major League Baseball attendance. SportMarketingQuarterly, 8(4),
23-34.
Boyd, T. C., & Krehbiel, T. C. (2003). Promotion Timing in Major League Baseball and the Stacking Effects of Factors That Increase Game
Attractiveness. SportMarketingQuarterly, 12(3).
Boyd, T. C., & Krehbiel, T. C. (2006). An analysis of the effects of specific promotion types on attendance at Major League Baseball
games. AmericanJournalofBusiness.
Bruggink, T. H., & Eaton, J. W. (1996). Rebuilding attendance in Major League Baseball: The demand for individual games. Baseball
economics:Currentresearch, 9-31.
Burnett, N. J., & Van Scyoc, L. (2008). Stadium attendance for baseball: A case study. JournalofEconomicsandEconomicEducation
Research, 9(3), 29.
Cairns, J., Jennett, N., & Sloane, P. J. (1986). The economics of professional team sports: A survey of theory and evidence. Journal of Economic
Studies, 13(1), 3-80.
Cebula, R. (2009). The potential role of marketing in promoting free enterprise in the US: A study involving minor league baseball and
ticket-sales revenue maximization. JournalofInternationalandGlobalEconomicStudies, 2(1), 31-45.
Cebula, R. J. (2013). A panel data analysis of the impacts of regional economic factors, marketing and promotions, and team performance
on minor league baseball attendance. TheAnnalsofRegionalScience, 51(3), 695-710.
Cebula, R. J., Toma, M., & Carmichael, J. (2009). Attendance and promotions in minor league baseball: The Carolina League. Applied
Economics, 41(25), 3209-3214.
Cebula, R. J., Coombs, C. K., Lawson, L., & Foley, M. (2013). The impacts of promotions/marketing, scheduling, and economic factors on total
gross revenues for minor league baseball teams. InternationalAdvancesinEconomicResearch, 19(3), 249-257.
Chupp, A., Stephenson, E. F., & Taylor, R. (2007). Stadium alcohol availability and baseball attendance: Evidence from a natural experiment.
InternationalJournalofSportFinance, 2(2), 36-44.
Cisyk, J. (2020). Impacts of performance-enhancing drug suspensions on the demand for Major League Baseball. JournalofSports
Economics, 21(4), 391-419.
Cisyk, J., & Courty, P. (2017). Do fans care about compliance to doping regulations in sports? The impact of PED suspension in
baseball. JournalofSportsEconomics, 18(4), 323-350.
Cisyk, J., & Courty, P. (2021). Stadium Giveaway Promotions: How Many Items to Give and the Impact on Ticket Sales in Live Sports. Journal
ofSportManagement, 35(6), 551-565.
DeSchriver, T. D. (2007). Much Adieu About Freddy: Freddy Adu and Attendance in Major League Soccer. JournalofSport
Management, 21(3).
Garcia, J., & Rodríguez, P. R. (2009). Sports attendance: A survey of the literature 1973-2007. Rivista di Diritto e di Economia dello Sport,
5(2), 112-151.
Hill, J. R., Madura, J., & Zuber, R. A. (1982). The short run demand for major league baseball. AtlanticEconomicJournal, 10(2), 31-35.
Howell, S. M., Klenosky, D. B., & McEvoy, C. D. (2015). Weather, timing, and promotions in Minor League Baseball: An examination of
attendance in the International League. JournalofAppliedSportManagement, 7(2), 9.
Kappe, E., Stadler Blank, A., & Desarbo, W. S. (2014). A general multiple distributed lag framework for estimating the dynamic effects of
promotions. ManagementScience, 60(6), 1489-1510.
Kelley, K., Harrolle, M. G., & Casper, J. M. (2014). Estimating consumer spending on tickets, merchandise, and food and beverage: A case
study of a NHL team. JournalofSportManagement, 28(3), 253-265.
Lemke, R. J., Leonard, M., & Tlhokwane, K. (2010). Estimating attendance at Major League Baseball games for the 2007 season. Journalof
SportsEconomics, 11(3), 316-348.
Marcum, J. P., & Greenstein, T. N. (1985). Factors affecting attendance of Major League Baseball: II. A within-season analysis. Sociologyof
SportJournal, 2(4), 314-322.
McDonald, M., & Rascher, D. A. (2000). Does bat day make cents?: The effect of promotions on the demand for baseball. JournalofSport
Management, 14.
Paul, R., & Chatt, R. (2011). Regional differences in fan preferences for minor league hockey: The AHL. NewYorkEconomicReview, 63.
Paul, R. J., Garrett, C., Barbuto, C., & Liotta, K. (2019). Celebrity attraction in the minors: The case of Tim Tebow. JournalofEconomicsand
Finance, 43(4), 795-812.
Paul, R. J., Paul, K. K., Toma, M., & Brennan, A. (2007). Attendance in the NY-Penn Baseball League: Effects of performance, demographics,
and promotions. NewYorkEconomicReview, 38(1), 72-81.
Paul, R. J., & Weinbach, A. P. (2011). Minor League Baseball Attendance in the Pacific Northwest: A Study of the Effects of Winning, Scoring,
Demographics and Promotions in the Northwest and Pioneer Baseball Leagues. EkonomikaaManagement,Universityof
Economics,Prague, 2011(2).
Paul, R. J., & Weinbach, A. P. (2013). Fireworks saturation and attendance in minor league baseball. InternationalJournalofSport
Finance, 8(4), 312.
Rascher, D. A., & Schwarz, A. D. (2012). Illustrations of price discrimination in baseball. TheOxfordHandbookofSportsEconomics:
EconomicsThroughSportsVolume2, 380-399.
Schreyer, D., & Ansari, P. (2022). Stadium attendance demand research: A scoping review. Journal of Sports Economics, 23(6), 749-788.
Schreyer, D., Schmidt, S. L., & Torgler, B. (2019). Football spectator no-show behavior. Journal of Sports Economics, 20(4), 580-602.
Siegfried, J. J., & Eisenberg, J. D. (1980). The demand for minor league baseball. AtlanticEconomicJournal, 8(2), 59-69.
13
Appendix
TableA1:DescriptiveStatistics
Variable Definition Mean St. Dev. Min Max
0.13 1.00
1
10,123 56,3105,265
00.22
29,957
0.05
0.53 0.25
Attendance Pai d attendanc e at event
Availability* Number of bobbleheads gi ven out as share of the stadi um’s
seating capacity.
Bobblehead
Game Indicator
Takes a value of one for observations of bobblehead games,
zero otherwi se.
Notes: * - descriptive statistics are conditional on observation being a bobblehead game. Sample includes all MLB games played during
the 2012 to 2019 regular-seasons excluding (a) games cancelled due to inclement weather and never re-scheduled; (b) games hosted in
neutral sites; (c) games cancelled due to extenuating circumstances (protests and civil unrest or extreme weather and hurricanes); and
bobblehead games featuri ng players that were traded from team i n the same seaso n prior to the bobblehead game. Num ber of
observations = 19, 402.
Other Type* Takes a value of one for observations of other-type
bobblehead games, zero otherwise. 0.07 0.26 0 1
Star Wars* Takes a value of one for observations of Star Wars-themed
bobblehead games, zero otherwise.
01
Other Theme* Takes a value of one for observations of other-themed
bobblehead type games, zero otherwise. 0.02
Marvel* Takes a value of one for observations of Marvel-themed
bobblehead games, zero otherwise.
Play er* Takes a value of one for observations of player-type
bobblehead games, zero otherwise.
Sports-Caster* Takes a value of one for observations of sports-caster type
bobblehead games, zero otherwise. 0.02
Mascot* Takes a value of one for observations of mascot-type
bobblehead games, zero otherwise. 0.02
Manager* Takes a value of one for observations of manager-type
bobblehead games, zero otherwise. 0.04 0
0
1
0.85 0.36 0 1
10.15
0.20
1
0.02 0.15 0 1
0.12 0
0.02 0.12
1
Themeles s
Bobblehead*
Takes a value of one for observations of themeless
bobblehead type games, zero otherwise. 0.94 0.23 0 1
0.13 0
112.61
Division Takes a value of one for games featuring two teams of the
same division, zero otherwise. 0.46 0.50 0 1
Predic ted Season
Wins Expected number of wins a team will earn by season’s end 81.01 10.59 46.16
1
Home Opener T akes a value of one for the eac h team’s firs t home game of
each season, zero otherwise. 0.01 0.11 0 1
Interleague Takes a value of one for games featuring two teams of
opposite leagues (American and National), zero otherwise. 0.12 0. 33 0
14
TableA2:Mainregressionresults
0.099*** -0.283*** -0.222** -0.281***
(0.008) (0.095) (0.101) (0.094)
0.040* -0.338** * -0.277*** -0.335***
(0.023) (0.093) (0.099) (0.092)
0.073*** -0.315*** -0.257** -0.313***
(0.028) (0.099) (0.105) (0.098)
0.118** * -0.252** -0.187* -0.250**
(0.039) (0.105) (0.110) (0.103)
0.069*** -0.315*** -0.249** -0.313***
(0.020) (0.097) (0.105) (0.097)
0.164** * -0.218** -0.153 -0.218**
(0.021) (0.099) (0.105) (0.099)
0.107** -0.264*** -0.199* -0.264***
(0.042) (0.101) (0.107) (0.101)
0.062 -0.321* ** -0.249** -0.321* **
(0.042) (0.116) (0.120) (0.115)
0.093** * -0.284*** -0.218** -0.284** *
(0.008) (0.096) (0.102) (0.095)
-0.023*** -0.023** *
(0.006) (0.006)
0.002** * 0.002***
(0.000) (0.000)
-0.003 0.000
(0.015) (0.015)
2.352** * 2.313*** 2.262*** 2. 190** * 2.344*** 2.314* **
(0.615) (0.627) (0.650) (0.659) (0.610) (0.622)
-4.246*** -4.168*** -4.063*** -3.924*** -4.230*** -4.170***
(1.182) (1.206) (1.251) (1.270) (1.170) (1.194)
2.294** * 2.251*** 2.188*** 2. 112** * 2.285*** 2.252* **
(0.663) (0.677) (0.702) (0.713) (0.656) (0.670)
0.011** * 0.011*** 0.011* ** 0. 011** * 0.011*** 0.011* ** 0.011*** 0.011** *
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001)
0.013** * 0.013*** 0.013* ** 0. 013** * 0.013*** 0.013* ** 0.013*** 0.013** *
(0.004) (0.004) (0.004) (0.004) (0.004) (0.004) (0.004) (0.004)
0.076** * 0.076*** 0.076* ** 0. 076** * 0.076*** 0.076* ** 0.076*** 0.076** *
(0.006) (0.006) (0.006) (0.006) (0.006) (0.006) (0.006) (0.006)
0.558** * 0.557*** 0.557* ** 0. 557** * 0.558*** 0.558* ** 0.557*** 0.557** *
(0.021) (0.021) (0.021) (0.021) (0.021) (0.021) (0.021) (0.021)
Da
y
of Week × Ni
g
ht FE Yes Yes Yes Yes Yes Yes Yes Yes
Month FE Yes Yes Yes Yes Yes Yes Yes Yes
Home Team × Year FE Yes Yes Yes Yes Yes Yes Yes Yes
O
pp
onent FE Yes Yes Yes Yes Yes Yes Yes Yes
Observations 19,402 19, 402 19,402 19,402 19,402 19,402 19,402 19,402
Ad
j
usted R
2
0.777 0.777 0.777 0.777 0.778 0.777 0.777 0.777
Cumulative count
of bobblehead
g
ames
Cumulative count
of bobblehead
g
ames, squared
(7)
Dependent V ariable:
Natural Lo
g
arithm of Attendance
(1) (4)(2) (3) (5) (6)
Play er
Star Wars
Marvel
Division
(8)
Former Type
Notes: * -
p
< 0.1; ** -
p
< 0.05; *** -
p
< 0.01. Standard errors cl ustered b
y
team and
y
ear in
p
arentheses .
Home Opener
Interleague
Manager
Mascot
Sports -caster
Other Type
Other
Themeles s
Availability
Avialability
2
Avialability
3
Predicted Season Wins
15
TableA3:Cannibalisation
-0.280*** -0.214**
(0.094) (0.098)
-0.335*** -0.260**
(0.092) (0.101)
-0.310*** -0.316***
(0.098) (0.115)
-0.249** -0. 279* **
(0.104) (0.095)
-0.311***
(0.096)
0.019*** -0.032
(0.007) (0.041)
0.011 -0.020
(0.029) (0.038)
0.054** 0.007
(0.027) (0.037)
-0.027 0.022** *
(0.028) (0.007)
0.039
(0.026)
0.013* 0.001
(0.008) (0.033)
-0.013 -0.025
(0.025) (0.051)
0.046 0.025
(0.044) (0.044)
0.003 0.013*
(0.029) (0.007)
-0.003
(0.021)
Availabilit
y
Controls Yes Availabilit
y
Control s Yes
Predicted Season Wi ns Control Yes Predicted Season Wins Control Yes
Division Indicato
r
Yes Division Indicato
r
Yes
Interlea
g
ue Indicato
r
Yes Interlea
g
ue Indicator Yes
Home O
p
ener Indicato
r
Yes Home O
p
ener Indicato
r
Yes
Da
y
of Week × Ni
g
ht FE Yes Da
y
of Week × Ni
g
ht FE Yes
Month FE Yes Month FE Yes
Home Team × Year FE Yes Home Team × Year FE Yes
O
pp
onent FE Yes O
pp
onent FE Yes
Observations 19,402 Observations 19,402
Ad
j
usted R
2
0.778 Ad
j
usted R
2
0.778
Other
Dependent Variable:
Natural Lo
g
arithm of Attendance
(1) (2)
Player
Manager
Dependent Variable:
Natural Lo
g
arithm of Attendance
Star Wars
Marvel
Mascot
Sports-caster
Other Type
Manager in game t-1
Mascot in game t-1
Notes: * -
p
< 0.1; ** -
p
< 0.05; *** -
p
< 0.01. Standard errors clustered b
y
team and
y
ear in
p
arentheses .
Player in game t+1
Manager in game t+1
Mascot in game t+1
Sports-caster i n game t+1
Other Type in game t+1
Player in game t-1
Sports-caster i n game t-1
Other Type in game t-1
Themeless
Star Wars in game t-1
Marvel in game t-1
Other in game t-1
Themeless in game t-1
Star Wars in game t+1
Marvel in game t+1
Other in game t+1
Themeless in game t+1
