DETERMINANTS OF RECORD BREAKS IN SWIMMING A THESIS ...

DETERMINANTS OF RECORD BREAKS IN SWIMMING

A THESIS

Presented to

The Faculty of the Department of Economics and Business

The Colorado College

In Partial Fulfillment of the Requirements for the Degree

Bachelor of Arts

By

Elizabeth Claire Preston

May 2011

Determinants of Record Breaks in Swimming

Elizabeth Claire Preston

May 2011

Economics

Abstract

This study examines record breaks in swimming in order to determine the factors of

athletic success. We use a regression analysis to observe the impact of several variables

on the frequency of record breaks ranging from 1969-2009. The study specifically

focuses on how innovations affect records when introduced to the competitive swimming

world. Proving a strong relationship between technology and record breaks, analysis of

the data shows the introduction of one average new technology results in .345 new

broken records. It also finds the following factors to have a positive significance on the

total number of record breaks: star athletes, a nation’s accessibility to sea, and cold

nations.

KEYWORDS: (Record, Technology, Swimming)

TABLE OF CONTENTS

ABSTRACT ii

1 INTRODUCTION

1

1.1 Background.................................................................................................... 5

1.2 Relevance....................................................................................................... 8

2 LITERATURE REVIEW 12

3 THEORY 22

4 DATA AND METHODOLOGY 31

4.1 Data................................................................................................................ 31

4.2 Technology Factor.......................................................................................... 37

4.3 Variable Correlations..................................................................................... 42

4.4 Methodology.................................................................................................. 43

5 RESULTS AND ANALYSIS

45

5.1 Results............................................................................................................ 45

5.2 Factors Influencing Lifespan of a Record ..................................................... 47

5.3 Factors Influencing Total Record Breaks and Star Athletes.......................... 53

5.4 Additional Consideration............................................................................... 57

5.5 Implications of Results................................................................................... 60

4.2 Further Research............................................................................................ 61

6 CONCLUSION

63

7 SOURCES CONSULTED 66

LIST OF TABLES

3.1 COUNTRY EXPENDITURE ON SWIMMING PER CAPITA IN 2008…. 26

4.1 SUMMARY STATISTICS FOR INDIVIDUAL RECORD BREAK DATA

SET……………………………………………………………………

34

4.2 RECORDS BROKEN EACH YEAR COMPARED WITH TECHNOLOGY

4.3 SUMMARY STATISTICS FOR TOTAL RECORD BREAK DATA SET…

4.4 SUMMARY STATISTICS FOR COUNTRY DATA SET…………

4.5 CHRONOLOGY DESCRIPTION OF SWIMSUIT INNOVATIONS

RANGING FROM 1969-2009………………………………………………

4.6 DESCRIPTION OF VARIABLES………………………………………

4.7 MULTICOLLINEARITY OF VARIABLES ON A COUNTRY BASIS…

4.8 MULTICOLLINEARITY OF VARIABLES ON AN INDIVIDUAL BASIS

5.1 TECHNOLOGY IMPACT ON RECORD BREAKS ON AN ANNUAL

BASIS……………………………………………………………………

5.2 IMPACT ON THE DAY LAPSE OF RECORD BREAKS…………………

5.3 FACTORS INFLUENCING TOTAL RECORD BREAKS FOR EACH

COUNTRY……………………………………………………………………

5.4 FACTORS INFLUENCING STAR ATHLETES FOR EACH

COUNTRY……………………………………………………………………

35

36

37

40

41

42

43

46

48

53

56

LIST OF FIGURES

2.1 PRODUCTION FUNCTION OF AN ATHLETES SPEED

PERFORMANCE GIVEN THE NECESSARY INPUTS OVER TIME……..

18

5.1 TECHNOLOGY IMPACT ON AVERAGE LIFESPAN OF A RECORD…… 51

5.2 STAR ATHLETES IMPACT ON RECORD BREAKS FOR EACH

COUNTRY………...……………………………………………………..........

55

1

CHAPTER I

INTRODUCTION

In the recent Beijing 2008 Olympic Games, swimmers broke twenty-five world

records and sixty-six Olympic records, with only a single previous Olympic record

surviving.1 Some races in the Games even resulted with the first five finishers all

beating the old record, stirring much speculation about the introduction of new swimsuit

technologies.2 More recently, the results in the 2010 Vancouver Olympics indicate a

clear advantage rich countries have in winning medals as the powerhouse nations like

the United States and Germany captured the lead in medal standings with ease. As sport

success continues to cater to the more advanced countries, technology should not only

prove to be a positive factor in past athletic achievement, but its value should also

increase in the future as our society becomes more dependent on technology.3

Many studies have analyzed athletic performance in its most competitive form,

most concluding that the development of superior physical performance is in part a

result of technological innovation. Although many factors contribute to success in sport,

technology appears to be most controversial because it questions the legitimacy of

performance times due to its apparent unfair advantage. It may be the reason why the

1 Ross Tucker. “Swimmers Credibility Crisis: How FINAS‟s Blind Eye is Affecting the Purity of

the Sport.” The Science of Sport (2008): Date Accessed: October, 2010 on

http://www.sportsscientists.com/2008/11/swimsuit-controversy.html.

2 IBID

3 “The Olympic Games.” Date accessed: October 2010 on www.olympic.org

2

richest countries like United States, Germany and China earned the highest medal

counts in recent Vancouver Olympics. Entire industries devote their business to the

development of sport, as national governments and international corporations annually

invest billions in order to sustain public interest in recreation and health.4 How much

difference does technology make? This study is novel as it specifically focuses on how

world and American records are broken by analyzing the causes for improvement in

swimming performance times.

Modern sport celebrates the most exceptional of performances, as summarized

in the Olympic motto, citius, altius, fortius. In obtaining maximum output, society

requires an increase of “scientific intrusion” into the sporting body in order to deliver

the “elusive „edge‟” in competition.5 Sport technology ranges from body techniques,

traditional sport equipment, substances or methods used outside of the competitive

setting, and performance enhancing machines.6

Success in athletics, shown through record breaks at an international level,

resembles a public good. The Coe Report7 argues three reasons why this may be the

case and why governments care about success among their athletes. First, the loud

celebration that comes with breaking records creates pride in one‟s own national

identity. Second, a record break builds positive energy within the swimmer‟s country,

4 Tara Magdalinski. Sport, Technology and the Body: The Nature of Performance.

London, New York. Routledge. (2009): 1-157

5 IBID

6 S. Loland, “Technology in Sport: Three Ideal-Typical Views and Their Implications.”

European Journal of Sport Science 2, no. 1 (2001).

7 I. A. Moosa, L. Smith, “Economic Development Indicators as Determinants of Medal Winning

at the Sydney Olympics: An Extreme Bound Analysis” Australian Economic Papers 43, no.3 (2004):

288–301.

3

helping boost the nation‟s image to the rest of the world, while also potentially

contributing to the sales of their own products and consumption. Finally, success helps

increase athletic participation not only in sport, but also recreation, strengthening the

overall health of society.8

Studies covering performance in swimming have focused specifically on

controlled trials9 or specific case studies.

10 This study differs as it examines each

occurrence of a record break in swimming ranging specifically from 1969-2009.

Analyzing swimming permits more data and analysis, as swimmers perform in many

more events then other professional athletes, providing more opportunities for stars to

set world records. For example, seventeen separate events are available for each male

swimmer in the 2011 ConocoPhillips National Championships, allowing superior

athletes like Michael Phelps to qualify and compete for any event available. 11

For

example, Phelps alone set eleven new records during the 2008 Beijing games. In the

200m butterfly, he set new Olympic records in both the heat and semifinal, and then a

world record in the final.12

Swimmers perform horizontally, which not only promotes

better circulation of oxygen and nutrients throughout the blood, but also avoids hard

impact on muscles, tendons, and ligaments. This type of competition allows for quicker

8 IBID

9 E. Tiozzo, G. Leko and L. Ruzic, “Swimming Bodysuit in All-Out and Constant-Pace

Trials.” Biology of Sport 26, no. 2 (2009): 149-156

10

David B. Pyne, Cassie B. Trewin and William G. Hopkins, “Progression and Variability of

Competitive Performance of Olympic Swimmers.” Journal of Sports Science 22, no.7 (2004): 613-620

11

USA Swimming. National Events. Order of Events. (2010). Date accessed: February, 2010 on

http://www.usaswimming.org/DesktopDefault.aspx?TabId=1476&Alias=Rainbow&Lang=en

12 Brenkus, John. The Perfection Point. New York. HarperCollins Publishers. 2010: 66

4

recovery relative to track and field, allowing competitors to be involved in more events.

In the twelve-month swim year starting September 1st, 2009, there were 6,334

sanctioned, approved and observed meets with 7,237,454 performance times loaded. 13

This amount of data gives researchers much to analyze and competitors many

opportunities to break a record.

Representing an indicator of optimization, record breaks signify the rate of

change in athletic performance, which can potentially correspond to the rate of change

in our global economy. Some economists use record theory to explain the issuance of

patents.14

Economists evaluate sports by using marks of improvement in economic

activities such as manufacturing and software advancements, analyzing technological

change through the frequency of record breaks to better understand discrete decisions

like workers mobility within jobs or success within firms. Record breaks can be more

relevant in answering these questions than actual performance.15

Some people may object to the idea that change in a sport can parallel change in

more central economic activities, because athletics are restricted to human abilities.

However, just as swimming is an individual effort and physical skill, so is writing

software or designing machines. No evidence supports these activities as being different

from one another, thus society is improving in both areas in similar directions. Due to

the many regulations athletes must follow in order to compete, one would expect the

13

Sharon Loving, Swimming Technology Research. Marketing Director. Tallahassee, FL.

(2010): Date Accessed: October, 2010 on swimmingtechnology.com

14

Munasinghe, Lalith, Brendan O‟Flaherty and Stephan Danninger. “Globalization and the Rate

of Technological Progress: What Track and Field Records Show.” Journal of Political Economy 109, no.

5 (2001): 1134

15

IBID

5

rate of technical progress to be slower in athletics. On the other hand, technical

advancements in sports are much clearer than any other economic activity in terms of

their single reason for function: obtaining a better time.16

Background

In order to better understand the evolution of swimming achievement, we must

examine which factors have had the biggest impact on athletic success in the past.

Advanced technology has long been a factor in the development of swimming.

Beginning in the early twentieth century, exercise science emerged in order to observe

the human body and hope to augment performance by expanding biological parameters

and physiological abilities.17

Adolph Kiefer, a 1936 U.S. Olympic swimming champion,

contributed some of the first key innovations to the sport. 18

Founder of aquatics

company Adolph Keifer & Associates, Kiefer helped with the progress of goggles and

non-turbulent lane lines, inventing the first nylon racing suit in 1948. The evolution of

competitive swimsuits in particular proves notable; initially made from wool, they

progressed to Mosquito netting in 1920s, board shorts in the 1940s, racing briefs in the

1950s, lycra in the 1970s, and then finally full bodysuits in 1998.19

A huge technical

jump occurred with the introduction of the first bodysuit: Adidas JetConcept. Adidas

adopted technology used by commercial aircrafts, which helps reduce pool drag and

16

IBID

17

Magdalinski, 5

18

Margaret Schauer, “How Much is Too Much? Evaluating the Impact of Swimming

Innovations” USMS Swimmer (2006): Date Accessed: October, 2010 on www.usmsswimmer.com.

19

IBID

6

influence how water flows around the swimmer‟s body. Designers also incorporated

small riblets into the suit in order to channel water, causing a shift in turbulence and

decreasing the amount of water carried by the swimmer.20

Since the introduction of full

body suits, the media has both marveled and criticized the suits‟ extraordinary light

weight, the welded seams, and even the fabric which attempts to mimic the skin of a

shark or the shape of a jet, luring many competitors into believing they can breaks more

records than ever before. However, the real secret of the new suits lies in their ability to

compress the swimmers body into a more streamlined shape while also prohibiting any

fat or muscle from protruding into the water.21

Matt Zimmer, director of TYR Sport

Inc., believes there is no question as to whether or not these body suits make swimmers

faster: “the proof is in the decreasing times.”22

Furthermore, once Speedo introduced

Fastskin technology, thirteen out of fifteen world records were broken that same year at

the Sydney 2000 Olympics. Swimmers donning the new suit won eighty-three percent

of all medals.23

When Inge de Bruin broke ten records in 2000, the Netherlands

swimmers accredited her huge improvements to modern training regimes and the “lift

provided by a new bodysuit.”24

20

Adidas. (2011): Date Accessed: October, 2010 on www.adidas.com

21

Brenkus, 65

22

Schauer: Date Accessed: October, 2010 on www.usmsswimmer.com.

23

Speedo. Explore the World of Speedo. (2011): Date Accessed: October, 2010 on

http://www.speedousa.com/shop/index.jsp?categoryId=3642946

24

Christopher Clarey. “Syndey 2000: Swimming; Beginning Tomorrow, All Eyes Will Be

Focused on One Pool.” Sydney. The New York Times. Sports. Date accessed: January, 2011 on

http://select.nytimes.com/gst/abstract.html?res=F20D1FF7385C0C768DDDA00894D8404482

7

By declaring the LZR suit legal for competition, FINA opened a Pandora‟s Box

in the swimming world. Suit manufacturers switched completely to polyurethane fabric

and discovered ways of trapping air bubbles in the suit for better buoyancy. Critics

accuse the body suits of “technologic doping” by the unfair increase in buoyancy, which

is a direct violation of the performance enhancing rules set by the international

swimming federation known as FINA.25

In reaction to this swimsuit craze and the

aftermath of the World Championships in Rome, FINA acknowledged that further

measures needed to be taken because the huge surplus in record breaks, thus they

banned full body suits and the polyurethane material. 26

Controversy stirred as a main

concern still lingered; would all the records broken by swimmers wearing the full body

suits still stand or be noted with an asterisk (dubious merit)? FINA then sent out a list of

approved suits, the LZR racer being one of them, to solve this issue.27

Technology and marketing not only helps athletes achieve higher goals, but also

attracts new interest into the sport. However, the question remains how much does

technology affect performance and when will it be considered unfair to the game?

When does scientific intrusion become inappropriate? Are all faster times due to the

increase and higher quality of technology? How do economists even measure sport

technology? In this study, I will not answer the ethical questions, but will focus on the

quantitative measurable one. I expect to see technology have a significantly positive

25

Gina Kolata and Jere Longman, “As Swimming Records Fall, Technology Muddies the

Water” The New York Times (2008): A1

26

IBID

27

Brenkus, 65

8

impact on better trials times resulting in a higher rate of record breaks in swimming

events from 1969-2009.

Relevance

Reviewing the rate of record breaks since 1969, apparent jumps in frequency

occur whenever a technical change is introduced. Overall, the rate of record breaks has

increased significantly over time especially in swimming events.28

Based on data

provided by usaswimming.com, the average number of records broken each year

amount to nearly thirty-one. When the first full bodysuits launched in 2000,29

forty

record breaks occurred in the following year. The records continued to grow and when

most every swimmer conformed to the high quality body suits in 2008,30

swimming

times resulted in eighty-four record breaks that year and seventy-five the following

year. During the World Championship games in Rome, a total of forty-three world

records were broken in the span of eight days while most swimmers at the games were

donning the new high-tech bodysuits made by TYR, Jaked, Arena, and Speedo.31

In the

1500 LCM freestyle, American Kate Ziegler smashed the oldest held record held by

Jane Evans by nearly ten seconds in June, 2007.32

Coincidentally, Speedo‟s new

Fastskin Pro had just been released, which Ziegler donned during her groundbreaking

28

“USA Swimming.” In Depth Event History. (2010): Date accessed: October 2010 on

http://www.usaswimming.org/DesktopDefault.aspx?TabId=1476&Alias=Rainbow&Lang=en

29

Schauer: Date accessed: October 2010 on www.usmsswimmer.com.

30

Kolata: A1

31

“Roma 2009.” (2009) Date accessed: February, 2011 on www.fina.org

32

Saslow, Eli. “New Suit Makes Splash in Debut.” Washington Post Staff Writer. The

Washington Post. (2007): Date accessed: October 2010 on www.washingtonpost.com

9

race.33

Technology appears to be having a huge effect on the performance in swimming,

but the question still remains on the degree to which technology is accountable for the

record breaks.

Although the occurrence of record breaks keeps fans interested and the game

exciting, the multitude of records broken after the introduction of the full bodysuit is

beginning to devalue and question such an accomplishment. In February 2008, the

world record for the 100m swim was 47.84 seconds, and in the previous thirty-six years

the record dropped .09 seconds a year on average.34

When swimmers first donned the

Speedo LZR racer in the Beijing Olympics six months later, the record was broken five

times and the average dropped by .13 seconds every month.35

Nearly four dozen world

records were broken in a six-month span after major swimming companies introduced

these high-tech suits.36

Usually multiple swimmers would break the record in many

events, but only the winner‟s time was counted. After the games, only two records

remained unbroken, both set four years earlier at the Sydney Games: Ian Thorpe‟s 400m

freestyle and Inge de Brujin‟s 100m butterfly.37

All records broken in the Beijing

Olympic Games were by swimmers wearing the new LZR racer suit.38

The NASA

approved suit not only costs $550, but it also takes nearly an hour to get into with the

33

IBID

34

Brenkus, 64

35

IBID

36

Kolata, A1

37

Brenkus, 66-69

38

IBID

10

help of assistants. With the new Jaked suit introduced post-Olympic games, more

records are predicted to fall.39

In addition to swimsuits, newer pools are designed to help absorb wave motion.

The standard competitive pool has a typical minimum depth of two meters. A deeper

pool will reduce the impact of reverberated waves created by the swimmer.40

For

example, the Water Cube pool, which held the 2008 Beijing Olympics, is three meters

deep relative to the standard two meter deep pool in order to alleviate resistance.

Furthermore, Olympic pools are now ten lanes wide for eight swimmers in order to

have the outside lanes serve as buffers to keep waves from reverberating. Officials now

employ plastic buoys as lane dividers in order to redirect water downward instead of

outward. Starting blocks are now nonskid in order to activate a faster take-off and

videos are available in order to monitor stroke counts, distance per stroke, split times,

and biomechanics of takeoffs. The ears of swimmers are even pricked post-race to test

for the difference in lactic acid levels.41

Technology appears to be the most progressive

change affecting swimming and controversy closely follows.

Critics argue about how technical innovations create an inappropriate advantage

for swimmers. The aid of new technology unfairly contrasts today‟s swimmers with

those of the past. Comparing current record times with times made in the 1960s appears

significant, but if you subtract the high-tech swimsuits, the non-turbulent lane lines, the

backwash absorbing gutter system, the deep water depth, flip turns, and goggles, it

39

IBID

40

IBID

41

Kolata, A1

11

would be interesting to see if these times are comparable. Sokolovas also wishes “he

could compare results from the ‟70s, ‟80s and ‟90s to the current results to know exactly

how they are improving because of the suits, technology and training.”42

In order to

provide this solution, this study will show exactly how much impact technology has on

record times, and if we can compare all swimmer abilities regardless of the decade in

which they competed. It also must be understood this paper does not seek to belittle the

natural talent of athletes, but just to understand the process of how excellence is

achieved.

The following chapter will review past studies relevant to athletic performance

in order to determine which factors are most important in influencing record breaks.

Chapter III discusses how this study applies past theory to the current issue, outlining

each significant variable and how we choose to define them in the final models. Chapter

IV then explains how the data was abstracted and what methodology is most

appropriate in handling the frequency of record breaks. Chapter V then presents the

results in several tables, discusses the possible implications, areas of analytical concern,

and any other additional consideration necessary to fully grasp the outcome. The paper

then closes with a final concluding section, summarizing each chapter and the

significance of the study.

42

Kolata, A1

12

CHAPTER II

LITERATURE REVIEW

Several factors must be considered when determining the success of a specific

athlete. In order to recognize which factors are most important for record breaks, we

analyze past sport studies. Past research has shown that the demographic and economic

characteristics of a country have significant explanatory powers for their athletes‟

abilities to prosper in athletic competition. Ever since Ball introduced the economic

model to determine athletic success in 1972,1 follow-up studies emerged, further

investigating relevant influences. In order to lay the foundation for this analysis, these

past studies found the following factors to be most influential in sport success: GDP,

income per capita, population, participation, geography, and technological

advancements.

Most studies recognize GDP and income per capita as the best predictors of

athletic performance because they measure the resources available to athletes regarding

health benefits, training, sponsorships, and infrastructure. Assuming athletic talent is

equally distributed across the world, Bernard and Busse show that success within a

nation defined by Olympic medal counts depends greatly on total GDP because income

per capita and population affect success at a similar level. For example, in the Atlanta

1Donald W Ball, “Olympic Games Competition: Structural Correlates of National Success”

International Journal of Comparative Sociology 12 (1972): 186-200

13

Games, the world‟s poorest 1.5 billion people won only three percent of the medals.2

Similarly, other studies handle GDP and income per capita on separate levels, though

still discover a positive relationship with medal counts. Moosa and Smith believe that

using total GDP is most valuable in determining sport success because it is the correct

measure of volume and a better variable when the objective is to measure the

effectiveness of medal winning. Total GDP is also quantitatively more related to total

medal winning than GDP per capita. Studies also use total GDP and population as

separate variables to see how each affect the dependent variable.3 For example, Johnson

and Ali prove how high income nations have a more pronounced affect in the winter,

while population has a higher affect in the summer.4 How a study defines athletic

success indicates which variable is better to use within the model: total GDP or GDP

per capita. If the output is total medal winning then total GDP is more appropriate, and

if the model tests medal winning per capita, then GDP per capita should be used. GDP

per capita is more appealing as it measures the country‟s ability to pay the costs needed

to send athletes to the Games, thus may be associated with a higher quality of training

success.5

Population of a nation and participation in events also contribute to athletic

success. A larger population size increases success as the nation has a greater talent

pool to choose from. Many researchers use population to predict the number of medal

2 I.A. Moosa and L. Smith, “Economic Development Indicators as Determinants of Medal

Winning at the Sydney Olympics: An Extreme Bound Analysis” Australian Economic Papers 43, no. 3

(2004): 288–301

3 IBID

4 D. K. N. Johnson and A. Ali. “A Tale of Two Seasons: Participation and Medal Counts at

the Summer and Winter Olympic Games.” Social Science Quarterly 85, no. 4 (2004): 974-993

5 Moosa, 293

14

counts for a particular nation. Specifically, Bernard and Busse prove the number of

medal worthy athletes should be proportional to the country‟s share of the world

population.6 Johnson and Ali, who also use population when estimating the total medal

counts, points out the variable‟s significance by showing how countries in 1956 that

won at least one medal averaged six times the population than those that did not win

medals. That figure only drops to five times the population in 1996.7 Conversely,

Rathke and Woitek only found population to be significant with medal counts for richer

countries.8

Furthermore, athlete participation increases the chance of winning more

Olympic medals as more opportunities are available. Kuper and Sterken focus on

significant factors contributing to participation like the host nation, the nation‟s distance

from the games, emancipation, income per capita and legal systems. Similarly, Johnson

and Ali estimate how income per capita affects participation as it costs $260 in GDP per

capita to send an athlete to the games,9 thus richer countries will have more success in

sport as they are able to involve more athletes in competition. Moosa and Smith verify

the impact athlete participation has on winning medals in the Olympic Games.10

Athletic success also depends on the geography of the athlete‟s nation. Johnson

and Ali distinguish nations by measuring the average amount of frost experienced in a

6A.B. Bernard and M.R. Busse, “Who Wins the Olympics: Economic Resources and Medal

Totals” Review of Economics and Statistics 86 (2004): 413–417

7 Johnson and Ali (2004): 974-993

8 Alexander Rathke and Ulrich Woiteck. Economics and Olympics: An Efficiency Analysis

(Working Paper No. 313.) University of Zurich (2007): 12

9 Johnson and Ali (2004): 974-993

10

Moosa, 288–301

15

winter month, in which he finds that colder climates do better than warmer ones.11

Kuper and Sterken also observe geography, though they measure this factor using

latitude and conclude that geographical data does explain athlete participation, affecting

overall success. Altering this definition, Pfau assigns a dummy variable based on the

past achievements of five regions: Soviet, Scandanavia, Germanic, Alpine, and North

America. Pfau discovers stronger performances from Germanic and North American

regions.12

Many different definitions are used in order to measure a countries location,

but most studies show how a country‟s geography has an effect on success in athletics.

Finally, the host nation factor greatly affects the outcome of athlete performance

as most studies agree it may increase familiarity with infrastructure, influence biased

referee calls, offer different events, follow home regulations and time-zone, while also

increasing participation and morale with reduced travel costs and more audience

support.13

Johnson and Ali show how the host nation affects not only medal counts, but

also participation within the games.14

Pfau even proves how Italy collected seven

additional medals due to their host city advantage in the 2006 Olympics.15

Neighboring

nations also have higher medal counts than their peers, but only during the summer

games.16

While every relevant study includes the host nation variable due to its

11

Johnson and Ali (2004), 974-993

12

Wade D. Pfau, “Predicting the Medal Wins By Country at the 2006 Winter Olympic

Games: An Econometrics Approach.” The Korean Economic Review 22, no. 2 (2006): 233-247

13Bernard, 413–417

14

Johnson and Ali (2004), 974-993

15

Pfau, 233-247

16

IBID

16

significance, the values differ in each study as the varying explanatory variables react

differently with each other in every model.

Many other variables need to be taken into consideration when analyzing the

reasons for success. The difference in political systems or gender appear to have a small

effect on athlete participation as Johnson demonstrates,17

while other studies believe the

soviet socialist countries increase the chance of success.18

Both Kuper and Johnson/Ali

find that socialist countries send more athletes into athletic competition resulting in

better success with medal counts at the Olympic Games.19

Pfau also examines past

share of medals in order to take into account the momentum of a country and existing

infrastructure.20

This variable can be thought of as a veteran factor where countries and

athletes who have performed well in the past expect some inertia in winning future

medals. Other factors include the importance and availability of an event to society.21

A

country offering more opportunities for citizens to compete in sport will perform better

at a higher level due to experience. Furthermore, society will be more engaged in events

popular within the country, thus the country should expect to perform better within

those particular events.

17

Johnson and Ali (2004), 974-993

18

Rathke: 12 and Johnson and Ali (2004), 974-993

19

Gerard Kuper and Elmer Sterken, The Olympic Winter Games: Participation and

Performance (Working paper). Department of Economics. University of Groningen (2001): 11

20

Pfau, 233-247

21

Rathke, 12

17

Government expenditure on athletics also appears to influence sport success.22

Moosa and Smith feel that if a country invests more money on health and education, as

opposed to defense (military expenditure), then more money should also be expected to

support athletic expenses.23

Furthermore, boycotted Games,24

the presence of doping, 25

world wars, accurate timing,26

consumption of cigarettes, unemployment and gender

equality27

have all been shown to have an impact on success in athletics.

Kuper and Sterken provide a figure (FIGURE 2.1) in order to better understand

how an athlete maximizes his/her performance.28

Over time, athletic success should

occur with the average speed consistently increasing in the upward sloping curve,

representing the natural trend of athlete development. The function also illustrates

diminishing returns to any given input, thus over a period of time, the rate of speed will

begin to decelerate as it reaches its maximum point. The only factor that can shift the

function higher is a major technical innovation, thus technology should be valued

highly because it raises the production function to a new higher level. The vertical jump

22 Moosa, 300

23

Moosa, 288–301

24

Kuper, 5

25

W. Maennig, “On the Economics of Doping and Corruption in International Sports”

Journal of Sports Economics 3, no.1 (2002): 61-89

26

Munasinghe, Lalith, Brendan O‟Flaherty and Stephan Danninger. “Globalization and the Rate

of Technological Progress: What Track and Field Records Show.” Journal of Political Economy 109, no.

5 (2001): 1143

27

Moosa, 298

28

Gerard Kuper and Elmer Sterken. “Endurance in Speed Skating: The Development

of World Records.” European Journal of Operational Research 148, no. 2 (2004): 293-301

18

also signifies an increase in marginal productivity as the slope steepens at any fixed

level of input.

FIGURE 2.1

PRODUCTION FUNCTION OF AN ATHLETES SPEED PERFORMANCE GIVEN

THE NECESSARY INPUTS OVER TIME

Few previous studies analyze how technology impacts medal counts in the

Olympic Games. However, a few studies focus on specific technologic advances in

determining success. Kuper focuses on the positive impact technical innovations like

the klapskate have on speed skating using world records as the dependent variable from

1893 to 2000 for both men and women.29

He also approximates a benchmark for future

29

Kuper, Gerard H. and Elmer Sterken. “Endurance in Speed Skating: The Development of

World Records.” European Journal of Operational Research 148, no. 2 (2004): 299

19

world records given the current level of technology based on the klapskate‟s impact on

the long-run limit value.30

Similarly, Munasinghe tests the influence of technology and globalization on

both international records set by anyone and local records set by members of a fixed

population (United States and New Jersey Highschool). 31

This study defines technology

as better equipment, more accurate timing, elimination of smoking, improved running

surfaces, nutrition, medical care, and training techniques. Including men‟s records from

1896, Munasinghe shows how technology maintains the frequency of record breaks in

track and field by identifying technical changes with the “number of draws,” and then

applying a computational assumption with logarithmically proportional models.32

However, no evidence indicates this process as speeding up or slowing down.

Comparably, Haake also assesses the effect of technology in sport; however, he

measures success with records through a performance improvement index in only four

Olympic events.33

Haake examines specific occasions when a new technology was

introduced to the game and found its specific effects on performance using a parametric

study. Demonstrating that all events owe some improvement of performance to

technology, Haake uses data spanning 1845-2004 to find that cycling is most sensitive

to the influences of technology. The index shows how cycling times have improved by

221% over 111 years, 45% of which are accountable to technological improvements.

30

IBID

31

Munasinghe, 1132

32

Munasinghe, 1137

33

S.J. Haake, “The Impact of Technology on Sporting Performance in Olympic Sports” Journal

of Sports Sciences 27, no. 13 (2009): 1421

20

Specifically, he focuses on the introduction of the new carbon fiber frame in the 1980s

and how it increases bike speed by 5km/h, 24% of the improvement due to its

enhancement in aerodynamics.34

The performance improvement index proves to be

useful in evaluating the effect of an external intervention in sports measured using time

or distance.

A case study conducted by Pyne estimates progression and variability in swim

times in order to show which factors affect performance.35

This study examined 676

official race times comprised of twenty-six US and twenty-five Australian Olympic

swimmers in the twelve- month period leading to the 2000 Olympic Games.

Progression was defined as the percent change in mean performance, while variability

was measured as the coefficient of variation in performance of an individual swimmer.

Within their sample, results show how additional enhancement has improved the

performance times of athletes by approximately 0.4%.36

A more recent study conducted by Tiozzo, et al. proved how new body suits

worn by competitive swimmers improved performance in controlled trials.37

They

specifically focused on Speedo‟s Fastskin suit in the 50m crawl race. The research

analyzed fifteen male national and international level swimmers who swam two trials,

one in a regular suit and one in the body suit. The test resulted in an overall faster race

34

Haake, 1421-1431

35

David B. Pyne, Cassie B. Trewin and William G. Hopkins, “Progression and Variability of

Competitive Performance of Olympic Swimmers.” Journal of Sports Science 22, no.7 (2004): 613-620

36

IBID, 618

37

E. Tiozzo, G. Leko, and L. Ruzic, “Swimming Bodysuit in All-Out and Constant-Pace

Trials.” Biology of Sport, 26, no. 2 (2009): 149-156

21

time for the body suit, specifically in the turn time, split time, streamlining and turn

kicking. This study concluded that body suits enhance performance by 1.6% (.41s).

They also measured the differences in fatigue with each swimsuit. While swimming in

the Fastskin, the racers‟ blood lactate concentrations and heart rate were significantly

lower even though the number of strokes was the same in both suits.38

In conclusion, past research shows which variables should be incorporated in the

final models in order to best predict athletic success. We also validate the focus on

technology by providing studies that have already discovered the degree to which

technology makes a difference in sport success. Due to the limited approaches and

scenarios existing, this study hopes to add another perspective to the current research by

redefining technology and achievement through record breaks.

38

IBID

22

CHAPTER III

THEORY

The purpose of this research is to study which factors affect record breaks in

swimming. In other studies, technological innovations have proven to be a major reason

why athletes continue to break records consistently and at a higher frequency. However,

this study differs from the previous literature as it observes all record breaks from 1969-

2009, while controlling for economic factors. With so many different types of technical

alterations, it is nearly impossible to account for every change that has shaped

swimming into its current form. Thus, I specifically narrow the definition by including

all changes in swimsuits and fabrics as they have created heated controversy in recent

swimming news. Other major technologic changes to the sport have been included by

discretion; however, equipment serves as the major characterization of technology.

By analyzing the performance times of past record breaks, the study examines

the progression of optimal times. Along with technology, this paper will also be adding

significant economic factors into the models in order to better explain the frequency of

record breaks. Gross Domestic Product and population of the swimmer‟s home nation

prove to be big contributors in sport performance. From the literature review, older

studies prove these variables significance when examining factors for athletic success

specifically with medal wins in the Olympic Games. GDP of a nation for that year will

allow the model to take into account the wealth of the swimmer‟s national country.

23

Population will show how many people have the opportunity to break a swimming

record within that nation. However, they may not play as significant a part in the

occurrence and quality of record breaks because the observations include mainly the

wealthiest countries. In fact, only twenty-five countries have provided record-breaking

athletes since 1969 with most breaks earned by USA (288), Germany (154), and

Australia (103).1

Studies also show how location affects athletic performance. An athlete proves

to do better if they compete within their home country. A similar, but less significant

influence holds true for an athlete competing in a country neighboring his/her home

nation. I consider these factors by assigning dummy variables to an athlete competing in

his/her home or a neighboring nation. A record will more likely be broken close to the

athlete‟s home because of reasons discussed previously: national support, travel costs

and availability.

Experience in record breaking also has a huge influence on whether a swimmer

breaks another record. If a swimmer has proven his/her ability to break a record in the

past then that athlete has the highest chance of breaking another record because they

have already proven to be the fastest at that time. Thus, I have assigned another set of

dummy variables to the record breakers who were already the past current record holder

or if they were the holder for the past four years of that particular event. This factor

serves as a proxy for the veteran variable as it suggests past record breakers are more

likely to break another record due to momentum and proven capability.

1 “USA Swimming.” In Depth Event History. (2010): Date accessed: October 2010 on

http://www.usaswimming.org/DesktopDefault.aspx?TabId=1476&Alias=Rainbow&Lang=en

24

Specific swimmers tend to break many records, thus accounting sport success to

natural born athletes. Swimmers such as Mark Spitz (USA), Kornelia Ender (GER), and

Michael Phelps (USA), are partly responsible for the high record breaks in 1972, 1976

and 2009, respectively. In order to account for these athletes, I have created a star

athlete dummy variable by denoting a “1” to every athlete that has broken at least three

records in any event or year. Furthermore, since star athletes are responsible for many

record breaks, we also explore this variable as a dependent variable to examine which

economic factors help exceptional athletes compete at their fullest potential.

Geography is suggested as another factor contributing to record breaks. I define

geography as the latitude of a countries capital in order to find the vertical distance

between the swimmers home nation and the equator. Colder nations tend to outperform

competitors in both summer and winter events, thus the higher the latitude, the more

likely an athlete should break a record. Again, due to the exclusion of all other

competing nations, the results for latitude may not be as accurate if most of the

countries included are already considered cold nations. Also, athletes will most likely

not reside in the exact latitude specified especially depending on the size of the country.

For example, USA‟s latitude is pinpointed at 41.61°N, but if the athlete lives and trains

in San Diego, California, then the corresponding latitude should be 32.49°N, nearly ten

degrees lower then that athletes assigned latitude.

There are many factors this study should consider, yet we lack sufficient data to

include. However, we will proceed to outline and discuss the potential impact of the

following variables: the number of official meets offered each year, popularity,

government expenditure on sports, sponsorships, doping regulations, technique changes,

25

sports nutrition and the quality of technological innovation. As mentioned before, the

opportunities available for swimmers to break a World or American record are

abundant; however, our data does not cover how opportunities have changed since

1969. Availability of a certain sport within a specific town, nation or even the world can

certainly contribute to the number of records broken each year.

Even if a country is rich, how much of their expenditure and focus is spent on

swimming? In other words, the importance of a sport reflects how well a country will

do because they will offer more opportunities to participate and compete. If one

swimmer has only one chance a year to break a record relative to another swimmer who

has twenty chances to break a record, odds of breaking a record favor the swimmer with

more opportunities based on pure probability. For example, swimming is popular in

Australia due partially to its close relationship with the ocean and long-standing

tradition. Even though USA is a richer nation, the Australian government spends more

money per capita on swimming activities as shown in TABLE 3.1, rewarding the nation

with many record-breaking swimmers. USA has a higher population and GDP, yet both

countries sent the same amount of swimmers to the 2008 Beijing Olympic Games.

Furthermore, a higher government expenditure on sports can lead to better training and

coaching, both essential elements to the development of an athlete.

26

TABLE 3.1

COUNTRY EXPENDITURE ON SWIMMING PER CAPITA IN 2008

POPULATION GDP per

capita

SPORT

DEVELOPMENT

COSTS per capita

EVENT COSTS

per capita

SWIMMERS SENT

TO BEIJING GAMES

AUS 21,007,310 36,614.746 .026 .065 43

USA 304,830,000 43,397.303 .012 .019 43

SOURCES: “Swimming Australia Annual Report.” 2008. Finance Report. Date accessed: February

2011 on http://swimming.org.au/

“United States Swimming, Inc.” Financial Statements and Supplemental Schedules.

Year 2008. Date accessed: February 2011 on www.usaswimming.org

The value of a specific sport to a particular society can be measured with a

country‟s expenditure on the sport, but information regarding every country‟s

expenditure on sport since 1969 proves to be impossible. Moosa and Smith show how

expenditure on health can serve as an indicator of that country‟s expenditure on sport,

yet difficulty lies in obtaining this data as well. However, aside from rare exceptions

like Australia and USA, Gross Domestic Product per capita will usually coincide with

availability because richer nations can afford to hold and promote more meets and

events. Lack of data regarding the number of meets held by each country limits

investigating this point further. Furthermore, richer nations have the ability to send their

athletes to meets outside of their nation.

Popularity of sport and availability of practice venues would also affect the

number of competing swimmers within a nation. The number of swimmers a nation

generates will certainly affect record breaks because not only will the competition be

higher, but the chances to break a record will increase. For example, Germany, one of

the richest countries, sent four hundred and thirty-six athletes to the Sydney Olympics,

while Vietnam, a country of similar population, but one of the poorest countries, sent

27

seven athletes.2 No athlete from Vietnam has ever broken a swimming record.

However, accounting for every open meet and competing swimmer each nation

produces each year since 1969 proves to be impossible due to the absence of

information. Thus, we must assume GDP per capita will capture some of these

influences in the model.

The emergence and growth of sponsorships could also have an effect on annual

record break counts. As sponsorships reward more competitive swimmers with an

increase in funds, more record breaks should occur due to the newly created

opportunities and higher quality of equipment and coaching. However, little information

exists on the number of sponsored swimmers each year especially for every country.

The presence of doping in swimming events appears to be the hardest variable to

measure. An athlete taking performance enhancing drugs would certainly affect record

breaks. Again, it would be impossible to observe which athletes violate or abide by the

drug policy.3 We could potentially examine how each country‟s doping regulations

differ in severity, but this information would not only be difficult to quantify, but also

inaccessible for each country and year. Thus, we must acknowledge this variable as a

plausible factor, but it cannot be applied to the model without avoiding large

assumptions about the athlete.

Changes in technique are also abundant in swimming. Shaving off milliseconds

from a swimmer‟s lap time could make the difference in breaking a record or not.

2 I. A. Moosa and L. Smith, “Economic Development Indicators as Determinants of Medal

Winning at the Sydney Olympics: An Extreme Bound Analysis” Australian Economic Papers 43, no. 3

(2004): 288–301

3 Maennig, W. “On the Economics of Doping and Corruption in International Sports.”

Journal of Sports Economics 3, no. 1 (2002): 61-89

28

Unless we interview each competing swimmer since 1969, all technique changes would

be impossible to account for as no public knowledge exists about the technique used by

every swimmer. In fact, most technique changes occur secretly in order to have an edge

over the competition. For example, an engineering professor of Rensselaer Polytechnic

Institute, Tim Wei, has been conducting confidential experiments on how water flows

and reacts to a swimmer‟s body.4 By sending his results to top swimming authorities,

coaches are able to better instruct swimmers for maximum performance by eliminating

superfluous milliseconds from their lap times. This type of innovation creates more

potential to break records; however, technique change cannot be accounted for in the

model due to its subtlety and differences among all swimmers.

Nutrition and weightlifting also serve as major components in enhancing an

athlete‟s performance. With the growing knowledge of sports nutrition, athletes are able

to energize their bodies properly for high competition.5 Furthermore, weight training

varies for different sports, thus as our knowledge base grows towards appropriate lifting

for specific sports, more swimmers will build certain muscles for optimal performance

in the pool. For example, most of the twenty-seven swimming records Germany broke

in 1976 were achieved by German women like Kornelia Ender. When searching for the

reason in Germany‟s swimming success, The Tuscaloosa News reported the apparent

difference in American and German swim programs, claiming most “American women

do not lift weights, while the German and Russian women can almost always be picked

4 Michael Hill. “Swimmers „See‟ the Water with Help of Professor.” (2008): Date accessed:

October 2010 on http://swimming.teamusa.org/news/2008

5 Charles B. O‟Neil and Peter G. Akintunde. Optimizing Athletes’ Performance Excellence and

Wellness through Nutrition and Exercise. (Working Paper). University of Calabar (2010)

29

out of a swimming lineup because their shoulders resembles those of pro football

players.”6

Finally, this study focuses on technology as the main variable in improving

athletic performance. As discussed in Chapter II, the production function in TABLE 2.1

measures an athlete‟s potential in success with technology as the only factor to

completely raise the output level. Record breaks can be regarded as the output resulting

from a production function in which the explanatory variables serve as the inputs

creating the base function of an athlete‟s potential. Diminishing returns also

appropriately apply to the growth of record breaks as we would expect the difficulty in

breaking records to increase as every additional record was broken, thus the frequency

would slowly decline. In order to sustain the occurrences of record breaks, technology

needs to improve in order avoid diminishing returns. In order to define innovations, this

study obtains information on every introduction of a new swimsuit, fabric, or other

major innovation since 1969. The value of each technological innovation should have a

huge effect on the incidence and quality of record breaks, but this study fails to answer

how the quality of technology changes due to the difficulties in quantifying the

influential values of each innovation.

The technology variable has many limitations and assumptions. First, it is

impossible to include every single technology innovation made for the sport of

swimming due to its abundance, confidentiality and time. Thus, the innovations used

are based on pure discretion, which may bias results. Also, the study must assume every

6 “Name That Tune.” The Tuscaloosa News. (1976): Date Accessed: February, 2011 on

http://news.google.com/newspapers?id=rUYgAAAAIBAJ&sjid=650EAAAAIBAJ&pg=6816,74269&dq

=history+of+germany+competitive+swimming&hl=en

30

technology introduced is available to every competing athlete, however this may not be

the case as poorer countries or unsponsored athletes may not be able to afford expensive

swimsuits. Finally, the data assumes each technology introduced has an equal effect on

a swimmer‟s potential performance. We further explain exactly how technology is

defined, found, and implemented in the next section.

31

CHAPTER IV

DATA AND METHODOLOGY

Many factors contribute to the frequency and magnitude of record breaks in

athletics. By reviewing the existing literature, this chapter will summarize the data

collected for each variable utilized later in the models and mention all possible

limitations. The data will be described using summary statistics and confirmed through

pair-wise correlations between all variables. We will then discuss the methodology used

in this study based on the different dependent variables tested.

Data

The data focuses on world swimming record breaks ranging from 1969-2009.

Three different data sets are included in the analysis: one includes the number of total

record breaks, both American and world; another incorporates each countries annual

record break; while the last contains every individual record break. The total record

break data set includes only 41 observations, using the annual number of records broken

each year since 1969. Due to limited observations, the other data sets hope to strengthen

the results by supplying more observations. The individual record break data set

includes 666 observations, examining every athlete who broke a World record, and the

country data set contains 1025 observations by including the twenty-five countries that

have ever contributed to a world record in swimming. This data set includes each

32

involved country every year over the forty-one year span, thus resulting in 1025

observations.

The data set describing every individual record break is provided by

usaswimming.org1 and confirmed by Christer Magnusson.

2 These two sources provide

the specific athlete‟s name, nationality, event, gender, performance time, location and

date of the break. In order to obtain GDP per capita for each athlete, ers.usda.gov

provides the measures of both population3 and Gross Domestic Product.

4 The data year

range was limited from 1969 to 2009 because ers.usda.gov only stretched their

international observations to 1969 and usaswimming.org only provided record breaks

up to 2009. Furthermore, the data combines all German athletes, East and West,

because the GDP/population source only provides measures for Germany. The study

acknowledges that merging East and West Germany may skew results due to the post-

reunification phase; however, public sources limit the data to this option. TABLE 4.1

provides the summary of statistics for the individual record break data set. Specific

latitudes of countries were taken from Daniel Johnson‟s personal data set online used

1 “USA Swimming,” In Depth Event History. (2010): Date accessed: October 2010 on

http://www.usaswimming.org/DesktopDefault.aspx?TabId=1476&Alias=Rainbow&Lang=en

2 Christer Magnusson, Records. (2010): Date accessed: October 2010 on

http://www.scmsom.se/records/statistics/World%20Record%20Progression.htm

3 Matthew Shane, Historical Population and Growth Rates in Population for Baseline

Countries/Regions. Economic Research Service. (2010): Date accessed: October 2010 on

http://www.ers.usda.gov/Data/Macroeconomics/

4 Matthew Shane, Real Historical Gross Domestic Product (GDP) and Growth Rates of

GDP. Economic Research Service (2010): Date accessed: October 2010 on

http://www.ers.usda.gov/Data/Macroeconomics/

33

for his Olympic study,5 while also being aided by the World Latitudes and Longitudes

online.6 All additional variables except technology were manipulated using the

information given from the sources mentioned.

Reviewing the numbers of TABLE 4.1, the average life span of a record broken

by individuals proves to be twenty days, which seems very low. Excessive zeros in the

data set may be the reason for the low average. Zeros and many low day count lapses

occur because multiple records are broken in one day or in the span of a week due to

large international events like the Olympic Games. Thus, a high standard deviation

makes sense as records break at an inconsistent frequency. The high GDP per capita

average means that more record breaking athletes come from richer nations, isolating

the poor nations as clear outliers. The average latitude also appears high relative to the

minimum number, meaning that more of these record breaking athletes live in colder

climates. Interestingly though, the average latitude is approximately the same latitude as

USA and the low deviation may signify USA as a leading nation in producing the most

record breaking athletes. On average, one technology is introduced for every record

broken that year, which makes the introduction of five technologies in one year a rare

occurrence.

5 Johnson, D. K. N. and A. Ali. “A Tale of Two Seasons: Participation and Medal Counts at the

Summer and Winter Olympic Games.” Social Science Quarterly 85, no. 4 (2004): 974-993

6 “World Latitude and Longitude.” MapXL. (2000): Date accessed: October 2010 on

http://www.mapsofworld.com/lat_long/

34

TABLE 4.1

SUMMARY STATISTICS FOR INDIVIDUAL RECORD BREAK DATA SET

Variable: Obs Mean Standard

Deviation

Minimum Maximum

Day Lapse

Since Last

Record

735 20.2449 49.12781 0 342

GDPpercapita 736 25031.12 13685.07 28.19356 248404.1

Latitude 736 42.50902 9.035581 11.82827 63.45496

Technologies

Introduced

736 1.110054 1.549231 0 5

The briefest data set includes the total number of records broken every year and

the corresponding number of technologies introduced as shown in TABLE 4.2. It is

interesting to observe how the total record counts correspond to the technologies

introduced that year or the year prior. For example, while records are shattered in 2008

and 2009, a total of eight new technologies are introduced over those two years with

two technologies introduced the year before.

35

TABLE 4.2

RECORDS BROKEN EACH YEAR COMPARED WITH TECHNOLOGY

YEAR TOTAL

RECORDS

WORLD

RECORDS

AMERICAN

RECORDS

TECHNOLOGIES

INTRODUCED

TECHNOLOGY

PROGRESSION 1969 26 15 11 0 0

1970 31 18 13 0 0

1971 30 18 12 0 0

1972 72 41 31 0 0

1973 38 28 10 2 2

1974 57 39 18 0 2

1975 48 25 23 1 3

1976 82 53 29 0 3

1977 17 11 6 0 3

1978 39 30 9 0 3

1979 24 16 8 1 4

1980 40 28 12 0 4

1981 19 12 7 0 4

1982 12 8 4 0 4

1983 27 14 13 0 4

1984 32 19 13 0 4

1985 15 11 4 0 4

1986 21 13 8 0 4

1987 36 20 16 0 4

1988 21 10 11 0 4

1989 14 8 6 0 4

1990 11 5 6 1 5

1991 23 15 8 0 5

1992 22 11 11 0 5

1993 3 2 1 1 6

1994 17 13 4 1 7

1995 9 6 3 0 7

1996 18 16 2 1 8

1997 8 5 3 1 9

1998 3 2 1 1 10

1999 20 15 5 0 10

2000 40 21 19 5 15

2001 20 13 7 1 16

2002 18 10 8 0 16

2003 35 18 17 3 19

2004 32 14 18 4 23

2005 20 12 8 2 25

2006 22 14 8 1 26

2007 29 15 14 2 28

2008 84 45 39 4 32

2009 75 47 28 4 36

36

The record break numbers were acquired from the previous data set by

combining the number of record breaks that occur every year by each swimmer

disregarding nationality. It also takes into account American record breaks and World

record breaks separately. TABLE 4.3 displays the summary statistics for this small data

set. On average, nearly thirty records are broken each year internationally with nearly

one new technology introduced. Due to the high standard deviation, total records broken

prove to be inconsistent at any given year, which may signify the impact of external

factors on the occurrence of record breaks.

TABLE 4.3

SUMMARY STATISTICS FOR TOTAL RECORD BREAK DATA SET

Variable: Obs Mean Standard

Deviation

Minimum Maximum

Total Records 41 29.5122 19.91497 3 84

American

Records

41 11.56098 8.561684 1 39

Technologies

Introduced

41 .8780488 1.326558 0 5

The last data set uses all previous sources in order to assign records broken,

GDP per capita, sea access, latitude, star athletes and average lifespan of a record that

year to each of the twenty five countries since 1969. The summary statistics for the

country data set are presented in TABLE 4.4. The mean for the average life span of

records each year is about thirty-four days, while the total records broken by each

country is approximately one. Obvious outliers exist on the maximum end with the 184

day average lifespan and 27 total records broken, resulting in high standard deviations.

Excessive zeros may be the reason for such a low total record average because every

37

nation is included at every year. Many nations only contribute a small amount of record

breaking athletes, thus because they are observed every year regardless of contribution,

a large amount of zeros arise. A clear outlier also occurs with the country providing

eight star athletes. On average, the number of star athletes a nation provides on any

given year is .2, which proves to be consistent due to its low standard deviation. This

makes sense as many countries fail to provide any star athletes at all, thus an excessive

amount of zeros occur. For instance, if a country never produces a star athlete, a zero

will show up in the data set forty-one times, a zero for every year. The remaining

variables follow similar trends as explained and shown in TABLE 4.1. Finally, the

technology variable which is tested in each data set must be properly explained.

TABLE 4.4

SUMMARY STATISTICS FOR COUNTRY DATA SET

Variable: Obs Mean Standard

Deviation

Minimum Maximum

Average

Record

Lifespan

1025 33.64098 33.85737 6.51 184

Star athletes 1025 .2390244 .7559154 0 8

Total Records 1025 .7180488 2.353836 0 27

GDPpercapita 1025 16145.97 13608.2 28.57657 52190.78

Latitude 1025 42.70709 13.13687 11.82827 63.45496

Technologies

Introduced

1025 .8770732 1.311201 0 5

Technology Factor

In order to determine the number of swimming technologies introduced every

year, information on the release of competitive swimsuits and fabrics were provided by

the top swimming company websites such as Speedo, Arena, Nike, Jaked, TYR, and

38

Adidas.7 Follow-up questions were asked via phone calls with individual company

workers. A chronology of the major introductions of swimsuits, fabrics and other

significant changes is displayed in TABLE 4.5. If a new fabric introduced was designed

for a specific swimsuit, then both technologies only count once in the total technology

column. Furthermore, the table includes the Beijing Water Cube in 2008 due to the

abundant records breaks broken in this specific pool. The pools significance lies in its

increased depth and wider lane lines relative to all other pools because it alleviates the

reverberations hindering a swimmer‟s pace.8

The table also includes other key technological moments. We discussed the

difficulty in defining technique changes, but when FINA publically approved the

dolphin kick in competitive swimming, I deemed it necessary to make an exception and

include it within the data due to its open release.9 We also must consider the

introduction of the starting block in 2007 as it has proven to shave off seconds to a

swimmers race.10

Accurate timing increases the frequency of records because the

smaller the unit of reporting, the smaller the margin by which a record must be broken

to be recognized especially in short distanced events. Finally, the data set includes the

introduction of the body suit concept as the idea itself sparked a flood of new

7 See sources for TABLE 4.5

8 “Venues.” Beijing 2008 Olympics. (2008): Date accessed: October 2010 on

http://en.beijing2008.cn

9 Howard Berkes. “Dolphin Kick Gives Swimmers Edge.” NPR. (2008): Date accessed:

February 3, 2011 on http://www.npr.org/templates/story/story.php?storyId=93575235

10

“USA Swimming.” Sport Science. Date accessed: October, 2010 on

http://www.usaswimming.org/DesktopDefault.aspx?TabId=1685&Alias=rainbow&Lang=en

39

innovations.11

After presenting how each factor is taken into account, definitions of all

variables in the data sets exist in TABLE 4.5.

11

Bernd Feldmann. “The Jet Concept.” Date accessed: October, 2010 on

http://www.swim.ee/index.html

40

TABLE 4.5**

CHRONOLOGY DESCRIPTION OF SWIMSUIT INNOVATIONS RANGING

FROM 1969-2009

YEAR* SWIMSUIT

INTRODUCED

NEW

FABRICS

OTHER TOTAL

TECHNOLOGY 1972 - Speedo

discovers

nylon/elastine

- 1

1973 Arena Skinfit - 1

1975 - Speedo

discover lycra

material

- 1

1979 Arena Flyback - - 1

1990 Arena Aqua Racer - - 1

1993 - Speedo S2000 1

1994 Speedo Endurance Speedo Four-

Way Stretch

- 1

1996 - Aquablade - 1

1997 Arena Xflat - - 1

1998 - - Adidas

introduces body

suit concept

1

2000 Speedo Fastskin, Diana

Submarine, Nike Lift, TYR

Aquapel, Arena Powerskin

- - 5

2001 - Speedo FS - 1

2003 Adidas Jet Concept, TYR

Aquashift, Nike Swift

- - 3

2004 XD Skin, Arena Powerskin

Xtreme, Speedo FastskinII,

Arena Powerskin X-treme

Speedo FSII - 4

2005 TYR Fusion - Fina allows

dolphin kick

2

2006 Speedo Fastskin Pro - - 1

2007 - Speedo

FSPRO

**FINA allows

creation of

starting block by

Omega

2

2008 TYR Tracer Light, TYR

Tracer Rise, Speedo LZR

Racer

Speedo LZR

Racer Pulse

Polyurethane

*Beijing Water

Cube increases

pool depth by 3

meters

4

2009 Arena Powerskin X-Glide,

Jaked 01, Adidas HydroFoil,

Arena Powerskin R-Evolution

- - 4

*Excluded all years without an introduction of a new swimming technology

**Sources to follow on next page

41

SOURCES: -“Jaked.” Press. (2010): Date accessed: October, 2010 on http://www.jaked.it. (Translated

Version)

-“Speedo History.” (2011): Date accessed: October, 2010 on

http://www.speedo.com/speedo_brand/insidespeedo/history/index.html,

-“TYR.” News/Events. (2010) Date accessed :October, 2010 on http://www.tyr.com/news/ -“Arena History.” 2010. Date accessed October, 2010 on

http://arenainternational.com/en/company/over-30-years-of-success

-Missing technology and confirmation were abstracted from personal phone calls to the

companies

TABLE 4.6

DESCRIPTION OF VARIABLES

Variable Name Description

Techintro The amount of swim suits, fabrics and other major innovations

introduced that year

GDPpercapita The GDP of the athletes nation divided by the nations population

Latitude The latitude of the athlete‟s home nation

Daylapse The number of days since the last swimming record was broken

Avgreclife The average day lapse between record breaks in that specific year

Percenttimelapse The difference between the new record performance time and the

former record performance time divided by the new record time.

Totalrecords The amount of total records broken that month

Americanrecords The amount of American records broken that month

GDP The GDP of the athletes countries in 2005 base million dollars

Population The total population of the athletes country

Techpro The progression of swimsuits, fabrics and other major

innovations since 1969

Starathletes The number of athletes competing for the country that year who

have broken 3 or more records

Year Time series of the data from 1969-2009

Dummy Variables

Seaaccess Given a “1” if the athlete home nation had access to the sea

Star3 Given a “1” if the athlete broke 3 or more records

Legend9 Given a “1” if the athlete broke 9 or more records

Neighbornation Given a “1” if the athlete broke the record in a neighboring nation

Hostnation Given a “1” if the athlete broke the record in his home nation

Currentrecholder Given a “1” if the athlete broke his/her own event record

Fourrecholder Given a “1” if the athlete broke this events record within the past

four swimmers

42

Variable Correlations

In order to properly account for each variable, pair-wise correlations must be

conducted in order to examine the relationship between each factor. If two variables are

perfectly linearly related (1.000), the model fails to differentiate any distinct effects

because they comprise the exact same information. Thus, correlations should be no

greater than the absolute value of 0.5 because the variables significance will be

inaccurate. Two variables with multi-collinearity will either deprive or inflate each

other of their true significance, resulting in artificially high or low t and f-statistics.

Table 4.7 shows the correlation of variables on a country basis, while TABLE 4.8

displays variables on an individual basis.

TABLE 4.7

MULTICOLLINEARITY OF VARIABLES ON A COUNTRY BASIS

Techintro Latitude GDPpercapita Starathletes Seaaccess

Techintro 1.0000 - - - -

Latitude 0.0009 1.0000 - - -

GDPpercapita 0.1574 0.3591* 1.0000 - -

Starathletes 0.0476 0.0009 0.1995 1.0000 -

Seaaccess -0.0004 0.0515 0.1592 0.1018 1.0000

43

TABLE 4.8

MULTICOLLINEARITY OF VARIABLES ON AN INDIVIDUAL BASIS

Techintro Latitude GDPpercapita Star3 Seaaccess Hostnation

Techintro 1.0000 - - - - -

Latitude -0.1634 1.0000 - - - -

GDPpercapita 0.3596* -0.0175 1.0000 - - -

Star3 -0.0309 -0.0022 0.0720 1.0000 - -

Seaaccess 0.0279 -0.0149 0.2340 0.0673 1.0000 -

Hostnation -0.0688 -0.0541 0.0316 0.0152 0.1625 1.0000

A moderately significant correlation exists between latitude and GDP per capita

in TABLE 4.7. However, when the model was tested without latitude, barely any

change occurred with GDP per capita, thus we can dismiss an issue of multi-

collinearity. Similarly, a correlation exists between techintro and GDPpercapita in

TABLE 4.8. Again, when testing the model without the technology variable the GDP

per capita value changed insignificantly.

Methodology

This section discusses the method used in order to observe the previously

presented data through different regression equations using a command-based program

called Stata. In order to test what influences the number of record breaks, a negative

binomial regression will be used because the model takes on any non-negative integer

value. Negative binomials are more often used for dispersed abnormal data because it

allows for more variability within the model. When testing the day lapse for the last

record break, we will use the poisson because the dependent variable also measures an

occurrence, does not follow a normal distribution, but the dependent variable is not as

44

dispersed at the day lapse variable.12

We avoid heteroskedasticity issues in all the

models by applying the robust error to each regression. This method automatically

changes the standard errors in order to guarantee constant variance. We also will test for

autocorrelation using the Durbin-Watson statistic. If any of the regressions contain

autocorrelation, the error variables are not independent of each other, which violate an

assumption regarding the data, thus questioning the legitimacy of the model. We do not

test for normality in any of the data sets because it does not curve around a central mean

due to the elimination of negative integers. We would only test for normality if our

dependent variable measures the change in world record counts. Even though the study

did examine how explanatory variables affect the change in record breaks, the T-

statistics resulted in insignificant values, thus we dismissed this definition as a valid

dependent variable. Thus, we use the total number of record breaks as our dependent

variable, resulting in a highly skewed data set where testing for normality would be

useless.

12 “Introduction to SAS.” UCLA: Academic Technology Services, Statistical Consulting

Group. Date Accessed: February, 2011 on http://www.ats.ucla.edu/stat/sas/notes2/

45

CHAPTER V

RESULTS AND ANALYSIS

This chapter provides the final results of the regression equations tested while

also providing an in-depth analysis of each equation. By reviewing the results of the

regression models, we attempt to better understand what exactly influences the

frequency and quality of breaks. A primary equation is presented first in order to

discover if technology has any significant effect on the occurrence of records breaks.

Two supplementary models will then follow hoping to further explain the primary

equation with additional data.

Results

In order to answer whether or not technology has an effect on record breaks, we

first run a simple negative binomial regression with total record breaks as the dependent

variable and number of swim technologies as the sole independent variable. Due to

problems with autocorrelation, we must include a second independent variable: the time

in years. Furthermore, we test the same independent variables, but with a different

dependent variable: number of American records broken. We also add a time variable

for this equation in order to fix autocorrelation issues. The final equations are displayed

below:

Nbreg totalrecords= f(techinto, year), robust

Nbreg Americanrecord= f(techintro, year), robust

46

By using year a second independent variable, we eliminate autocorrelation, but

also unfortunately create an issue with multi-collinearity. When the variables year and

techintro are highly correlated with each other (0.6088), the true coefficient values are

deflated. However, the purpose of this primary equation is just to show if technology

has any significance on record breaks, not how much. Thus, it makes more sense to

include year and avoid autocorrelation in order to examine the t-values accurately. The

results to these equations are displayed in TABLE 5.1.

TABLE 5.1

TECHNOLOGY IMPACT ON RECORD BREAKS ON AN ANNUAL BASIS

TOTAL RECORD

BREAKS

AMERICAN RECORD

BREAKS

TECHINTRO .345 (4.52)*** .396 (4.83)***

YEAR -.033 (-4.73)*** -.033 (-4.03)***

CONSTANT 68.124 (4.95)*** 67.266 (3.29)***

Wald Chi2(2) (28.76)*** (25.13)***

OBSERVATIONS 41 41

DURBIN WATSON 1.83 1.64

By reviewing TABLE 5.1., we obtain the Wald Chi2(2) values of 28.76 for total

records and 25.12 for American records with two degrees of freedom. Associated with

the chi-squared values are the p-values, both at 0.0000. The Wald Chi2(2) values assess

the “fit of the model” and are analogous to an F-statistic in a linear regression model. It

shows the significance of the explanatory variables jointly. The p-values are both less

than the generally used criterion of 0.05, thus we reject the null hypothesis, indicating

that the coefficients are not simultaneously equal to zero. From these results, we can

conclude that both techintro and year are statistically significant in determining record

47

break counts. However, due to issues with multi-collinearity, we cannot confirm how

much of record breaks are influenced by the introduction of swimming innovations.

The critical values for the Durbin-Watson statistic depend on the degrees of freedom

and number of observations. Given both these equations have two degrees of freedom

and forty one observations, the critical values lie at 1.391(dL) and 1.600 (dU) at the 5%

significance level.1 The Durbin-Watson test abides by the following rules:

d>dU= statistical evidence proves the error terms are not positively autocorrelated

d>dL= statistical evidence proves the error terms are positively autocorrelated

dL <d< dU=the test is inconclusive

With both Durbin-Watson statistics above the dU measure, we can conclude no

autocorrelation exists.

This model also draws data not only on an annual basis, but also on a monthly

basis in order to include more observations (492). However, the explanatory variables

do not correspond with the dependent variables because monthly record breaks cannot

be compared to the technology variable due to its limitation in defining introductions of

innovation only on an annual basis. When tested on a monthly basis however,

technology does prove to have an impact on record breaks, but the coefficient cannot be

compared with the models tested in TABLE 5.1 due to the divergence in variables.

Factors Influencing Lifespan of a Record

In order to better understand the frequency of record breaks in further detail, we

incorporate additional explanatory factors, which have proven in the past to have an

influence on athletic performance. In our next equations, we test the dependent variable,

day lapse of records, using poisson regression because we are measuring the count

1 Durbin-Watson Table. (2011): Date accessed: February, 2011 on

http://www.nd.edu/~wevans1/econ30331/Durbin_Watson_tables.pdf

48

between record occurrences. We first observe this dependent variable using the

individual record break data set. All variables cleared the multi-collinearity test as

displayed in TABLE 4.8, thus are all incorporated in the following equation:

Poisson daylapse= f(techintro, GDPpercapita, latitude, star3, seaaccess,

hostnation), robust

In order to test a similar dependent variable using the country data set, we use

the same variables except hostnation to test the average day lapse of a record that year.

We eliminate host nation because the location of an exact swimmer breaking a record is

lost at the country level. The following equation shows how we tested this similar

dependent variable, but with a different data set:

Poisson avgreclife= f(techintro, latitude, GDPpercapita, starathletes,

seaaccess) robust

Fortunately, these two equations escape the issue of autocorrelation, thus no

time variable is added. Results to these equations are displayed in TABLE 5.2.

TABLE 5.2

IMPACT ON THE DAY LAPSE OF RECORD BREAKS

INDIVIDUAL COUNTRY

TECHINTRO -.120 (-2.09)** -.081 (-5.98)***

LATITUDE -.023 (-2.06)** -.003 (-0.95)

STARS .094 (0.49) -.190 (-6.21)***

GDPpercapita -2.24x10-6

(-0.25) 6.65x10-06

(2.49)**

SEAACCESS -.115 (-0.22) .013 (0.11)

HOSTNATION .623 (3.28)*** ---

CONSTANT 3.811 (5.55)*** 3.610 (30.06)*** Wald Chi2(6)(5) (20.04)*** (80.11)***

Pseudo R2 0.0522 0.0305

OBSERVATIONS 735 1025

DURBIN WATSON 2.08 1.72

49

Based on the Wald-Chi2 values, the variables tested in the country data set are

more significant jointly in determining the lifespan of a record, but the combined

variables in the individual data set still proves to be significant. The p-values are both

less than 0.05, thus we can reject the null hypothesis, indicating that the coefficients are

not simultaneously equal to zero. The pseudo R-squared does explain the proportion of

the total variability of the outcome that is accounted for by the model like the typical

OLS R-squared; however, pseudo R-squareds cannot be interpreted independently or

compared across datasets. They can only be used in evaluating multiple models

predicting the same outcome on the same data set. A pseudo R-squared only has

meaning when compared to another pseudo R-squared with the same data and

predicting the same outcome. In this particular condition, the higher pseudo R-squared

indicates which model better predicts the outcome. Thus, the Pseudo- R-squareds are

irrelevant when analyzing these results.2

Given the individual data equation has six degrees of freedom and 735

observations, the critical values for the autocorrelation lie at 1.707 (dL) and 1.831 (dU)

at the 5% significance level. However, the Durbin-Watson statistic scored a 2.08, thus

no issues with autocorrelation exist. The country data equation scored a lower value of

1.72. Given its five degrees of freedom and higher number of observations, the critical

values lie at 1.623(dL) and 1.725 (dU) at the 1% significance level. Thus, we fail to

completely reject autocorrelation, but issues regarding this test are highly improbable.

By confirming the validity of the overall model, we can now scrutinize the z-

values and their corresponding coefficients. Both models prove how the introduction of

2 “Introduction to SAS.” UCLA: Academic Technology Services, Statistical Consulting Group.

Date Accessed: February, 2011 on http://www.ats.ucla.edu/stat/sas/notes2/

50

technology has a negative influence on the day lapse of breaks. Even though the

swimtech variable was more significance in the country data set, both models show how

an introduction of swimming technology decreases the day lapse of a record break. In

other words, new innovations in swimming increase the frequency or lifespan of a

record break. The country model shows that for every technology introduced, the

average lifespan of a record will decrease by .081 days. Similarly, the individual model

confirms the results of the country model by explaining how the average lifespan of a

record will decrease by .120 days with the presence of one new innovation. Overall, the

swimtech variable proves to have significance on the lifespan of a record especially

because both models derived from different data sets support this variable. Figure 5.1

shows how the higher number of swimsuit technologies not only maintains the

frequency of record breaks, but the average lifespan of a record also slightly decreases

on a country level with a linear fit on a scatter plot. For example, if this year introduces

four swimming technologies, we may expect the average lifespan of a record to be forty

days at most in year 2011. In other words, approximately nine additional records will

fall in year 2011 due solely to new developments in technology.

51

FIGURE 5.1

TECHNOLOGY IMPACT ON AVERAGE LIFESPAN OF A RECORD

For the remainder of the variables, the equations disagree on which factors have

an influence on the frequency of record breaks. However, this divergence in the models

do not make the significant factors any less valuable to the success in record breaks

because the models are abstracting information from two completely different data sets.

Within the individual model, latitude and hostnation appear to have an impact on the

occurrence of record breaks. As the latitude increases with each athlete, the day lapse

between each break decreases by .023 days. This outcome proves that athletes from

colder nations increase the frequency of record breaks. Unexpectedly though, if the

athlete breaks the record from his own home nation, the day lapse from the last break

will increase by .623 days. This number seems surprising as one would think athletes

would be breaking more records within their home nation because of reasons explained

in the literature review. The seaaccess and GDPpercapita variables, although negative,

05

01

001

502

00

0 1 2 3 4 5

Number of Swimsuit Technologies

AverageLifespanofrecord Fitted values

Average Lifespan of

a Record

52

have no effect on an individual breaking a record within a smaller amount of time lapse.

The constant explains that if no additional explanatory variables were used within the

equation, the day lapse between each break would be 3.811 days.

For the country model, the starathletes variable has a huge significance on the

average lifespan of a record break. For each country with a star athlete breaking a

record, the average lifespan of a record decreases by .190 days. Countries with star

athletes have a huge impact on influencing an increased occurrence of record breaks.

Furthermore, countries with a higher GDP per capita inversely prove to be decreasing

the average lifespan of a record break by a minuscule amount. A reason for this oddity

in the results could be that the countries used are only the ones who have had athletes

break swimming records. Thus, the data set excludes all other countries. If we assume

only countries with the highest GDP per capita are included based on previous literature

stating GDP per capita has the highest influence on athletic success, the model fails to

differentiate between the rich and poor countries. Thus, the outcome of the GDP per

capita variable fails to be reliable as the data only contains athletes from countries with

the highest GDP per capita relative to the average GDP per capita. GDP per capita may

be decreasing the average lifespan of record breaks because of countries like Australia

who are not necessarily the richest of countries, but as explained before, have spent

more money on swimming activities relative to richer countries because the popularity

of swimming within that nation.

Latitude and seaaccess variables have no effect on the frequency of record

breaks within the country model, which may be because most countries have higher

latitudes and are accessible to the sea. In fact, the average latitude for the countries

53

included is 42.70709 ºN, while the United States latitude falls just below that at

41.60879 ºN, thus countries included within the data are naturally located in colder

temperate areas. The constant coefficient corresponds similarly with the constant

variable in the individual model. Thus, if no additional explanatory variables were used

within the equation, the average lifespan of a record break would be 3.61 days.

Factors Influencing Total Record Breaks and Star Athletes

We then explore some final dependent variables using our largest country data

set. We first observe which variables contribute to the total record breaks on a country

level using the negative binomial regression with the results displayed in TABLE 5.3:

Nbreg totalrecords= f(techintro, latitude, starathletes, GDPpercapita,

seaaccess) robust

TABLE 5.3

FACTORS INFLUENCING TOTAL RECORD BREAKS FOR EACH COUNTRY

TOTAL RECORD BREAKS

TECHINTRO .164 (3.22)***

LATITUDE 7.991x10-3

(-0.13)

STARS 1.563 (11.50)***

GDPpercapita 2.23x10-06

(0.35)

SEAACCESS .901 (3.59)***

CONSTANT -2.719 (-8.32)***

Wald Chi2(5) (205.39)***

# OF OBSERVATIONS 1025

DURBIN WATSON 1.99

54

With a high Wald Chi2(5) and a low p-value, this model proves valid in

determining the factors influencing total record break. Furthermore, with five degrees of

freedom and 1025 observations, the critical values for the Durbin-Watson statistic lie at

1.718 (dL) and 1.820 (dU) at the 5% significance level. Since the actual value (1.99)

exceeds the upper limit, we can dismiss any bias due to autocorrelation.

The techintro variable appears to have a significant influence on total record

breaks in the world. For every introduction of technological innovation, the total record

breaks broken by each country increases by .164. The number of star athletes generated

from each country also shows to have a huge impact on an individual countries total

record break count as shown in FIGURE 5.2. For every one star athlete a country

generates every year, the total record break count for that country increases by 1.563.

This variable seems reasonable as data shows the same athlete breaking records

multiple times like USA stars Mark Spitz and Shane Gould in 1972 or Kornelia Ender

(GER) in 1976 and even Michael Phelps (USA) in the most recent years. FIGURE 5.2

shows how much star athletes have an impact on the number of World records at the

country level. For example, the existence of eight star athletes competing in 2011 would

result in nearly nineteen additional record breaks for that year.

55

FIGURE 5.2

STAR ATHLETES IMPACT ON RECORD BREAKS FOR EACH COUNTRY

A final significant variable proves to be seaaccess. This model proves that for

every country with direct accessibility to the sea, their total record break count increases

by .901. GDPpercapita and the latitude of a country both show no impact on the total

records broken that year, which again may be because the exclusion of all countries.

Finally, if the model excluded all explanatory variables within the equation, the total

record breaks would be -2.719 a year. Since a negative record break count would be

impossible, the explanatory factors included prove to be especially vital in determining

the total amount of record break.

All previous models tested prove the introduction of technology takes a

significant role in determining an increase in record break occurrences. Other variables

like latitude, seaaccess, and starathletes prove to be significant on more erratic terms.

56

One aspect of the model that remains in question is GDPpercapita. Not only does this

variable reveal itself as insignificant for the majority of times, but it also shows to

hinder athletic record break shown in TABLE 5.2 for countries. This outcome seems

daunting as most studies reviewed in chapter II prove that GDP per capita is the single

most important factor in determining athletic success. Furthermore, we suspect other

variables such as availability, coaching, training, and sponsorships would be higher in a

richer country, thus we expect this variable to capture some of these influences with

record breaks, but it is not the case. Due to this insignificance of what we know is an

essential factor in determining athletic performance, we suspect another variable may be

robbing the actual value of GDPpercapita. Since we now know star athletes are held

accountable for a mass number of records broken, we test what factors determine a star

athlete in this negative binomial regression with the results shown in TABLE 5.4:

Nbreg starathletes= f(seaaccess, latitude, GDPpercapita, techintro) robust

TABLE 5.4

FACTORS INFLUENCING STAR ATHLETES FOR EACH COUNTRY

STAR ATHLETES

TECHINTRO .042 (0.30)

LATITUDE -.007 (-0.64)

GDPpercapita 5.68x10-5

(7.79)***

SEAACCESS .801 (1.05)

CONSTANT -3.070 (-6.15)***

Wald Chi2(4) (122.22)***

# OF OBSERVATIONS 1025

DURBIN WATSON 0.717

57

As assumed, this model shows evidence that GDPpercapita of a country has a

strong influence in generating star athletes, but due to issues with autocorrelation, the

coefficient cannot be validated. We attempted to fix autocorrelation in several ways by

adding a time series or changing the dependent variable to the change in star athletes

over time, but these modifications failed to adjust the Durbin-Watson statistic.

Regardless, these results indicate that the starathletes variable does capture most of the

GDPpercapita influence, thus GDP per capita does indirectly have an effect on the

incidence of record breaks.

Additional Consideration

After reviewing the results of all models, additional reflection needs to be

discussed in order to thoroughly interpret the data. Although the introduction of

technology proves to be significant with record breaks in swimming, the effect is shown

to be minimal. However, we defined technology in broad terms and made many

assumptions about its use within the swimming world as discussed in chapter IV. Thus,

the fact that the models even show any value for innovation shows that if defined more

accurately within the data set, technology could have an even bigger impact.

We also tested many other independent and dependent variables, but since they

resulted in less significant values to the ones presented, the study eliminated them from

the final models. GDP and population were tested separately at first, but due to issues

with multi-collinearity, we combined the two variables to obtain GDP per capita.

However, when they were tested separately, GDP proved to be more significant in

58

determining swimming record breaks than population. Thus, including population may

be another cause for the non-impact of GDP per capita within the models.

Other variables were tested as dummies: currentrecholder and fourrecholder. To

review, a “1” was given to a certain athlete if they broke their own record within that

event or if they broke that events record within the last four athletes to break that

event‟s record time. Both variables were correlated, but when used independently, they

did prove to be significant in determining record breaks. However, the models replaced

these variables with the star3 and legend9 variables because athletes can break records

in multiple events. Michael Phelps may break nine records in one day, but all in

different events. Thus, the previous variables would not be able to account for his

natural experience and athleticism. The star3 variable takes this sweep into account by

adding all of his records ever broken. The model also drops the legend9 variable due to

its correlation with star3. We chose to present the star3 variable due to its higher

influence on total record breaks.

In order to take into account technology as a constant development for

swimming performance, we attempted to test the progression of swimming innovations

as seen in TABLE 4.2. As time passes, new technologies introduced constantly build off

the most recent innovation in order to keep improving. For example, the Speedo LZR

racer may have not been invented if Adidas did not introduce the idea of bodysuits in

1998.3 In order to account for the consistent growth of technology, we added the

Techpro variable, which considers every past innovation since 1969 into the total

technology count of the current year. By obtaining this variable, the study hoped to

apply current technology to swimmers performance times in the earlier years in order to

3 See TABLE 4.5

59

fairly compare current swimmers with past swimmers of equal technology opportunity.

However, this progression of technology showed to be barely significant in determining

record breaks. We even attempted to better match this explanatory variable with the

progression of record breaks, but the value was still insignificant, thus the variable was

dropped from the model.

The seaaccess variable was added last in an attempt to proxy for the popularity

of swimming within the nation. This variable surprisingly demonstrated significance in

TABLE 5.3 when influencing the total record breaks. However, this variable could have

been more significant if the country data set included more countries because most of

the twenty-five countries bordered a sea. In fact, only five countries are completely

land-locked: Hungary, Poland, Serbia, Switzerland, and Zimbabwe. In order to improve

this variable, we could have measured the percentage of the countries border that

included a coastline. However, due to time and absence of public information, we are

unable to take this into account in the models.

We also attempted to examine another dependent variable: percenttimelapse.4

This aspect of record breaks captures how sizeable each record break was. For example,

a record broken by .0001 seconds differs in significance than a record broken by 4.000

seconds. The specific event also needs to be considered when analyzing the quality of a

record break. Breaking the 100LCM freestyle by .0001 second may be the same

proportionally as breaking the 1500LCM by 4.000 seconds. Thus, when measuring the

value of a particular record break, we calculated the percentage of the time difference in

order to account for the specific event. We then tested all the explanatory variables on

percenttimelapse in a tobit model in order to set a lower and upper bound at 0 and 1.

4 See TABLE 4.6

60

However, no relationship existed between the dependent variable and techintro, perhaps

because this explanatory variable measures the existence of an innovation rather than

the “worth.”

Implications of Results

This study shows how an increase in record breaks can partly be explained by an

introduction of new swimming technology, sea access, star athletes, and latitude.

Naturally, these models only represent a small portion of what truly affects record

breaks, as many other factors must be considered, but are nearly impossible to account

for as discussed in Chapter III

As swimming technology continues to grow, swimming records should keep

falling. The pressure of introducing new innovations and shaping better athletes should

increase for major swim companies and wealthy countries as the future of swimming

depends on it. However, John Brenkus believes a specific “perfection point” exists, thus

at some point these variables will have no significance on the number of record breaks

because the dependent variable will have hit its maximum point. A limit must exist for

performance times because no swimmer can complete a lap in zero seconds. As record

times near our maximum potential, will the proven effect of a new technological

innovation slowly diminish in value?

It will be interesting to see how FINA continues to regulate innovations within

competition. They have already begun releasing annual lists of approved swimsuits to

solve the bodysuit controversy.5 FINA struggles to balance its two obligations of

providing the highest performance possible for the world to enjoy, while also preserving

5 “FINA Approved Swimsuits.” (2011): Date accessed: February, 2011 on www.fina.org

61

the purity of the sport for the sake of its main intentions. Technology plays a major part

in the future of swimming especially in a world so dependent and fixated with technical

development. When does innovation cross the line of appropriateness in sport? Should

Oscar Pistorius be allowed to compete in the 2012 London Olympic with two prosthetic

limbs?6

Further Research

This study leaves many other possibilities for further research. The technology

factor will continue to grow as more athletes depend on its availability in order to break

records. Further improvements may include including more countries or athletes in

order to add depth and show the true significance of what factors influence a record

break. Since the data only comprises the frontier of swimming since 1969, we only ask

what stimulates record breaks, while we could strengthen the argument by asking how

other countries and athletes fail to ever break records.

An improved and more advanced definition for technology could also improve

the results of this study. Not only does our technology variable take on many

assumptions and limitations, but it fails to show how much of an athlete‟s success is

credited towards technology. Thus, factoring in the quality of an innovation is crucial in

determining the eminence of a particular record break. Thus, I suggest applying Haake‟s

proven performance index in order to measure a technologies impact on the quality of a

record break for future research.

6 Majendie, Matt. “For Oscar „Blade Runner‟ Pistorius, Fast is a State of Being.” (2011).

The National. Date Accessed: March, 2011 on http://www.thenational.ae/lifestyle/motoring/for-oscar-

blade-runner-pistorius-fast-is-a-state-of-being

62

Analyzing a different sport with the same methodology may also be interesting

for comparison. For example, Haake used four sports in his study in order to show how

his approach with technology differed for each event.7 Analyzing cycling under the

same methodology with technology may be interesting in order to reaffirm or enhance

the results.

Finally, the study hoped to be able to compare swimmers across time in order to

justify the times of swimmers who competed without equal technical enhancements in

1969. However, we only test the introduction of technologies, not the progression, thus

we cannot apply the technology factor to performance times. For later studies, some sort

of technologic progression index must be used in order to analysis performance times

across time and compare athletes‟ natural abilities regardless of external factors.

7 S.J Haake. “The Impact of Technology on Sporting Performance in Olympic Sports.” Journal

of Sports Sciences 27, no. 13 (2009): 1421-1431

63

CHAPTER VI

CONCLUSION

Record breaks represent the ultimate success in athletics as they continue to

shatter barriers deemed impossible by society. Record breaks create a dynamic

interaction among fans, athletes, and media in order to maintain the thrill of a sport.

Thus, this thesis has examined the factors influencing record breaks as they influence

the future for most athletic events. Several variables were tested in order to discover

what affects the frequency of breaks. This study hoped to discover the relationship

between technology and record breaks. The high controversy of body suits and

dominating power of technology in the new century has made this research a point of

interest to those familiar with the athletic world.

Chapter I provides an introduction of the swimming world and why record

breaks occur most frequently compared to other record-breaking sports. The

background section of this chapter offers a history of the sport and how technology has

affected its progression throughout time. Finally, this chapter not only explains why this

subject is relevant to current times, but also why the study is essential to review for

future consideration.

Chapter II reviews past literature written regarding what affects success in

athletics. Analysis of past literature is crucial in understanding what research has

already been conducted and how we can possibly build off what is already discovered.

64

This section also shows how this study adds to the current range of literature. After

evaluating the relevant studies, I was able to gather a number of variables that have

proved to be significant in athletic success and utilize them differently within the

models.

Chapter III reviews the theory of this study and how it differs from previous

studies. This section discusses what variables will be used and which variables should,

but due to lack of information, cannot be used in the models. This chapter serves to

distinguish this study as an original piece, weaving both personal ideas and concepts

established by past authors.

Chapter IV then explains the data and methodology used in order to gain insight

regarding the studies focus. This section provides reasons for the disclosed data

selection and the sources used in order to obtain the data discussed. The chapter then

transitions to clarify why certain regressions and econometric tests are used in order to

ensure the validity of the models.

Chapter V presents the results and an analysis of the models tested using the

data in the previous chapter. All models used the negative binomial or poisson

regressions depending on the dependent variable. The results of this chapter show that

the introduction of technology does increase the number of record breaks in all models.

Furthermore, sea access, star athletes, and latitude prove to have an impact on the

number of record breaks. Evidence shows how GDP per capita creates star athletes, thus

this variable indirectly causes an increase to record breaks. These findings support the

main purpose of the study which was to determine the influences on record breaks.

65

Analysis of data has discovered a strong relationship between technology and

record breaks by measuring the size and strength. Generally, the introduction of one

new technology would result in .345 new broken records. An innovation also proves to

decrease the life span of a record by .12 days for individual swimmers and .081 days for

a country. Furthermore, for every one star athlete competing, 1.5 new records will be

broken.

This study has attempted to explain why athletes continue to break records.

Research has found that the introduction of technology has a direct impact on the

number of records broken that year. It was also found that star athletes, countries

accessible to sea, and individuals belonging to colder nations have the potential to

increase the number of record breaks. For the most part, results hint that the future of

swimming could be dominated by the effects of technology, especially since the study

loosely defines this variable. Swimming should expect either a slowdown in record

breaks if FINA bans all future technology or a consistent trend if society continues to

introduce new innovations to the sport.

66

SOURCES CONSULTED

“Arena.” Our History. (2009): Date accessed: October, 2010 on

http://arenaswimming.co.za/about/our-history/

Ball, Donald W. “Olympic Games Competition: Structural Correlates of National

Success.” International Journal of Comparative Sociology 12 (1972): 186-200.

Battese, G. E. and Coelli, T. J. “A Model for Technical Inefficiency

Effects in a Stochastic Frontier Production Function for Panel Data.” Empirical

Economics 20 (1995): 325–332.

Barzel, Toram. “The Rate of Technical Progress: The „Indianapolis 500.‟” Journal of

Econonmics Theory 4 (1972): 72-81

Bernard, Andrew B. and Meghan R. Busse “Who Wins the Olympics: Economic

Resources and Medal Totals.” Review of Economics and Statistics 86 (2004):

413–417.

Brenkus, John. The Perfection Point. New York. HarperCollins Publishers. 2010: 54-

75

Chandler, K. N. “The Distribution and Frequency of Record Values.” Journal of Royal

Statistical Sociology Series B, 14, no. 2 (1952): 220-28.

Clarey, Christopher. “Sydney 2000: Swimming; Beginning Tomorrow, All Eyes Will

Be Focused on One Pool.” Sydney. The New York Times. (2000): Date

accessed: January, 2011 on

http://select.nytimes.com/gst/abstract.html?res=F20D1FF7385C0C768DDDA00

894D8404482

Connors, Martin, Diane Dupuis and Brad Morgan. The Olympic Factbook. Detroit.

Visible Ink Press. 1992.

Grimes, Ray A., William J. Kelly and Paul H. Rubin. “A Socioeconomic Model

of National Olympic Performance.” Social Science Quarterly 55 (1974): 777–

782

Grubb, H.J. Models for Comparing Athletic Performances (Working paper).

Department of Applied Statistics. The University of Reading (1997)

67

Haake, S.J. “The Impact of Technology on Sporting Performance in Olympic Sports.”

Journal of Sports Sciences 27, no. 13 (2009): 1421-1431

Hill, Michael. “Swimmers „See‟ the Water with Help of Professor.” (2008): Date

accessed: October, 2010 on http://swimming.teamusa.org/news/2008

“Introduction to SAS.” UCLA: Academic Technology Services, Statistical Consulting

Group. Date Accessed: February, 2011 on

http://www.ats.ucla.edu/stat/sas/notes2/

Johnson, Daniel K. N. and Ayfer Ali. “A Tale of Two Seasons: Participation and Medal

Counts at the Summer and Winter Olympic Games.” Social Science Quarterly

85, no. 4 (2004): 974-993

Johnson, Daniel K. N and Ayer Ali, Coming to Play or Coming to Win: Participation

and Success at the Olympic Games (Wellesley College Working Paper) (2000):

2000-10

Johnson, Daniel. “Vancouver 2010 Medal Predictions.” (2009). Date Accessed:

October, 2010. http://theolympianblog.com/2010/02/vancouver-2010-medal-

predictions-daniel-johnson.html

Kolata, Gina and Jere Longman “As Swimming Records Fall, Technology

Muddies the Water.” The New York Times. (2008): A1.

http://www.nytimes.com/2008/08/12/sports/olympics/12records.html

Kortum, Samuel. “Research, Patenting, and Technological Change.”

Econometrica 75 (1997): 1389-1419

Kuper, Gerard H. and Elmer Sterken. The Olympic Winter Games: Participation and

Performance (Working paper). Department of Economics. University of

Groningen. (2001)

Kuper, Gerard H. and Elmer Sterken. “Endurance in Speed Skating: The Development

of World Records.” European Journal of Operational Research 148, no. 2

(2004): 293-301

Kuper, Gerard H. and Elmer Sterken. Modeling The Development of World Records in

Running (Working Paper). Department of Economics. University of Groningen.

(2006)

Levine, Ned. “Why Do Countries Win Olympic Medals? Some Structural Correlates

of Olympic Games Success: 1972.” Sociology and Social Research 58 (1974):

353-360.

68

Loland, S. “Technology in Sport: Three Ideal-Typical Views and Their

Implications.” European Journal of Sport Science 2, no. 1 (2001)

Loving, Sharon. Swimming Technology Research. Marketing Director. Tallahassee, FL.

(2010): Date accessed: October, 2010 on www.swimmingtechnology.com

Magdalinski, Tara. Sport, Technology and the Body: The Nature of Performance.

London, New York. Routledge. 2009.

Maennig, W. “On the Economics of Doping and Corruption in International Sports.”

Journal of Sports Economics 3, no. 1 (2002): 61-89

Magnusson, Christer. “Records.” (2010): Date accessed: October, 2010 on

http://www.scmsom.se/records/statistics/World%20Record%20Progression.htm

Majendie, Matt. “For Oscar „Blade Runner‟ Pistorius, Fast is a State of Being.” (2011).

The National. Date Accessed: March, 2011 on

http://www.thenational.ae/lifestyle/motoring/for-oscar-blade-runner-pistorius-

fast-is-a-state-of-being

Moosa, I. A. and L. Smith. “Economic Development Indicators as Determinants

of Medal Winning at the Sydney Olympics: An Extreme Bound

Analysis.” Australian Economic Papers 43, no. 3 (2004): 288–301.

Munasinghe, Lalith, Brendan O‟Flaherty and Stephan Danninger. “Globalization and

the Rate of Technological Progress: What Track and Field Records Show.”

Journal of Political Economy 109, no. 5 (2001): 1132-1149

“Name That Tune.” The Tuscaloosa News. (1976): Date Accessed: February, 2011 on

http://news.google.com/newspapers?id=rUYgAAAAIBAJ&sjid=650EAAAAIB

AJ&pg=6816,74269&dq=history+of+germany+competitive+swimming&hl=en

O‟Neil, Charles B. and Peter G. Akintunde. Optimizing Athletes’ Performance

Excellence and Wellness through Nutrition and Exercise. (Working Paper).

University of Calabar (2010)

Pfau, Wade D. “Predicting the Medal Wins By Country at the 2006 Winter Olympic

Games: An Econometrics Approach.” The Korean Economic Review

22, no. 2 (2006): 233-247

Pyne, David B., Cassie B. Trewin and William G. Hopkins. “Progression and

Variability of Competitive Performance of Olympic Swimmers.” Journal of

Sports Science 22, no.7 (2004): 613-620

Rathke, Alexander and Ulrich Woiteck. Economics and Olympics: An Efficiency

Analysis (Working Paper No. 313.) University of Zurich (2007)

69

Roan, Shari. “Have Olympic Athletes Done all They Can?” The Los Angeles Times

col. 1 (2010): Date Accessed: September, 2010 on www.latimes.com

“Roma 2009.” (2009): Date accessed: February, 2011 on fina.org

Saslow, Eli. “New Suit Makes Splash in Debut.” Washington Post Staff Writer. The

Washington Post. (2007): Date Accessed: October, 2010 on

www.washingtonpost.com

Schauer, Margaret. “How Much is Too Much? Evaluating the Impact of Swimming

Innovations” USMS Swimmer (2006): Date Accessed: October, 2010 on

http://www.usmsswimmer.com/200611/innovations.pdf

Shane, Mathew. Historical Population and Growth Rates in Population for Baseline

Countries/Regions. Economic Research Service. (2010): Date Accessed:

October, 2010 on http://www.ers.usda.gov/Data/Macroeconomics/

Shane, Mathew. Real Historical Gross Domestic Product (GDP) and Growth Rates of

GDP. Economic Research Service. (2010): Date Accessed: October, 2010 on

http://www.ers.usda.gov/Data/Macroeconomics/

“Speedo USA.” Speedo History. (2010): Date accessed: October, 2010 on

http://explore.speedousa.com

“Swimming Australia Annual Report.” Finance Report. (2008): Date accessed:

February, 2011 on at http://swimming.org.au/

Tiozzo, E., G. Leko and L. Ruzic. “Swimming Bodysuit in All-Out and Constant-Pace

Trials.” Biology of Sport 26, no. 2 (2009): 149-156

Tucker, Ross “Swimmers Credibility Crisis: How FINAS‟s Blind Eye is Affecting the

Purity of the Sport.” The Science of Sport (2008): Date Accessed: October, 2010

on http://www.sportsscientists.com/2008/11/swimsuit-controversy.html

“TYR Sport Inc.” (2010): Date accessed: October, 2010 on www.tyr.com

“United States Swimming, Inc.” Financial Statements and Supplemental Schedules.

(2008): Date accessed: February 2011 on www.usaswimming.org

“USA Swimming.” In Depth Event History. (2010): Date accessed: October, 2010 on

http://www.usaswimming.org/DesktopDefault.aspx?TabId=1476&Alias=Rainbo

w&Lang=en

Quarrell, Anne. “Adidas presents new bodysuit: the JETCONCEPT.” Adidas

America. (2003): Date accessed: October, 2010 on

http://www.eurekalert.org/pub_releases/2003-07/aa-apn071803.php

70

“World Latitude and Longitude. MapXL.” (2000): Date accessed: October, 2010 on

http://www.mapsofworld.com/lat_long/