Content uploaded by Dae Hee Kwak
Author content
All content in this area was uploaded by Dae Hee Kwak on Jan 14, 2015
Content may be subject to copyright.
ORIGINAL PAPER
The Overestimation Phenomenon in a Skill-Based
Gaming Context: The Case of March Madness Pools
Dae Hee Kwak
ÓSpringer Science+Business Media New York 2015
Abstract Over 100 million people are estimated to take part in the NCAA Men’s Bas-
ketball Tournament Championship bracket contests. However, relatively little is known
about consumer behavior in skill-based gaming situations (e.g., sports betting). In two
studies, we investigated the overestimation phenomenon in the ‘‘March Madness’’ context.
In Study 1 (N =81), we found that individuals who were allowed to make their own
predictions were significantly more optimistic about their performance than individuals
who did not make their own selections. In Study 2 (N =197), all subjects participated in a
mock competitive bracket pool. In line with the illusion of control theory, results showed
that higher self-ratings of probability of winning significantly increased maximum will-
ingness to wager but did not improve actual performance. Lastly, perceptions of high
probability of winning significantly contributed to consumers’ enjoyment and willingness
to participate in a bracket pool in the future.
Keywords Uncertainty Skill-based gaming Illusion of control Enjoyment
Risk-taking
Introduction
Making predictions a priori about as many game winners as possible in the National
Collegiate Athletic Association’s (NCAA) Men’s Basketball Tournament (a.k.a. March
Madness) has become one of the most popular betting activities in homes, offices, and
schools in the United States (McCrea and Hirt 2009). The March Madness tournament is
the most widespread wagered-on sporting event in the U.S. (Linn 2013). Over 100 million
people are estimated to take part in tournament bracket contests, which accounts for one-
third of the U.S. population (Jessop 2014). In addition, more than $12 billion is wagered in
D. H. Kwak (&)
University of Michigan, 1402 Washington Heights #2118, Ann Arbor, MI 48109, USA
e-mail: kwakd@umich.edu
123
J Gambl Stud
DOI 10.1007/s10899-015-9520-7
legal and illegal ‘‘pools’’ on the tournament (McCarthy 2012), suggesting that the March
Madness tournament exceeds the $10 billion amount bet on the National Football League’s
Super Bowl, considered the biggest single betting day in sports (Jessop 2014). Illegal
betting on the men’s college basketball tournament is so widespread that only one percent
of the amount wagered on the tournament comes from the state of Nevada (Jessop 2014),
one of four states in the U.S. where sports gambling is legal. This further illustrates the
sheer magnitude of the number of people engaging in collegiate sports wagers during
March. Given the popularity of the tournament, numbers of advertisers are keying on it to
reach a diverse audience. According to Ad Week (Crupi 2011), March Madness is the
second biggest postseason sports showcase for advertisers, surpassing the Major League
Baseball playoffs and World Series, the National Basketball Association playoffs and
Championship Series, and all 35 college football bowl games. Now that all games are
available live on mobile devices and personal computers, marketers are jumping on the
‘‘bracket’’ bandwagon to tie in their brands with the excitement of making predictions
about the tournament.
In a typical March Madness pool, consumers predict the outcome of all 63 tournament
games and gain points based on the number of correct predictions. More sophisticated
pools award higher points for correct predictions in later rounds of the tournament. Prizes
(e.g., entry fees from participants) are then usually awarded to the player with the highest
point total. While the tournament draws millions of American consumers to place bets, it is
almost impossible to make perfect predictions for 68 games being played during the three-
week tournament. A person with modest knowledge about team rakings and other infor-
mation has about a one in 128 billion chance of correctly picking each of the 63 games in
the tournament (Linn 2013). This chance is far below that of purchasing a winning lottery
ticket. Even the highest scoring bracket among ESPN.com subscribers has still gotten 18
games wrong, suggesting that it is extremely difficult to make accurate predictions. If
making accurate predictions is notoriously difficult, then, what makes March Madness
pools so popular? More importantly, what drives people to overestimate their control over
the outcome when the known statistical probability of making all predictions correctly is
close to zero? Does so-called skill really matter in making more accurate predictions? As
such, consumer behavior in uncertain contexts remains underexplored in the marketing
literature (Lam 2007; Sierra and Hyman 2009). The present study attempts to find answers
to these questions.
One could argue that the allure of the tournament is that it is very difficult so that
anyone without basketball knowledge can have a chance to outperform knowledgeable or
so called ‘‘skilled’’ participants. Therefore, we contend that it is not the actual statistical
probability that matters, but individuals’ perceived probability of winning that does. Pre-
vious research has documented that individuals with higher perceived control tend to invest
more time and money to achieve the desired outcome (e.g., Chau and Phillips 1995; Kwak
et al. 2010). Illusion of control theory (Langer 1975) suggests that skill-relevant factors
(e.g., involvement, choice, prior knowledge, competition, etc.) increase individuals’ con-
trol beliefs (see Thompson et al. 1998, for a review). In March Madness pools, personal
selections (i.e., filling out a bracket) coupled with uncertainty of outcome might increase
perceptions of control and increase individuals’ beliefs that they have a better chance of
making correct predictions. While previous research has considered illusion of control as
erroneous cognition that is frequently associated with problem or pathological gamblers
(e.g., Ladouceur 2004; Moore and Ohtsuka 1999), research has been equivocal about
whether such inflated control beliefs indeed result in better performance (e.g., Andersson
et al. 2005; Cantinotti et al. 2004). In particular, research has been sparse in examining the
J Gambl Stud
123
overestimation phenomenon among casual or social gamblers in a skill-based gaming
context.
To elucidate the overestimation phenomenon in a skill-based gaming context, we report
on empirical findings regarding how consumers develop inflated winning expectancy in
predicting the outcomes of the NCAA men’s basketball tournament brackets. In addition,
we examine whether enhanced winning confidence increases financial risk-taking and
actual performance. Finally, we explore the role of illusion of control in predicting
enjoyment and willingness to participating in a future tournament bracket.
Background and Review
Overestimation of Control in Chance-Based Events
People often overestimate their control over the desired outcome in chance-based situa-
tions. This is known as the illusion of control, which is defined as ‘‘an expectancy of a
personal success probability inappropriately higher than the objective probability would
warrant’’ (Langer 1975; p. 313). The evidence of illusory control has been extensively
documented in various chance-based gaming contexts where outcomes are unrelated to the
actions of the participants—such as lotteries (e.g., Ladouceur and Se
´vigny 2005; Langer
1975; Rogers 1998), slot machines (e.g., Griffiths 1995), and dice-throwing (e.g., Dunn and
Wilson 1990).
The notion of illusion of control has also been explored in other gaming contexts
believed to be associated more with participants’ skill/knowledge than chance, including
horse-racing (e.g., Allcock 1987; Ladouceur et al. 1998), sports betting (e.g., Cantinotti
et al. 2004), and fantasy sports (e.g., Kwak et al. 2010,2013). Research has consistently
shown that illusion of control is closely related to financial risk-taking (e.g., Chau and
Phillips 1995; Kwak et al. 2010), gaming frequency (e.g., Kwak et al. 2010), and gambling
behaviors (e.g., Kwak et al. 2013; Moore and Ohtsuka 1999). Given the widespread
overestimation phenomena in various gambling situations, researchers have explored
antecedents that influence illusion of control. Studies have found that people develop
illusory control perceptions when skill-relevant factors (e.g., familiarity, knowledge,
choice, etc.) are presented with the task (Thompson et al. 1998). For instance, if individuals
are more familiar with a task (via practice or simply imagining the task), they develop
illusions of control (Ayeroff and Abelson 1976; Dykstra and Dollinger 1990).
Choice is another skill-related factor that increases illusion of control. When partici-
pants were allowed to make their own choices such as selecting lottery numbers, as
opposed to being given numbers randomly by the experimenter, they believed their per-
sonal choices had a better chance of winning (Langer 1975). In the March Madness
tournament bracket context, completing one’s own bracket would amplify the belief that
one’s prediction is more accurate than picks made by other contestants. Why would
making personal selections increase optimistic biases regarding future events? The desire
to control events in one’s life seems to be closely tied to this erroneous cognitive pro-
cessing. Psychologists have proposed theories around a notion of ‘‘control motivation.’’
Similarly, DeCharms (1968) asserted that the individual’s primary motivational disposition
is to become a causal agent for his or her environment. In addition, Kelley (1971) con-
tended that cognitive biases in our expectations are in part due to individuals’ strong desire
for control. Likewise, individuals possess such innate desire for control and attribute
themselves as causal agents of desirable outcomes. In an empirical study, Burger and
J Gambl Stud
123
Cooper (1979) found that desire for control is positively associated with illusion of control.
Langer (1975) also found that people overestimate their control over desirable outcomes
when they are personally involved in the selection process. For instance, when people
select lottery numbers on their own, they tend to overestimate the value of their selection
over randomly selected numbers, while the probability of winning a lottery ticket remains
unchanged. Likewise, we posit that the act of the selection process is what increases
individuals’ estimation of their chances in uncertain contexts. Consistent with the existing
literature, we expect that individuals filling out their own brackets would likely to be more
optimistic about their performance than those who did not make their own selections.
In addition, previous research suggests that knowledge or experience increases illusion
of control (e.g., Cantinotti et al. 2004; Kwak et al. 2010; Langer 1975). For instance,
Cantinotti et al. found that experts’ skills do not translate into better financial gains than
random selection, concluding that experts’ skills are cognitive distortions that reinforce
illusion of control. Considering the wealth of statistical information and analysis available
to consumers, sports betting provides a unique opportunity to explore the role of knowl-
edge/skill in the development of illusion of control. It seems plausible that individuals’
perceived knowledge about sport would build confidence in their predictions. Kwak et al.
(2010) found that perceived knowledge about sport is a significant driver of confidence
among fantasy sports participants. However, one important question that remains to be
answered is whether confidence about winning leads to better performance in reality. If
confidence actually leads to better performance, the notion of ‘illusion’ should be avoided
in discussing the phenomenon.
Perceived Skill and Accuracy in Predictions
In the gambling literature, there remain mixed findings about whether participants’
experience or so called ‘‘skills’’ actually produce better performance in sports betting. One
could argue that knowledge or experience indeed help gamblers perform better than
choosing at random. Recently, d’Astous and Gaspero (2013) found that online sports
bettors’ return on investment (ROI) was positively associated with their experience and the
degree of information search and analysis they conducted. Their findings indicate that skill-
based factors such as experience or extensive information search can actually enable
bettors to achieve better (financial) outcomes, which counters previous findings (cf.
Cantinotti et al. 2004). Similarly, Forrest and Simmons (2002) found that sports experts
made more accurate predictions of the result of a set of soccer matches than the predictions
made by a random system. The results suggest that experts’ knowledge and experience
help in making better predictions. In another study, Andersson et al. (2005) examined the
predictions of five groups of people with different levels of expertise in soccer. Participants
predicted the outcome of the 2002 FIFA World Cup soccer tournament, and five groups
outperformed the completely random model. However, the group with most knowledge did
not perform better than the group having the least knowledge in soccer. The experts
overestimated their performance and tended to be more overconfident than their coun-
terparts with less knowledge. The researchers also manipulated participants’ access to
domain-relevant information and found that providing non-experts with information did
not improve their performance, but increased their confidence. Their findings highlight that
overconfidence does not translate into better performance. Therefore, the current study
aims to extend the literature by examining whether confidence leads to better performance
in another popular skill-based gaming context—the March Madness bracket.
J Gambl Stud
123
In our study, we propose that personal selections will amplify individuals’ confidence
when participating in a competitive pool. We also propose that such inflated confidence is
directly linked to enjoyment, which is found to be a primary reason people engage in
gambling behavior (e.g., Gupta and Derevensky 1998; Nower and Blaszczynski 2010).
Considering that enjoyment is a critical motivating factor, the more an individual believes
he or she has a better chance of winning, the more likely the person will find the game to be
entertaining.
Overview of Studies and Summary of Predictions
As discussed previously, both skill and luck play a role in predicting the outcome of a
sporting event. Given that making an accurate prediction is not driven by either purely luck
or skill per se, we contend that skill-relevant factors (e.g., personal selection and domain-
specific knowledge) will increase consumers’ confidence about winning. Two studies are
presented here to explore the overestimation phenomenon in a popular sports betting
context (i.e., March Madness pools). Study 1 examines the notion that the personal
selection process increases the illusion of control (Langer 1975). In the case of March
Madness bracket pools, we expect that individuals allowed to make their own picks will
report a greater probability of winning than individuals who did not make any selections,
after controlling for their previous bracket experiences and perceived knowledge about
sport (i.e., college basketball). As previously discussed, we hypothesize that the personal
selection process coupled with the innate desire for control would amplify one’s belief that
his or her selection will lead to the desirable outcome (Burger and Cooper 1979; Kelley
1971).
Hypothesis1 Making personal selections will increase individuals’ projection of their
probabilities of winning.
After establishing that the personal selection procedure increases one’s expectations of
winning, Study 2 sought to investigate the predictive role of expectations of winning on
various cognitive and behavioral outcomes. For instance, study 2 was designed to examine:
(1) the impact of confidence on maximum willingness to wager, (2) the relationship
between confidence and actual performance, and (3) the role of confidence in predicting
consumer enjoyment and willingness to participate in a bracket pool in the future. In line
with previous research on illusion of control and gambling behavior, we expect that more
confident individuals will wager more money than less confident individuals (Chau and
Phillips 1995). Chau and Phillips (1995) found that bettors tend to use minimal compu-
tational effort when placing bets, suggesting that subjective confidence would serve as a
cue that affects their wagering. To examine the relationship between confidence and actual
performance, we calculated each participant’s actual bracket score at the end of the
tournament. Given the difficult nature of achieving a better score through making delib-
erate choices than through random selections (McCrea and Hirt 2009), we expect that
confidence will not necessarily increase accuracy in predictions. Therefore, we propose the
following hypotheses.
Hypothesis 2 Individuals with high confidence in winning will intend to wager more
money than individuals with low confidence.
Hypothesis 3 Confidence will not improve actual performance.
J Gambl Stud
123
We also posit that inflated winning confidence is an important driver of consumer
enjoyment of participating in a bracket and willingness to play a bracket pool again.
Previous research has suggested that overconfidence is closely associated with positive
feelings such as enjoyment and arousal (e.g. Dunn and Wilson 1990; Kwak et al. 2010). In
addition, social cognitive theories of achievement and motivation suggest that task
enjoyment is a critical factor that facilitates intrinsic motivation (Elliot and Harackiewicz
1994; Puca and Schmalt 1999). As such, we posit that task enjoyment will enhance
intrinsic motivation, which will motivate consumers to participate in a similar contest in
the future. Based on a review of the literature, we advance the following hypotheses.
Hypothesis 4 Confidence in winning will have a positive effect on task enjoyment.
Hypothesis 5 Task enjoyment will have a direct effect on willingness to play again.
Hypothesis 6 The effect of confidence on willingness to play again will be mediated by
task enjoyment.
Study 1
Study 1 is designed to examine the impact of making personal selections on the devel-
opment of illusion of control. We compared individuals’ winning probability in two
groups: a group of participants given the opportunity to fill out their own brackets and a
group not given that opportunity.
Participants and Procedures
A total of 81 undergraduate students (56.8 % female) at a large Midwestern university in
the United States participated in this study. Subjects received course credit for participating
in this study. The majority of respondents were Caucasians (85.2 %) and the mean age of
the sample was 19.5 (SD =1.06). Data collection started one day after the 2012 March
Madness tournament bracket was released to the public and ended before the beginning of
the first round. This was to control for potential confounding from the results of early
rounds. Each participant who agreed to participate in the study was told that he or she
would be one of approximately 100 subjects competing in the pool. They were then
randomly assigned to one of two conditions—personal selection condition and control
condition. In the personal selection condition, participants were given the official printable
bracket sheet available from NCAA.com and were instructed to make their own predictions
for all 63 games. After completing their brackets, they responded to questionnaires asking
about their perceived probability in winning. In the control condition, participants were
given the blank official bracket sheet but were not told to make any predictions. After
viewing the bracket of the tournament, they were asked to respond to questionnaires
projecting their winning probabilities if they completed the bracket. Therefore, the only
difference between the two groups was the presence/absence of the personal selection
process. In both conditions, participants were also asked about their past bracket behaviors
and perceived knowledge about men’s college basketball. These measures were utilized as
control variables since these skill-relevant factors (e.g., foreknowledge, familiarity, past
behavior) have been found to increase illusion of control (Thompson et al. 1998). In
addition, these measures were used to ensure that participants in the two conditions did not
differ in terms of past bracket behaviors and perceived knowledge about college basketball.
J Gambl Stud
123
Measures
Perceived Probability of Winning
The illusion of control has been operationalized in different ways in the literature (see
Presson and Benassi 1996, for a review). For instance, studies have used various measures
such as using discrepancy scores between expected and actual performance, self-ratings of
perceived control, and self-ratings of confidence in winning. Presson and Benassi (1996)
found that larger effect sizes were reported in studies that measured participants’ perceived
ability to predict outcomes than participants’ perceived ability to control outcomes.
Therefore, we employ self-ratings of perceptions of likelihood of winning to assess indi-
viduals’ illusion of control in bracket participation. A single item was used to directly
measure respondents’ perceived probability of winning (‘‘How would you estimate your
chance (from 0 to 100 %) to become one of the top ten percent of scorers in this chal-
lenge?’’). Participants rated their winning probability using numerical values.
Past Bracket Behaviors
Two items were used to measure individuals’ past bracket behaviors. The first item asked
how often they participate in March Madness pools on a five-point scale (1 =never;
2=rarely; 3 =sometimes; 4 =frequently; 5 =every year). The second item asked the
average number of different brackets they fill out every year on a five point scale
(1 =none; 2 =one bracket; 3 =two brackets; 4 =three brackets; 5 =four brackets or
more).
Perceived Basketball Knowledge
Three-items were adapted from Kwak et al. (2010) to measure participants’ knowledge
about college basketball (‘‘I am knowledgeable about college basketball compared with the
average basketball fan,’’ ‘‘I am confident in using college basketball knowledge compared
with the average basketball fan,’’ and ‘‘I have better ability to comprehend college bas-
ketball information compared with the average basketball fan’’). The measures used a
seven-point Likert-type scale (1 =strongly disagree; 7 =strongly agree) and were
internally consistent (a=.97).
Results and Discussion
Table 1illustrates the sample description and mean differences of control variables (past
bracket behaviors, perceived basketball knowledge) across conditions. Fratios indicated
that there were no significant differences in past bracket behaviors and perceived knowl-
edge across groups. However, age was significantly different across groups [F(1,
81) =17.64, p\.001] and was thus included as a covariate in hypothesis testing.
In order to test the research hypothesis, analysis of covariance (ANCOVA) was con-
ducted by employing age, past bracket behaviors, and perceived basketball knowledge as
covariates. Levene’s test of equality of variance was nonsignificant (p[.26), indicating
that the assumption of homogeneity of variance was met for this sample. ANCOVA results
showed significant main effects of personal selection [F(1, 81) =4.74, p\.05, g
2
=.06].
Among covariates, only perceived knowledge [F(1, 81) =18.53, p\.001, g
2
=.20] was
significant. Frequency of participating in brackets [F(1, 81) =2.11, p=.15], number of
J Gambl Stud
123
brackets filled out [F(1, 81) =2.42, p=.12], and age [F(1, 81) =1.27, p=.26] were not
significant. As shown in Fig. 1, participants in the personal selection condition
(M=51.83, SD =26.24) felt more confident about their probability of winning than
participants in the control condition (M=34.80, SD =25.13). Therefore, Hypothesis1,
that making personal selections inflates individuals’ expectations of desired outcomes, was
supported after controlling for individuals’ past bracket behaviors and perceived basketball
knowledge.
This finding confirms the notion that personal selection leads to erroneous beliefs about
one’s choices (Langer 1975). While previous research on illusion of control has tended to
Table 1 Description of participants, past bracket behaviors, and perceived knowledge across conditions
(Study 1)
Variables Experimental condition
(N =41)
Control condition
(N =40)
Gender Female (56.1 %) Female (57.5 %)
Age 19.93 (SD =.88) 19.03 (SD =1.05)
Past bracket behavior
Frequency 3.34 (SD =1.46) 3.13 (SD =1.57)
Number of brackets 2.61 (SD =1.18) 2.43 (SD =1.34)
Perceived basketball knowledge 3.95 (SD =1.67) 3.63 (SD =1.98)
Mean differences of all past bracket behaviors and perceived basketball knowledge across conditions were
not statistically significant (Fs \.65, ps [.42). Age difference was statistically significant (F(1,
81) =17.64, p\.01). Thus, age was included as a covariate in the main analysis
Fig. 1 Box plots relevant to the comparison of the experimental and the control conditions
J Gambl Stud
123
focus on purely chance-based gaming contexts (e.g., lottery, slot machine), our finding
contributes to the literature by demonstrating that consumers’ personal choice increases
illusion of control in a skill-based gaming context as well. We also demonstrate that
domain-specific perceptions of knowledge contribute to the development of illusion of
control (Kwak et al. 2010; Thompson et al. 1998). Researchers agree with the notion that
individuals have innate desires to become causal agents of future outcomes (e.g., Burger
and Cooper 1975; deCharms 1968; Kelley 1971). Our findings show that the personal
selection procedure augments such expectations. We can speculate that participants felt
more confident about the desired outcome because they knew their performance would
hinge on the selections they made. In addition, one could also argue that lack of personal
involvement in predictions reduced illusion of control which might result in under-confi-
dence for those in the control condition. However, our data suggest that controlling for
individuals’ past bracket behaviors and perceived domain-specific knowledge, perceived
winning projection is positively associated with personal selection.
Study 2
Having demonstrated that making personal selections inflates individuals’ confidence,
Study 2 was designed to investigate the effects of self-ratings of probability of winning on
wager amounts, actual performance, and perceived enjoyment. In Study 2, we created a
‘‘March Madness Bracket Challenge’’ event and invited participants to participate in a real
contest. All participants filled out their brackets for the 2011 March Madness tournament,
and five participants among the top ten percent of scorers were randomly selected to be
rewarded with a gift card ($100 VISA debit card).
Participants and Procedures
Undergraduate and graduate students (N =197) attending a large Midwestern university
were recruited online from various courses in Communications Studies, Movement Sci-
ence and Sport Management. Forty-six percent of the sample were female and the mean
age of the sample was 25.9 (SD =4.92). Eighty-seven percent of the sample reported that
they have participated in March Madness pools in the past, and 69 % reported that they
have placed bets on the bracket (Avg. wager amount =$19.66, SD =19.88). Data col-
lection started one day after the 2011 March Madness tournament bracket was announced
and ended before the beginning of round 64. Each participant who agreed to participate in
the study was told that he or she would be one of approximately 200 subjects competing in
the pool. Participants were also told that the top 10 % of scorers would be eligible for a
random prize drawing of five $100 VISA gift cards (1/40 probability to become one of the
prize winners). Participants were told that their brackets would be calculated based on the
following scoring system: 2 points for each correct pick in the ‘‘Sweet 16’’; 4 points for
each correct pick in the ‘‘Elite 8’’; 8 points for each correct pick in the ‘‘Final Four’’; and
16 points for the correct pick in the ‘‘Final’’ (total eligible points add up to 128). After
reading the consent form, participants were given the official printable bracket sheet
available from NCAA.com and were instructed to make their predictions starting from the
round of 16, the round of 8, the final 4, and the championship game. After completing their
brackets, they responded to questionnaires asking about their probability of winning, wager
amount, past bracket behavior, perceived basketball knowledge, and perceived enjoyment
of the bracket competition. After the tournament was completed, each participant’s bracket
J Gambl Stud
123
performance was calculated based on the scoring system described above. Scores in each
round were added up to create a total score for each participant.
Measures
For perceived probability of winning, past bracket behaviors, and perceived basketball
knowledge, same measures used in Study 1 were utilized. Mean probability of winning a
prize was 46.38 (range 0-100, SD =27.62). The multi-item perceived basketball knowl-
edge scale was internally consistent (a=.97). They were asked to freely estimate their
chance (from 0 to 100 %) to become eligible for prize (probability range 0-100;
Mean =46.38, SD =27.62).
Maximum Willingness to Wager
A single item was used to directly measure participants’ maximum willingness to wager on
their own bracket. The item asked respondents ‘‘If you had an option to place a bet, what is
the maximum amount you would consider wagering on your current bracket?’’ Participants
responded to this questionnaire in dollar amounts (amount range $0-200; Mean =15.79,
SD =17.53).
Task Enjoyment
Respondents’ degree of enjoyment in participating in the March Madness bracket com-
petition was adapted from Kwak et al. (2010). The scale was used to capture the hedonic
value of the competition. Participants were asked to rate their overall enjoyment playing in
a fantasy football league using a measure consisting of five seven-point semantic differ-
ential items (e.g., ‘‘I feel the XXX Bracket Challenge is dull-exciting; delightful-not
delightful; thrilling-not thrilling; fun-not fun; unenjoyable-enjoyable’’ ; a=.89).
Actual Performance
Each participant’s bracket performance was calculated after the completion of the cham-
pionship game. As previously described, participants’ scores were calculated using the
following scoring system: 2 points for each correct pick in the ‘‘Sweet 16’’; 4 points for
each correct pick in the ‘‘Elite 8’’; 8 points for each correct pick in the ‘‘Final Four’’; 16
points for the correct pick in the ‘‘Final.’’ All scores for different rounds were added to
create a total bracket score (score range 4-66; Mean =30.05, SD =10.00).
Willingness to Play Again
Participants’ willingness to play a bracket pool again was measured with three seven-point
semantic differential items asking about participants’ likelihood of participating in the
bracket challenge contest next year (‘‘How likely is that you would play the XXX Bracket
Challenge again next year?’’ Improbable-probable; unlikely-likely; impossible-possible;
a=.96).
J Gambl Stud
123
Results and Discussion
Illusion of Control and Wager Amount: H2
We converted perceived probability of winning estimates into standardized scores
(z scores; mean =0, SD =1) to facilitate comparisons of wager amounts based on par-
ticipants’ perceived chances of winning. Participants with 1 SD above and 1 SD below the
mean perceived chances of winning were grouped into either a high confidence group
(1 SD above) or a low confidence group (1 SD below). Then ANCOVA was conducted to
examine whether being in the high confidence group resulted in being willing to wager a
greater amount of money than being in the low confidence group, while controlling for past
bracket behaviors, gender, and perceived basketball knowledge. ANCOVA results showed
a significant main effect of perceived probability of winning [F(1, 81) =8.56, p\.01,
g
2
=.10]. Among covariates, only number of brackets completed in the past year [F(1,
81) =3.84, p\.054] was marginally significant. All other covariates were nonsignificant
(Fs \1.10, ps [.30). As shown in Fig. 2, high-confidence individuals reported that they
would wager more money (M=$22.95, SD =12.91) than low confidence individuals
(M=$8.85, SD =10.26). Therefore, H2 was supported.
Illusion of Control and Actual Performance: H3
Similar to the test for H2, an ANCOVA was utilized to examine whether high confidence
individuals demonstrated better performance than low confidence individuals. Participants
were grouped into either a high confidence group or low confidence group based on
standardized z-scores. Again, gender, past bracket behaviors, and perceived basketball
Fig. 2 Mean of maximum wager amount as a function of confidence in winning (Study 2). Error bars
denote 95 % confidence intervals
J Gambl Stud
123
knowledge were entered as covariates. ANCOVA results indicated no significant effect of
perceived chances of winning on actual bracket scores [F(1, 81) =.38, p=.54]. Among
covariates, only past bracket behavior was marginally significant F(1, 81) =3.43, p=.07,
g
2
=.04]. Other covariates were statistically nonsignifcant (Fs \3.03, ps [.09). As
shown in Fig. 3, high confidence individuals’ average score (M=31.23, SD =8.34) was
not significantly higher than low confidence individuals (M=26.86, SD =12.35). Thus,
H3 was supported.
Illusion of Control, Enjoyment, and Willingness to Play Again: H4, H5, and H6
To examine the relationship between illusion of control, enjoyment, and willingness to
play again, we employed the Statistical Package for the Social Sciences (SPSS) macro for
testing simple mediation effects (Preacher and Hayes 2004). In order to better understand
the relationships among other variables, past experience and basketball knowledge vari-
ables were included as covariates in the mediation model. Figure 4shows that perceived
probability of winning is positively associated with task enjoyment (b=.44, t=6.04,
p\.001) and willingness to play again (b=.38, t=5.77, p\.001). Enjoyment also had
a positive impact on willingness to play again (b=.60, t=6.86, p\.001). Among
covariates, past experience had a positive impact on willingness to play again (b=.29,
t=3.70, p\.01) but basketball knowledge had no significant impact on willingness to
play again (b=.01, t=.13, p=.89). Therefore, both H4 and H5 were supported. When
controlling for covariates and enjoyment, the direct impact of perceived probability of
winning on willingness to play again became non-significant (b=.15, t=1.13, p=.12).
Furthermore, as hypothesized, perceived probability of winning had an indirect positive
effect on willingness to play again mediated by enjoyment (b=.26). We tested the
Fig. 3 Mean actual bracket scores as a function of winning confidence (Study 2). Error bars denote 95 %
confidence intervals
J Gambl Stud
123
significance of this indirect effect using bootstrapping procedures (Preacher and Hayes
2004). Unstandardized indirect effects were computed for each of 1,000 bootstrapped
samples, and the 95 % confidence interval was computed. The two-tailed significance test
(assuming normal distribution) demonstrated that the indirect effect was significant (So-
bel’s z=4.51, p\.001). This indirect effect was still significant event after controlling
for covariates. The 95 % confidence interval ranged from .16 (lower limit) to .40 (upper
limit). Therefore, enjoyment mediated the effect of confidence on willingness to play
again, in support of H6.
General Discussion
While consumers’ engagement in predicting outcomes of uncertain events (e.g., March
Madness, Super Bowl) has become a popular and widely accepted form of entertainment,
relatively little is understood about consumers’ gaming behavior (Lam 2007; Sierra and
Hyman 2009). In two studies, we demonstrated the existence of illusion of control in the
NCAA Men’s Basketball Tournament context. Personal selection had a biasing effect of
inflating participants’ estimation of the likelihood of winning (Study 1), and such per-
ception was indeed ‘‘illusory’’ in that it did not predict actual performance (Study 2).
However, more importantly, we showed that such inflated control beliefs were significantly
associated with participants’ maximum willingness to wager and enjoyment of partici-
pating in a competition.
In two studies, we demonstrated the existence and predictive utility of the illusion of
control effect, which occurs when personal selections are made, and is significant in
predicting maximum willingness to wager, consumer enjoyment, and willingness to play
again. Study 1 showed that making personal selections increased confidence in winning.
Participants who invested time to make actual predictions about the upcoming tournament
were more optimistic about their performance than those who did not make any prediction
(see Fig. 1). What psychological mechanisms underlie the increase in probability projec-
tion when participants complete their own brackets? In the literature review, we suggested
that the control motivation coupled with personal selection increases erroneous beliefs
about control over outcomes (e.g., Kelley 1971). As an alternative explanation, cognitive
Fig. 4 Standardized regression coefficients for the relationship between perceived confidence in winning
and willingness to play again as mediated by task enjoyment (Study 2). The standardized regression
coefficient between perceived winning confidence and willingness to play again, controlling for enjoyment
and other covariates (past experience and basketball knowledge), is in parentheses
J Gambl Stud
123
‘‘effort’’ involved in making predictions would likely predispose participants to be opti-
mistic about their choices. Given that we only allowed the participants in the experimental
condition to make their own selections, we believe this increase in confidence is in part due
to overvaluation of their personal effort.
Recently, Norton et al. (2011) proposed the ‘‘IKEA effect,’’ which asserts that personal
labor increases the valuation of completed tasks. They found that a positive link between
effort and perceived value of finished task was evident in both consumers who are inter-
ested in ‘‘do-it-yourself’’ projects and those who are relatively uninterested. As such,
consumers would overestimate the value of the finished task (bracket) when they are
actively engaged in doing the work (selection) process. The authors also contended that the
overvaluation of individuals’ effort could be more apparent when the task orientation is for
pleasure. Considering that participating in a March Madness pool is a hedonic experience,
it seems plausible that making predictions in each round might increase one’s valuation of
his or her own effort. Although their research context is not directly relevant to making
decisions under conditions of uncertainty per se, the notion that consumers overvalue their
effort in finishing a given task offers a valid potential explanation for this interesting
phenomenon. Future studies should examine individuals’ perceived effort to establish the
link between two judgments—evaluation of inputs and prediction of outcomes (Kahneman
and Tversky 1973).
Given the mixed findings in the sports betting literature (e.g., Andersson et al. 2005;
Cantinotti et al. 2004; d’Astous and Gaspero 2013), our findings provide additional
empirical evidence that skill-relevant factors in fact do not contribute to actual perfor-
mance. For instance, we found that domain-specific knowledge and perceived probability
of winning did not contribute to actual performance. Our findings confirmed that confi-
dence in winning is indeed overrated and has no relationship to actual performance.
However, we found that such inflated confidence in winning was the only significant
predictor of consumer enjoyment. Based on our findings, we posit that the illusion of
control is a critical motivator that enhances task enjoyment, which has a direct effect on
willingness to play again. As such, we show that irrational cognitive processing has utility
in predicting consumer enjoyment and willingness to play a tournament bracket pool again
in uncertain situations.
Enjoyment is the predominant factor that motivates non-pathological gamblers to
engage in gambling behavior (e.g., Gupta and Derevensky 1998). As social gamblers,
participants would play March Madness pools to increase their enjoyment of watching the
actual event (Mandel and Nowlis 2008; Lin et al. 2012). Our findings suggest that illusion
of control was significantly associated with participants’ enjoyment of playing a tourna-
ment bracket (Kwak et al. 2010). Given that illusion of control is positively associated with
enjoyment, more research should be conducted to identify if the enjoyment derived from
overconfidence reinforce erroneous perception about one’s control over the outcome. More
importantly, from a clinical perspective, one would wonder if there is a tipping point that
non-pathological gamblers show psychological and behavioral markers of problem gam-
blers. Obviously, much more research should be conducted to distinguish different gam-
bling patterns and behaviors among non-pathological and pathological gamblers. Recently,
Orgaz et al. (2013) found that pathological gamblers demonstrated greater illusions on their
daily lives not restricted to gambling than undiagnosed sample. Their findings indicate that
pathological gamblers tend to overestimate cause-effect relationships in general than non-
pathological gamblers. Future studies should be conducted to explore how filling out the
bracket develops illusions of control among pathological- and non-pathological gamblers.
Perhaps, the direct link between personal selection and overconfidence would be more
J Gambl Stud
123
salient among pathological gamblers than non-pathological gamblers. However, the role of
domain-specific knowledge and relevant experience would be driving illusion of control
for non-pathological gamblers. Likewise, it would be interesting to examine how patho-
logical gamblers and non-pathological gamblers differ on developing illusion of control in
skill-based gaming situations.
Limitations and Directions for Future Research
Several limitations should be acknowledged. First, participants were recruited from a
Midwestern university and therefore, the results cannot be generalized to a broader
audience. Targeting different populations based on previous gambling behaviors would be
useful to further extend the current study. For instance, it would be interesting to compare
pathological gamblers and non-pathological gamblers on their level of illusion of control in
a tournament bracket setting. Second, we used self-report measure when assessing indi-
viduals’ wager amount if s/he had an opportunity. While such approach might be appro-
priate in gauging participants’ interest in wagering, there could be discrepancies between
willingness to wager and actual wagering behavior. Therefore, future studies should
address this issue by establishing the validity of self-reported wagering intent (Hodgins and
Makarchuk 2003). Although our findings demonstrate that personal selection increased
participants’ estimation of the likelihood of winning, further research should be conducted
to examine mechanism underpinning the overvaluation process. In particular, are there any
processing elements that facilitate overvaluation of participants’ perceived control in
uncertain contexts? For instance, do participants who spend more time (more cognitive
effort) on completing the bracket report greater levels of perceived control than those who
spend less time, after controlling for their domain-specific knowledge? While we speculate
that time spent on completing the bracket would elicit the perception of control, the
question of whether confidence in winning is also a direct function of time (effort) spent on
filling out the bracket remains to be explored. Therefore, future studies should measure
time filling out the bracket or assess individuals’ cognitive effort exerted during the
selection process to assess the relationship between effort and participants’ gaming
behavior (Garbarino and Edell 1997).
Our study showed that confidence does not translate into better performance. This
finding enabled us to replicate the illusion of control theory in the March Madness context.
Previous research suggests that individuals must correctly predict an upset (a win by a
higher seeded, unknown team over a lower seeded, favored team) to achieve a higher score
in the tournament pools (McCrea and Hirt 2009). Future research could examine whether
the number of upsets predicted by an individual can enhance his or her confidence in
winning.
In addition, our study did not explicitly measure participants’ emotions. In the gambling
literature, arousal is closely associated with gambling behavior and is one of the highly
sought emotional rewards that gamblers find gratifying (e.g., Neighbors et al. 2002).
Therefore, assessing participants’ arousal level and establishing its link with overestima-
tion of control could advance the current findings. Future research could also incorporate
anticipated emotion (e.g., anticipated pleasure from winning or anticipated regret from
losing; Cowley 2013; Perugini and Bagozzi 2001) and examine its impact on probability
estimation as well as on individuals’ betting decisions. Understanding the interplay
between emotions and erroneous cognitions on the decision making process would further
advance our understanding of gambling behaviors in uncertain contexts. Furthermore, it
would be interesting to see if the illusion of control reduces followed by an undesired
J Gambl Stud
123
outcome (Matute and Blanco 2014). If most of the predictions in the first round matches
were incorrect, do inaccurate predictions reduce illusions? It would be interesting to
examine if illusions change throughout the course of actions during the Championship
tournament. In addition, it would be interesting to compare pathological gamblers and non-
pathological gamblers in their processing of undesired outcomes in a bracket prediction
setting.
Overall, the current research provides empirical evidence regarding the overestimation
phenomena in one of the most widely wagered events in the U.S.–the NCAA Men’s
Basketball Tournament. We show that personally engaging in the selection procedure of
filling out a bracket increases confidence in winning. Enhanced confidence motivates
participants to wager more money, but it has no impact on actual performance, suggesting
that the confidence depends on illusory perceptions. However, in an uncertain context in a
competitive environment, such inflated confidence is a key determinant of task enjoyment
and willingness to play a tournament bracket again.
Acknowledgments This research was funded by the Office of Research at the University of Michigan. The
author would like to thank Joon Sung Lee, for his help with data collection and coding.
References
Allcock, C. (1987). An analysis of a successful racing system. In M. Walker (Ed.), Faces of gambling (pp.
181–187). Sydney: National Association for Gambling Studies.
Andersson, P., Edman, J., & Ekman, M. (2005). Predicting the World Cup 2002 in soccer: Performance and
confidence of experts and non-experts. International Journal of Forecasting, 21(3), 565–576.
Ayeroff, F., & Abelson, R. P. (1976). ESP and ESB: Belief in personal success at mental telepathy. Journal
of Personality and Social Psychology, 34(2), 240.
Burger, J. M., & Cooper, H. M. (1979). The desirability of control. Motivation and Emotion, 3(4), 381–393.
Cantinotti, M., Ladouceur, R., & Jacques, C. (2004). Sports betting: Can gamblers beat randomness?
Psychology of Addictive Behaviors, 18(2), 143.
Chau, A. W., & Phillips, J. G. (1995). Effects of perceived control upon wagering and attributions in
computer blackjack. The Journal of General Psychology, 122(3), 253–269.
Cowley, E. (2013). Forgetting the anxiety: Gambler’s reactions to outcome uncertainty. Journal of Business
Research, 66(9), 1591–1597.
Crupi, A. (2011, March 1). March Madness still one of the biggest sports franchises. Ad Week. Retrieved
from http://www.adweek.com/news/television/march-madness-still-one-biggest-sports-franchises-
1258890
d’Astous, A., & Gaspero, M. D. (2013). Explaining the performance of online sports bettors. International
Gambling Studies, 13(3), 371–387.
deCharms, R. (1968). Personal causation. New York: Academic Press.
Dunn, D. S., & Wilson, T. D. (1990). When the stakes are high: A limit to the illusion-of-control effect.
Social Cognition, 8(3), 305–323.
Dykstra, S. P., & Dollinger, S. J. (1990). Model competence, depression, and the illusion of control. Bulletin
of the Psychonomic Society, 28(3), 235–238.
Elliot, A. J., & Harackiewicz, J. M. (1994). Goal setting, achievement orientation, and intrinsic motivation:
A mediational analysis. Journal of Personality and Social Psychology, 66(5), 968–980.
Forrest, D., & Simmons, R. (2002). Outcome uncertainty and attendance demand in sport: The case of
English soccer. Journal of the Royal Statistical Society: Series D (The Statistician), 51(2), 229–241.
Garbarino, E. C., & Edell, J. A. (1997). Cognitive effort, affect, and choice. Journal of Consumer Research,
24(2), 147–158.
Griffiths, M. (1995). Towards a risk factor model of fruit machine addiction: A brief note. Journal of
Gambling Studies, 11(3), 343–346.
Gupta, R., & Derevensky, J. L. (1998). Adolescent gambling behavior: A prevalence study and examination
of the correlates associated with problem gambling. Journal of Gambling Studies, 14(4), 319–345.
Hodgins, D. C., & Makarchuk, K. (2003). Trusting problem gamblers: Reliability and validity of self-
reported gambling behavior. Psychology of Addictive Behaviors, 17(3), 244–248.
J Gambl Stud
123
Jessop, A. (2014, January 17). The business of the bracket: How Vegas and businesses capitalize on March
Madness. Forbes. Retrieved from http://www.forbes.com/sites/aliciajessop/2013/03/20/the-business-
of-the-bracket-how-vegas-and-businesses-capitalize-on-march-madness/
Kahneman, D., & Tversky, A. (1973). On the psychology of prediction. Psychological Review, 80(4), 237.
Kelley, H. H. (1971). Attribution in social interaction. Morristown, NJ: General Learning Press.
Kwak, D. H., Lee, J. S., & Mahan, J. E, I. I. I. (2013). Ad-evoked illusory judgments in fantasy sports
participation: Effects of customization level and expert information. Journal of Sport Management,
27(5), 393–406.
Kwak, D. H., Lim, C. H., Lee, W. Y., & Mahan, J, I. I. I. (2010). How confident are you to win your fantasy
league: Exploring the antecedents and consequences of winning expectancy. Journal of Sport Man-
agement, 24(4), 416–433.
Ladouceur, R. (2004). Perceptions among pathological and nonpathological gamblers. Addictive Behaviors,
29(3), 555–565.
Ladouceur, R., Giroux, I., & Jacques, C. (1998). Winning on the horses: How much strategy and knowledge
are needed? The Journal of Psychology, 132(2), 133–142.
Ladouceur, R., & Se
´vigny, S. (2005). Structural characteristics of video lotteries: Effects of a stopping
device on illusion of control and gambling persistence. Journal of Gambling Studies, 21(2), 117–131.
Lam, D. (2007). An exploratory study of gambling motivations and their impact on the purchase frequencies
of various gambling products. Psychology & Marketing, 24(9), 815–827.
Langer, E. J. (1975). The illusion of control. Journal of Personality and Social Psychology, 32(2), 311.
Lin, C. H., Hung, H. H., & Li, Y. H. (2012). How confidence and uncertainty affect consumers’ enjoyment
of gambling. Social Behavior and Personality: An International Journal, 40(3), 425–432.
Linn, A. (2013, March 21). Going for a perfect NCAA bracket? You’re more likely to win Powerball. NBC
News. Retrieved from http://www.nbcnews.com/business/careers/going-perfect-ncaa-bracket-youre-
more-likely-win-powerball-f1C8981570
Mandel, N., & Nowlis, S. M. (2008). The effect of making a prediction about the outcome of a consumption
experience on the enjoyment of that experience. Journal of Consumer Research, 35(1), 9–20.
Matute, H., & Blanco, F. (2014). Reducing the illusion of control when an action is followed by an
undesired outcome. Psychonomic Bulletin & Review, 21, 1087–1093.
McCarthy, M. (2012, March 27). March Madness betting now tops Super Bowl. USA Today. Retrieved from
http://content.usatoday.com/communities/gameon/post/2012/03/march-madness-betting-bigger-than-
super-bowl-ncaa-las-vegas-nevada-ncaa-mens-final-four/1#.U5XB-vldV8E
McCrea, S. M., & Hirt, E. R. (2009). Match Madness: Probability matching in prediction of the NCAA
Basketball Tournament. Journal of Applied Social Psychology, 39(12), 2809–2839.
Moore, S. M., & Ohtsuka, K. (1999). Beliefs about control over gambling among young people, and their
relation to problem gambling. Psychology of Addictive Behaviors, 13(4), 339.
Neighbors, C., Lostutter, T. W., Cronce, J. M., & Larimer, M. E. (2002). Exploring college student gambling
motivation. Journal of Gambling Studies, 18(4), 361–370.
Norton, M., Mochon, D., & Ariely, D. (2011). The ‘IKEA effect’: When labor leads to love. Harvard
Business School Marketing Unit working paper, (11-091).
Nower, L., & Blaszczynski, A. (2010). Gambling motivations, money-limiting strategies, and precommit-
ment preferences of problem versus non-problem gamblers. Journal of Gambling Studies, 26(3),
361–372.
Orgaz, C., Este
´vez, A., & Matute, H. (2013). Pathological gamblers are more vulnerable to the illusion of
control in a standard associative learning task. Frontiers in Psychology, 4, 1–7.
Perugini, M., & Bagozzi, R. P. (2001). The role of desires and anticipated emotions in goal-directed
behaviours: Broadening and deepening the theory of planned behaviour. British Journal of Social
Psychology, 40(1), 79–98.
Preacher, K. J., & Hayes, A. F. (2004). SPSS and SAS procedures for estimating indirect effects in simple
mediation models. Behavior Research Methods, Instruments, & Computers, 36(4), 717–731.
Presson, P. K., & Benassi, V. A. (1996). Illusion of control: A meta-analytic review. Journal of Social
Behavior & Personality, 11(3), 493–510.
Puca, R. M., & Schmalt, H. D. (1999). Task enjoyment: A mediator between achievement motives and
performance. Motivation and Emotion, 23(1), 15–29.
Rogers, P. (1998). The cognitive psychology of lottery gambling: A theoretical review. Journal of Gambling
Studies, 14(2), 111–134.
Sierra, J. J., & Hyman, M. R. (2009). In search of value: A model of wagering intentions. The Journal of
Marketing Theory and Practice, 17(3), 235–250.
Thompson, S. C., Armstrong, W., & Thomas, C. (1998). Illusions of control, underestimations, and accu-
racy: A control heuristic explanation. Psychological Bulletin, 123(2), 143–161.
J Gambl Stud
123