Sunteți pe pagina 1din 26

John Carroll University

An Analysis of the Cleveland Indians Game-Day


Home Attendance: 2013-2014

Eric Smith
EC 499
Individual Research Project in Economics
Dr. Thomas Zlatoper

Table of Contents
I. Introduction.......................................................................................................... 1
II. Literature Review............................................................................................... 2
III. Model................................................................................................................... 6
IV. Data..................................................................................................................... 9
V. Estimation Results............................................................................................ 11
A. The Estimation Model........................................................................................ 11
B. Overall Model.................................................................................................... 11
C. Estimation Coefficients...................................................................................... 12
VI. Conclusions...................................................................................................... 15
VII. Appendix A...................................................................................................... 17
Table A1. Seasonal Attendance for the Cleveland Indians.....................................17
VIII. Appendix B..................................................................................................... 17
Table B1. 2013 Season Data Descriptive Statistics ( n = 80 )................................17
Table B2. 2014 Season Data Descriptive Statistics ( n = 78 )................................18
Table B3. 2013 & 2014 Data Combined Descriptive Statistics ( n =158 )............18
Table B4. 2013 & 2014 Model Summary ( n =158 ).............................................19
Figure B1. 2013 & 2014 Residual Plot ( n =158 ).................................................20
IX. Appendix C....................................................................................................... 21
Works Cited................................................................................................................ 22

I. Introduction
From June 12th, 1995 to April 4th, 2001, the Cleveland Indians sold out 455 consecutive
regular season home game (Castrovince). At the time this was a record not just for Major League
Baseball, but for all American sports. During this span the Indians organization had average
attendance of 41,582 fans per game and they won 58.7% of their games. Not only were the
Indians winning baseball games, they also debuted their new stadium, Jacobs Field, in 1994.
Following a brief period of three seasons with a sub-.500 record, the Cleveland Indians bounced
back and won 57.4% of their games in 2005 and 59.3% of their games in 2007. In fact, over the
span of 10 years (2005-2014) the Indians have won about 50% of their games, yet their
attendance over this same time period has been a dismal 22,413 fans per game. Speculation
about the poor attendance figures by the media and fans alike has spanned a variety of reasons,
including: a poor local economy, failure by the owners to spend money on the roster, and out-ofcontrol ticket prices. Another potential reason for the decline in attendance, or rather a reason for
the increased attendance from 1994-2001, is the honeymoon effect created by the construction of
the new baseball stadium, Jacobs Field.
The purpose of this empirical research paper is to examine the factors that have
influenced the Cleveland Indians home game-day attendance during the 2013 and 2014 regular
seasons. During these two seasons the Cleveland Indians had a winning record yet their
attendance was well below the league average. This paper is broken down into the following
sections: Introduction, Literature Review, Model, Data, Results, and Conclusion. The Literature
Review discusses some other relevant research and how it ties into this research paper. The
Model section presents the general form of the model and discusses all of the variables used. The
Data section contains a discussion of the data and how they were collected. The Results section
Page | 1

presents the estimation results from the model and all of the appropriate statistics. Finally, in the
Conclusion section this paper summarizes the important results and makes suggestions going
forward.

II. Literature Review


There exists an enormous amount of literature on the subject of sporting event
attendance. One of the challenges of reviewing this wealth of literature is finding the research
that is the most relevant to analyzing the Cleveland Indians attendance.
For a starting point, I have decided to begin with the article by Welki & Zlatoper and
structure my model using their article as a foundation (Welki and Zlatoper). Their research looks
into fan attendance of National Football League (NFL) games, but much of their work is still
relevant to Major League Baseball (MLB) fan attendance. Welki & Zlatopers research
concludes that the quality of the teams involved as well as weather conditions are important
predictors of fan attendance. A key difference though is that NFL games are typically played on
Sundays, whereas MLB games are played on every day of the week. Another important
difference is that there are only eight home games in an NFL season compared to 81 home games
for MLB teams. Despite these differences, the findings of Welki & Zlatoper stand as a key
starting point since their model provides a sound approach to modeling attendance. Because my
model examines game-by-game attendance instead of looking at each season as a whole, the
quality of the teams in each game is an important variable to measure in my model. The Welki &
Zlatoper article breaks down quality by including variables such as conference and division
matchups as well as the home teams winning percentage prior to the game. Another important
variable included in the Welki & Zlatoper article is the interaction term of temperature and

Page | 2

precipitation. Welki & Zlatopers article showed statistical significance in these variables, and
they will be strongly considered for inclusion in this model.
Gitter & Rhoads provide an interesting wrinkle into the literature by establishing minor
league baseball games as substitutes for major league games (Gitter and Rhoads). Their model
focuses on minor league home teams and looks at ticket pricing in both leagues, distances of
minor league and major league teams, and performance of the home team. Their paper finds that
more winning has a significant effect and leads to higher game-day attendance and that distance
of minor league teams acts as a statistically significant substitute and thus has a negative impact
on MLB attendance. Gitter & Rhoads findings show the importance of examining the substitutes
for baseball games in the Cleveland area. The Cleveland Indians have three minor league
affiliates within 150 miles. Using Gitter & Rhoads research, I can rule out the Columbus, Ohio
minor league affiliate of the Cleveland Indians because it is about 150 miles away and thus not as
likely to have an effect, but the other two Ohio affiliates, Akron and Lake County, are close
enough to include them as a likely substitute. Gitter & Rhoads research deals specifically with
minor leagues teams as substitutes, but one of the challenges with this papers model is choosing
which other potential substitutes to take into account.
Cebula analyzes minor league baseball game-day attendance by looking at many
variables, including: economic indicators*, marketing promotions*, day of the week*, month of
the year, and team performance* (Cebula).1 The approach by Cebula, though quite thorough,
raises some questions of collinearity as well as questionable reasoning for variable selection.
With so many similar variables (months, days, promotions such as fireworks, etc.) it would have
been helpful to see a section addressing potential collinearity. For example, firework promotions,
1 * Cebulas study found these variables significant at the 5% level.
Page | 3

summer months, and weekend games may overlap and have issues of collinearity, yet Cebula
does not address this in his article. Another potential problem with Cebulas article is that he
adds variables to his initial results even though his reasoning is questionable. For example, he
adds fielding errors errors made by players on defense even though he does not use
seemingly more obvious variable such as homeruns. In fact the only performance variables used
by Cebula are winning percentage, total runs scored, and fielding errors. Despite these issues,
Cebulas article is highly important to this papers research. His use of many variables along with
his reasoning for using them is helpful in identifying potential variables for my own research.
Also, the issues I have raised with his work provide useful warnings for my own research going
forward.
Another important aspect when analyzing the factors of Cleveland Indians fan attendance from
1994-2014 is the honeymoon effect. Zygmont & Leadley provide an excellent article on the
honeymoon effect and game-day baseball attendance (Zygmont and Leadley). The honeymoon
effect is given as the relationship between attendance and the age of the sports facility. Zygmont
& Leadleys literature review details the previous research involving the honeymoon effect and
explains how they have used this to improve their model. The conclusion of their research is that
a new baseball stadium improves attendance by approximately 6,000 fans per game in the first
year, and the improvement then declines to 1,100 additional fans by the tenth year and 500 fans
per game in the fifteenth year. Overall, Zygmont & Leadley report that there is a significant
honeymoon effect on new baseball stadiums that lasts approximately 8 years. Zygmont &
Leadley also find that teams raise the price of tickets by an average of 30% in the first year. This
honeymoon effect poses a serious issue with my research because the Cleveland Indians debuted
a new stadium in 1994 and experienced a large drop in attendance after the eighth season in the
Page | 4

new stadium. This paper focuses on the 2013 and 2014 seasons, which means the honeymoon
effect should have expired and perhaps explains some of the decreased attendance since 1994.
Butler addresses another nuance of Major League Baseball: interleague play (Butler).
There are two leagues within Major League Baseball, the American League and the National
League. Prior to 1997 teams from these two leagues could only meet in the championship, better
known as the World Series. Beginning in 1997 MLB introduced interleague play to improve
attendance. Butlers research estimates the impact of interleague play and finds that game-day
attendance increased by approximately 7% for interleague games. Because these games only
make up a small portion of the overall season, the overall increase was only about half of 1% for
the season. Another interesting contribution from Butler is his model. To isolate certain effects,
Butler used numerous dummy variables2. An example where Butler simplified his model using a
dummy variable not used in the other articles is temperature. He set the dummy variable to 1 if
game temperature was less than 55 degrees Fahrenheit and another variable if temperature was
higher than 94 degrees Fahrenheit. While this approach is not necessarily better, it is interesting
to consider a variety of uses for dummy variables.

2 Butlers dummy variables that were significant at the 5% level include: Games behind division leader,
opponents win percentage, opponents payroll, day of the week, month of the year, and weather.
Page | 5

III. Model
The general form of the model for this analysis is as follows:
AATTENDANCE = f (SWEEKEND, RDIV, TNIGHT, WSTREAK, LSTREAK, FTEMP , POTHER, PDOLLAR, PFIREWORKS,
IINTER, VRIVAL, YRIVAL).
The dependent variable is briefly defined as:
AATTENDANCE = the number of Cleveland Indians home tickets sold
The independent variables are briefly defined as follows:
SWEEKEND = A dummy variable defined as 1 if the game is on a Friday, Saturday, or
Sunday and 0 otherwise. ( + ) 3
RDIV = the rank of the team with respect to the five teams in the American League Central
Division. ( - )
TNIGHT = A dummy variable defined as 0 for a daytime game and 1 for a nighttime game. (
+)
WSTREAK = the number of consecutive games won by the Indians at the start of that home
game. ( + )
LSTREAK = the number of consecutive games lost by the Indians at the start of that home game. ( )
FTEMP = the temperature in degrees Fahrenheit at the start of the game. ( + )
POTHER = A dummy variable defined as 1 if the game had a promotional event besides
fireworks or Dollar Dog Night and 0 otherwise. ( + )
3 The expected impacts of each variable are indicated next to each definition. A positive impact is indicated by a
positive sign ( + ) and a negative impact is indicated by a negative sign ( - ).

Page | 6

PDOLLAR = A dummy variable defined as 1 if the game had an event called Dollar Dog
Night as a promotion and 0 otherwise. ( + )
PFIREWORKS = A dummy variable defined as 1 if the game had a fireworks event as a
promotion and 0 otherwise. ( + )
IINTER = A dummy variable defined as 1 if the game was an interleague matchup and 0
otherwise. ( + )
VRIVAL = A dummy variable defined as 1 if the game is a matchup against a division rival
and 0 otherwise. ( + )
YRIVAL = A dummy variable defined as 1 if the game is against the New York Yankees or
the Boston Red Sox.4 ( + )
SWEEKEND and TNIGHT are expected to have a positive impact on game-day attendance. It
may be easier for fans to attend a night or weekend game since they may have to work during the
day/week or have other errands to run.
RDIV is expected to have a negative impact on game-day attendance. An increase in this
variable means the Indians are falling behind with respect to other teams in their division. The
assumption is that a team in first place is winning and thus offers a higher quality of
entertainment for the ticket purchaser. This paper also assumes that a higher RDIV means the team
offers a lower quality of entertainment value for the ticket buyer.
WSTREAK is expected to have a positive impact on game-day attendance under the same
assumption that winning means a higher quality of entertainment. The dependent variable is
defined as tickets sold, which means fans may purchase their tickets ahead of time and thus prior
to any winning streak.

4 This paper was limited to using one variable for the two teams expected to have the largest impact.
Ideally this paper would utilize a separate variable for each team that the Cleveland Indians played
against during these seasons.
Page | 7

LSTREAK is similar to WSTREAK except it has the opposite expected impact. A higher value for
LSTREAK should result in a lower value for attendance since losing is the inverse of winning.
FTEMP is expected to have a positive impact on game-day attendance. The assumption is
that fans may be less likely to attend a colder game since the games are held outdoors.
POTHER, PDOLLAR, PFIREWORKS are expected to have a positive impact on game-day
attendance. The purpose of promotions is to drive ticket sales, so they should have a positive
impact on game-day attendance.
IINTER is expected to have a positive impact on game-day attendance. Butlers study found
that interleague games increase attendance by 7% on average.
VRIVAL is expected to have a positive impact on game-day attendance. In MLB the
divisional games have a larger impact on making the playoffs since the division leaders at the
end of the regular season get an automatic bid to the playoffs.
YRIVAL is expected to have a positive impact on game-day attendance. The Yankees and
Red Sox have large fan bases which should result in more buyers of tickets to games involving
these teams.

Page | 8

IV. Data
The empirical data used in this study were collected from the 2013 and 2014 MLB
seasons. All of the games were played in Cleveland, Ohio at Progressive Field. The data were
compiled from baseball-reference.com and mlb.com. The model uses data from the 2013 and
2014 seasons because the Cleveland Indians finished both seasons with a winning record but still
had below average attendance.5
AATTENDANCE refers to the number of tickets sold for a Cleveland Indians home game at
Progressive Field in Cleveland, Ohio. It does not account for actual game-day attendance and
thus it is possibly an overestimate of actual attendance.
SWEEKEND is a dummy variable indicating whether or not the game is played on weekend.
A value of 1 indicates that the game is on a Friday, Saturday, or Sunday. A value of 0 indicates
that the game is on a Monday, Tuesday, Wednesday, or Thursday.
RDIV is the ranking of the Cleveland Indians position relative to the other teams within
their division. This division is in the American Leagues Central Division and consists of five
teams (the Detroit Tigers, Chicago White Sox, Minnesota Twins, Cleveland Indians, and the
Kansas City Royals). A value of 1 indicates that the team is in first place and a value of 5
indicates that the team is in last place.
TNIGHT refers to the time of the game as being either a day game or a night game. It is a
dummy variable that takes on a value of 1 if the game is considered a night game and a value of
0 if it is considered a day game.
5 Seasonal data on attendance can be found in Appendix A.
Page | 9

WSTREAK refers to the current number of consecutive wins by the Cleveland Indians at the
start of the game. A value of 0 indicates that it was either the first game of the year or that the
team lost the previous game.
Similarly LSTREAK refers to the number of consecutive losses by the Cleveland Indians at
the start of the game. A value of 0 indicates that it was either the first game of the year or that the
team won the previous game.
FTEMP refers to the temperature in degrees Fahrenheit in Cleveland, Ohio at the start of the
game.
POTHER is a dummy variable which takes on a value of 1 if the team had a promotional
event for the game other than fireworks or Dollar Dog Night and 0 if they did not.
PDOLLAR is a dummy variable which takes on a value of 1 if the team had their promotional
event called Dollar Dog Night for the game and 0 if they did not.
PFIREWORKS is a dummy variable which takes on a value of 1 if the team had a promotional
fireworks event and 0 if they did not.
IINTER is a dummy variable which takes on a value of 1 if the team played against a team
from the National League (NL) and 0 if they did not. There are two leagues within MLB, the
American League (AL) and the National League. Prior to 1997 teams from each division did not
compete against each other during the regular season (MLB.com).
VRIVAL is a dummy variable defined as 1 if the game is against a team in the AL Central
Division and 0 otherwise.

Page | 10

YRIVAL is a dummy variable defined as 1 if the game is against either the Boston Red Sox
or the New York Yankees. These are two teams that are likely to draw larger crowds due to their
large fan bases.6

V. Estimation Results
A. The Estimation Model
For this 2013 & 2014 combined season estimation model, the functional form is a linear
specification.7

The estimated model for the 2013 & 2014 seasons is:
A

= 15,629.6 + 2,838.6

L
STREAK

1,950.7

+ 38.9

I
INTER

F
TEMP

528.6

S
WEEKEND

+ 2,523.9

V
RIVAL

1,329.5

P
OTHER

+5,013.9

R
DIV

446.0

+ 1,999.0

P
DOLLAR

NIGHT

+ 204.0

+ 7,372.9

W
STREAK

P
FIREWORKS

+ 284.7

Y
RIVAL

B. Overall Model
The R-square value for this model is 0.482. This means that, assuming this is the correct
model to estimate game-day attendance, then this model explains 48.2% of the variation in

6 A summary of the selected variables can be found in Tables B1-B3.


7 The results of the model are displayed in Table B4.
Page | 11

attendance. The overall model has an F-statistic of 11.225 with a p-value < 0.001, meaning it is
highly statistically significant.
C. Estimation Coefficients
This model estimates that attendance is affected in the follow ways:
For an increase of one unit in rank, the number of tickets sold for the next game will
decrease by 1,330. This is consistent with the prediction because a one unit increase in rank
means the team is falling behind the first place team, thus weakening their chances of making the
playoffs and ultimately lowering the perception of the team as a quality product. With a p-value
of 0.0025, this variable is highly statistically significant8.
If the game being played is a night game, then it estimates that 445 fewer tickets will be
sold than if it were a day game. The expectation was that night games would increase attendance
since fans are less likely to be at work. This variable has a p-value 0.364 which is not close to
0.05 and thus it is not statistically significant.
Weekend home games increase ticket sales by 2,838 more than game played on Monday
Thursday. This result agrees with the expected sign of weekend games having a positive effect
on attendance. With a p-value of 0.0175 it is statistically significant.
A one degree increase in temperature (Fahrenheit) will result in an increase of 39 more
tickets being sold. This result agrees with the expected sign, but with a p-value of 0.1815 it is not
statistically significant within this model.

8 In this paper, statistical significance means significance at a level of 0.05 or less. P-values are
reported as one-tail p-values.
Page | 12

For each additional consecutive win, there will be 203 more tickets sold for the following
home game. The result agrees with the expected sign, but the p-value of 0.284 means this
variable is not statistically significant within this model.
For each additional consecutive loss, there will be 284 more tickets sold for the following
home game. This result contradicts the expected sign of consecutive losses having a negative
impact on attendance. The p-value of 0.2535 means that this variable is not statistically
significant.
Promotional events increase attendance. When fireworks are used as a promotional event,
7,373 more tickets are sold for that home game. The Dollar Dog Night promotional event results
in 1,999 more tickets being sold for that home game. Other promotional events result in 2,524
more tickets being sold for those games. These events are not mutually exclusive, so if the
Indians have fireworks and Dollar Dog at the same game then 9,372 more tickets will be sold, on
average. Fireworks, Dollar Dog Night, and other promotions all have results that agree with the
expect signs. Fireworks and other promotions have p-values of <0.001 and 0.0095, respectively,
meaning they are both significant. Interestingly, Dollar Dog Night has a p-value of 0.0805 which
puts it close to the boundary of significance.
If the game is an interleague matchup, then 1,951 fewer tickets will be sold. This is an
interesting result because it contradicts Butlers findings of interleague play having a significant
and positive effect on home game-day attendance. The p-value for this variable is 0.082 though,
so it is outside the boundary for significance as classified within this paper.
If the game is against a division rival, then ticket sales will be decrease by 529. This
result is surprising since division games have more importance than other games when it comes
Page | 13

to making the playoffs. The result does not match the expected sign, but the p-value of 0.3105
means this variable is not significant.
If the game is against a non-division rival, then ticket sales will be increased by 5,014.
This result makes sense since the Red Sox and Yankees have large fan bases. This result matches
the expected sign and the p-value of 0.00045 means it is statistically significant.
D. Chow Test
The Chow Test failed to detect a significant difference between the 2013 and 2014 seasons used
for this report. This means that there is not significant evidence of structural change between the
2013 and 2014 seasons, which provides justification for pooling the two seasons of data in the
regression reported in Table B4. 9
E. Multicollinearity
This report examines potential multicollinearity by looking at the Variance Inflation Factor (VIF)
and Pearson correlation coefficients10. Most of the independent variables in this model had VIF
values below two. The main issue of collinearity within this model is the relationship between
weekends and fireworks. The VIFs for fireworks and weekends are 2.406 and 2.349,
respectively. This report concludes that a correlation coefficient for weekend and fireworks of
0.51 as well as the VIFs for each are not strong enough to indicate significant issues of
collinearity.
F. Residual Plot
9 The Chow Test can be found in Appendix C.
10 Pearson Correlations are shown in Table B5.
Page | 14

The residual plot shows a random pattern, indicating that this linear model is a good fit. 11

11 Figure B1 shows the residual plot for this model.


Page | 15

VI. Conclusions
This research paper examined the factors that have influenced the Cleveland Indians home
game-day attendance during the 2013 and 2014 regular seasons. The model used to estimate
these factors was highly statistically significant. The significant variables in this model were:
weekends, rank in division, firework promotions, other promotions besides Dollar Dog Night,
and non-division rivals. The Dollar Dog Night promotion was close to being significant. The rest
of the variables were not significant. This model explains 48.2% of the variation in game-day
attendance. This report did not find strong evidence of collinearity. The residual plot showed that
this model is a good fit.
In the future this report recommends examining the insignificant variables and possibly
excluding them from the model. Another important recommendation is to add new variables to
the model. One example may be a dummy variable for each opposing team to isolate the effect
each team has on game-day attendance. Another useful variable may be one that takes into
account other events taking place in Cleveland or the surrounding area.
One important issue for examining factors that influence game-day attendance is
distinguishing between individual games and entire seasons. Factors such as average income for
local residents may have a major influence on attendance, but would not be a useful variable
within this model since incomes typically change over long periods of time, not day-to-day.
Similarly, the honeymoon effect is not applicable to this paper due to the time periods used.

Page | 16

This report was not able to include the distance variable from Gitter & Rhoads study due
to time constraints. With more time it would be possible to collect the data on potential
substitutes and the effect they have on game-day attendance.

Page | 17

VII. Appendix A
Table A1. Seasonal Attendance for the Cleveland Indians

VIII. Appendix B

Table B1. 2013 Season Data Descriptive Statistics ( n = 80 )


Mean
Std. Deviation
AATTENDANCE
19,661.58
7,320.07
SWEEKEND
0.49
0.503
RDIV
2.29
0.86
TNIGHT
0.68
0.471
FTEMP
71.43
10.461
WSTREAK
1.71
1.766
LSTREAK
0.70
1.316
POTHERS
0.66
0.476
PFIREWORKS
0.24
0.428
PDOLLAR
0.19
0.393
IINTER
0.18
0.382
VRIVAL
0.46
0.502
YRIVAL
0.08
0.265

Page | 18

Page | 19

Table B2. 2014 Season Data Descriptive Statistics ( n = 78 )


Mean
Std. Deviation
AATTENDANCE
18,428.12
7,312.29
SWEEKEND
0.50
0.503
RDIV
3.47
0.950
TNIGHT
0.67
0.474
FTEMP
67.37
12.367
WSTREAK
1.08
1.204
LSTREAK
0.72
1.127
POTHERS
0.38
0.490
PFIREWORKS
0.21
0.406
PDOLLAR
0.17
0.375
IINTER
0.14
0.350
VRIVAL
0.45
0.501
YRIVAL
0.09
0.288
Table B3. 2013 & 2014 Data Combined Descriptive Statistics ( n =158 )
Mean
Std. Deviation
AATTENDANCE
19,052.65
7,319.09
SWEEKEND
0.49
0.502
RDIV
2.87
1.081
TNIGHT
0.67
0.471
FTEMP
69.42
11.585
WSTREAK
1.40
1.543
LSTREAK
0.71
1.222
POTHERS
0.44
0.498
PFIREWORKS
0.22
0.417
PDOLLAR
0.18
0.383
IINTER
0.16
0.366
VRIVAL
0.46
0.500
YRIVAL
0.08
0.276

Page | 20

Table B4. 2013 & 2014 Model Summary ( n =158 )

12 P-values shown are two-tail p-values and have been converted to one-tail p-values in the paper.
Page | 21

Figure B1. 2013 & 2014 Residual Plot ( n =158 )

Table B5. Full Model Correlation Coefficients

Page | 22

IX. Appendix C

Chow Test
H0: There is no structural change between the data for the 2013 season and the 2014 season.
Ha: There is structural change between the two groups.
F = [(RSSpooled (RSS2013 + RSS2014)) / # of parameters] / [(RSS2013 + RSS2014) / (N 2(# of parameters))]
Let :
A = RSSpooled
B = RSS2013 + RSS2014
P = # of parameters
N = # of observations
A = 4,359,997,597
B = 3,986,884,317
P = 13
N = 158
N -2P = 158 24 = 134
A B = 373,113,280
2013: RSS = 2,391,466,814
2014: RSS = 1,595,417,503
RSS pooled = 4,359,997,597
Top = 31,092,773
Bottom = 29,752,868
F = 1.04503
F(12,134) Using an F Table the boundary for rejection is 1.83 for a significance level of 0.05.
1.04 < 1.83 thus: fail to reject the null hypothesis that there is no structural change between the data for
the 2013 season and the 2014 season.
Page | 23

Works Cited
Baseball-Reference.com. 2014 Cleveland Indians. n.d. 1 March 2014
<http://www.baseball-reference.com/teams/CLE/2014.shtml>.
Butler, Michael R. "Interleague Play and Baseball Attendance." Journal of Sports
Economics 3.4 (2002): 320-34.
Castrovince, Anthony. Sellout Streak Etched in Tribe Lore. 4 September 2008. 4
March 2014 <http://m.indians.mlb.com/news/article/3418249/>.
Cebula, Richard J. "A panel data analysis of the impacts of regional economic
factors, marketing and promotions, and team performance on minor league
baseball attendance." The Annals of Regional Science 51.3 (2013): 695-710.
Cleveland Indians Attendance Data. n.d. 1 March 2015 <http://www.baseballalmanac.com/teams/cleiatte.shtml>.
Gitter, Seth R. and Thomas A. Rhoads. "Determinants of Minor League Baseball
Attendence." Journal of Sports Economics 11.6 (2010): 614-28.
MLB.com. Cleveland Indians. n.d. 3 March 2014
<http://cleveland.indians.mlb.com/schedule/>.
Welki, Andrew and Thomas Zlatoper. "U.S. Professional Game-Day Attendence."
Atlantic Economic Journal (1999): 285-98.
Zygmont, Zenon X. and John C. Leadley. "When Is the Honeymoon Over? Major
League Baseball Attendence 1970-2000." Journal of Sport Management
(2005): 278-99.

Page | 24

S-ar putea să vă placă și