Printer Friendly

Exercising willpower: differences in willpower depletion among athletes and nonathletes.

"In the absence of willpower the most complete collection of virtues and talents is wholly worthless. "

--Aleister Crowley


Participation in high school and college athletics has been linked to an earnings premium of between 4% and 10% (Barron, Ewing, and Waddell 2000; Cabane and Lechner 2014; Ewing 1998, 2007; Lechner 2009; Lechner and Downward 2013; Lechner and Sari 2014; Long and Caudill 1991; McCormick and Tinsley 1987; Rehberg and Schafer 1968; Rooth 2011; Spreitzerand Pugh 1973; Stevenson 2010). Stevenson (2010) identified a causal link between athletic participation and greater labor force participation and higher wages among women. What remains unclear, however, is the avenue through which this wage premium for athletes occurs. If the wage premium exists, holding constant measurable skills such as grade point average (GPA) and standardized test scores, do athletes possess other noncognitive skills that are valued by the market such as willpower, determination, competitiveness, and teamwork?

In this study, we consider athletes and willpower. Based on Stevenson's findings, it is plausible that an athlete's willpower reserves could be an important source of their earnings premium. The significance of willpower as an important noncognitive skill was first recognized by Walter Mischel (1958) with his famous "marshmallow experiments" in which he linked the ability to stave off immediate gratification among small children to larger rewards later in life such as higher grades, greater graduation rates and, ultimately, higher earnings. Recent research (Baumeister, Heatherton, and Tice 1994; Baumeister et al. 1998; Muraven, Tice, and Baumeister 1998) has established that individuals have a limited supply of willpower that can be depleted. Hence, willpower is like a muscle that becomes exhausted with use. One can, however, strengthen this muscle with exercise, and develop an ability to diminish willpower depletion. (1)

Based on the sports earnings premium literature, we hypothesize that athletes, by their participation in formal physical activities over many years, have developed and strengthened their willpower muscle, (2) and when tested, will exhibit less depletion as compared to nonathletes. In order to test this, we conduct a laboratory experiment based on the procedure laid out by Muraven, Tice, and Baumeister (1998) in which participants are timed while attempting to solve an unsolvable puzzle. Participants in treatment groups underwent a willpower-draining task of restraining from expressing emotions while watching an emotionally provocative video before attempting the puzzle, while control group participants watched the same video but were able to express emotions.

We randomly assigned athletes and nonathletes into control and treatment groups, resulting in four separate experimental groups: athlete-treatment, athlete-control, nonathlete-treatment, and nonathlete-control. We hypothesized that athletes in the treatment group would spend more time on the puzzle, thereby exhibiting less willpower depletion. We found that treated athletes worked significantly longer on the difficult task than nonathletes. Because of the random assignment of treatment group and control group participants, we interpret this finding as causal evidence that after expending some self-control, athletes experience less willpower depletion than nonathletes in subsequent tasks.

This article proceeds as follows: Section II describes the previous literature results on both the links between participation in athletics and an earnings premium as well as the literature on willpower and willpower depletion. Section III describes the experiment, Section IV describes the model and the results, and Section V concludes the article.


High school and college athletes earn more in labor markets (Barron, Ewing, and Waddell 2000; Cabane and Lechner 2014; Ewing 1998, 2007; Lechner 2009; Lechner and Downward 2013; Lechner and Sari 2014; Long and Caudill 1991; McCormick and Tinsley 1987; Rehberg and Schafer 1968; Rooth 2011; Spreitzer and Pugh 1973; Stevenson 2010). Long and Caudill (1991) document a 4% premium in annual incomes among males who participated in U.S. college athletics. Ewing (2007) finds a wage effect of 6% using the National Longitudinal Survey of Youth while Lechner (2009) finds that sports activities increase the long-term earnings of German males by 5%-10%. Rooth (2011), relying on a sibling fixed effects model, finds an earning premium to good physical fitness of nearly 4%, and 1.7% with noncognitive controls added.

It is clear that participation in athletics is correlated with abilities that determine wages and which are not measured by other observable variables (Stevenson 2010, 5). The avenue through which this relationship occurs, however, is not as clear. Are the benefits associated with participation in sports caused by the participation or are they associated with the particular characteristics of the individual who chooses to participate in athletics? "Athletes tend to be more extroverted, aggressive, and achievement oriented. Are these traits they bring to athletics or are these traits athletics bring to them? Are they learning valuable skills? Or are the high skilled simply more likely to participate in sports?" (Stevenson 2010,1). For example, Hyytinen and Lahtonen (2013) found that physical activity can make people more persistent at work-related activities, translating into higher earnings.

Relying on a natural experiment driven by the legal changes associated with the introduction of Title IX, (3) Stevenson (2010) identified a causal relationship between participating in high school sports and earnings stemming from the large-scale policy change requiring female participation in sports to meet the level at which boys participate on a state-by-state basis (Stevenson 2010). Exploiting the differences across states in boys' participation, Stevenson concluded that being a high school athlete translates into an 8% wage premium for women (Stevenson 2010, 24).

The research presented in this article relies on Stevenson's (2010) work establishing that participation in athletics fosters the development of noncognitive skills valued by the market. "Athletics is a highly regulated system in which social conflict is displayed in a positive light. From this, players learn how to compete and how to operate successfully under a formal code of rules and procedures. Furthermore, players are taught to function as a team" (Stevenson 2010, 6). In particular, we hypothesize that athletics fosters willpower and that athletes develop a stronger willpower muscle than nonathletes that is less prone to depletion.

Willpower is traditionally defined as the ability to control impulses and to persevere with tasks. Mischel (1958) pioneered the research on willpower through an experiment in which he measured the extent to which children, aged four to six could delay their gratification from eating a treat of their choice (Oreo cookie, marshmallow, or pretzel stick). Mischel posited that only through the exertion of willpower could the participants delay the immediate reward (a single marshmallow) for a greater reward (two marshmallows). Perhaps more interesting than Mischel's initial findings, however, are the follow-up studies in which he recruited the same research participants from the marshmallow study 20 years later as adults. Mischel, Shoda, and Peake (1988) found that the adults who displayed the most willpower in the marshmallow experiment as preschoolers went on to score higher on college entrance exams, earn higher salaries, display fewer problems with drugs and alcohol, and even have lower body mass indices. (4)

More recent research has shown that exerting willpower is like using a muscle that can get exhausted with use, leaving individuals unable to carry out tasks that require further self-control or willpower (Baumeister, Heatherton, and Tice 1994; Baumeister et al. 1998; Pocheptsova et al. 2009; Schmeichel, Vohs, and Baumeister 2003; Vohs and Heatherton 2000; Vohs et al. 2011). Baumeister used the term "ego depletion" to explain the phenomenon. Ego depletion refers to the idea that willpower draws upon a limited pool of mental resources and that this pool of energy can be exhausted.

Resisting temptation seems to have produced a psychic cost, in the sense that afterward participants were more inclined to give up easily in the face of frustration. It was not that eating chocolate improved performance. Rather, wanting chocolate but eating radishes instead, especially under circumstances in which it would seemingly be easy and safe to snitch some chocolates, seems to have consumed some resources and therefore left people less able to persist at the puzzles. (Baumeister and Tierney 2011, 20)

Experiments inside and outside the laboratory consistently demonstrate two outcomes: People have a finite amount of willpower that depletes with use, and we rely on that stock of willpower for many different tasks that require self-control (for a review, see Muraven and Baumeister 2000).

While various measures suggest that there are stable individual differences in willpower stamina (Costa and McCrae 1992; Gosling, Rentfrow, and Swann 2003; Mischel 1974; and Tangney, Baumeister, and Boone 2004), there is also some evidence that one's "bank" of willpower can be enhanced, stockpiled, and rebuilt. Baumeister et al. (2006) found that regular exercises of willpower can improve self-control and make it so that willpower is less quickly depleted. Muraven, Baumeister, and Tice (1999) conducted a study in which subjects engaged in one of three self-regulatory exercises: tracking food eaten, improving mood, or improving posture for 2 weeks. When compared to a baseline stamina task performed after ego depletion, there was an overall improvement in self-regulation compared to a control group that did no exercises.

Significant to our hypothesis, Oaten and Cheng (2006a, 2006b) found that regular exercise improved self-control. Generally, successful exertion among athletes may require willpower to force oneself to continue despite physical fatigue, so we could expect that athletes, who have presumably participated in regular exercise for years, will be less likely to experience willpower depletion than the nonathlete. It is important to note that in these studies "improvement was in the form of increased stamina (resistance to the debilitating effects of resource depletion), rather than increased capacity" (Oaten and Cheng 2006a, 2006b).


A. Procedure

To test our hypothesis, we closely mirrored the experimental procedure laid out by Muraven, Tice, Baumeister (1998). Specifically, after having reviewed the experimental procedures and completing a short demographic questionnaire, (5) we exposed subjects to an "emotionally evocative film" (a series of funny videos). Participants in the treatment group were required to stifle their emotional response, a willpower-draining task (Baumeister et al. 2007, 352). Participants in the control group were not asked to stifle their responses. (6)

After exposing subjects to the funny videos, both the control and experimental groups completed a feedback survey regarding the videos. Participants in each condition were asked to rate the degree to which they found the two videos funny. The outcome of the survey among those treated is a plausible proxy measure to control for the amount of willpower necessary to maintain self-control and hence a control across individuals to relative willpower depletion.

After completing the survey, participants were asked to solve a "challenging puzzle." (7) As in the Muraven, Tice, and Baumeister (1998) study, we attempted to include a measure of willpower that was "conceptually and subjectively distinct" from the self-control involved in the initial task (p. 776). The laboratory operator demonstrated a correct answer to a simpler, solvable, version of the challenging puzzle and then helped answer questions and guide participants through a correct answer to a more difficult, but still solvable puzzle which had all of the same rules as the challenging puzzle. The problem-solving task (puzzle) was adopted from Glass, Singer, and Friedman (1969) and Feather (1961), requiring the participant to trace a geometric figure without retracing any lines and without lifting his or her pencil from the paper. Multiple slips of paper were provided for each figure, so the person could take multiple attempts. After the practice puzzles were complete, the experimenter gave the participants the main test figure with the following instructions:
   You will have up to 30 minutes to complete the puzzle; however, if
   at any point during these 30 minutes you choose to give up you may
   do so. A stopwatch will be provided so that you can monitor the
   amount of time you have remaining. If you decide to give up before
   the 30 minutes has concluded please stop the stopwatch by clicking
   the button on the top right of the watch and ring the bell to
   inform the researcher who will be outside of the room. You will
   then be paid for your participation and your participation will be

The experimenter collected information on the amount of time (in seconds) spent working on the puzzle as a measure of persistence. All participants were paid $ 10 for their participation, and were notified of this payment prior to their participation. The payment was not conditional on the time spent on the task. (8)

B. Participants

Data were collected in individual sessions from 75 undergraduate students (9) at the University of Wisconsin-La Crosse (UW-L). (10) UW-L is a comprehensive 4-year institution within the University of Wisconsin-System, awarding bachelors, masters, and one doctoral degree with a student population of nearly 11,000, of which nearly 10,000 are undergraduates. The student body is 57% female and 88% Caucasian with an average American College Testing (ACT) score of 25. (11) Our sample reflected the University itself in its gender and ethnic makeup, consisting of 57% women and 99% Caucasian participants.

The university sponsors 19 Division-Ill intercollegiate athletic teams, (12) as well as a host of additional intramural clubs in which students compete with other students at UW-L. There are approximately 570 student athletes participating in intercollegiate athletics at UW-L and 3,300 students participating in intramural athletics. (13)

Participants were recruited under the guise of a Universal Humor Study. Individuals were verbally prescreened with the question "Did you participate in varsity athletics in your senior year in high school?" and were then randomly assigned to a treatment and control group. This verbal screen was designed to isolate the more serious athletes in the sample as proxied by their level of participation in a sport. In the United States, "varsity teams" are the highest level of athletics at the high school level, broken down by classes within states that are based on the sizes of the high school. It is estimated that 55.5% of all students play a sport at some point during the 4 years of high school (Koebler 2011), but fewer play at the varsity level and through their senior year in high school. Fewer still go on to play sports on an intercollegiate college team. (14)

For the purpose of this study, we define "athlete" in this sample as an individual who participated in varsity athletics in their senior year in high school and/or participates in intercollegiate athletics at UW-L. (15) Because the sample includes both those that played in their senior year in high school and intercollegiate athletes, the current level of intensity of athletic participation will vary between participants, even among those participants who are identified as an athlete. We did not delineate between the type of sport in which the participant played, nor did we make a delineation with respect to ability. (16) Table 1 provides the breakdown of our sample, including our definition of athletes (in bold) and nonathletes by gender. As noted above, this level of participation in competitive sports in the United States represents a small proportion of students and requires a high degree of commitment, talent, and athleticism.

We expect that athletes have built up their willpower reserves and should be less subject to the effects of willpower depletion. Therefore, we hypothesize that athletes will experience less "willpower drain," as measured by time spent working on the difficult puzzle.


A. Model

This experiment tested for differences in willpower depletion among athletes and nonathletes. Specifically, we hypothesized that the willpower reserves in athletes would drain less quickly when subjected to a prior mentally/emotionally strenuous task. We estimated the following regression models in an attempt to progressively examine the degree to which athletes exhibit differences in the outcome variable:

(1) Time = [[alpha].sub.0] + [[beta].sub.1] x Treatment + [gamma]X + [epsilon]

(2) Time = [[alpha].sub.0] + [[beta].sub.2] x Athlete + [gamma]X + [epsilon]

(3) Time = [[alpha].sub.0] + [[beta].sub.1] X Treatment+ [[beta].sub.2] x Athlete + [[beta].sub.3] (Treament x Athlete) + [gamma]X + [epsilon]

The dependent variable, "Time," is measured in seconds that the subject attempts the undoable puzzle (zero to 1,800 s). "Treatment" is a zero-one indicator that equals one if the subject received the instruction to resist responding emotionally to the videos. "Athlete" is a zero-one indicator that equals one if the individual was an athlete in his/her senior year in high school and/or is an intercollegiate athlete at the university. A- is a vector of individual controls, including "Female," a zero-one indicator that equals one if the participant is a female, 0 if male; "Work," a zero-one indicator that equals one if the participant works during the semester; "Hours Worked" are the number of hours the week conditional on whether or not the participant answered "yes" to working; "Hours UW-L Sports" is a continues variable indicating how many total hours per week an individual is dedicated to sports at UWL (note, this is "zero" for the nonathletes). This variable is one measure of intensity associated with participation in athletics. (17) "Last Meal" is the number of hours since the participant's last meal in order to attempt to control for the effect of glucose on individual willpower reserves; (18) "Non-Freshman" = I if the participant was not a freshman and zero otherwise; "Video 1" and "Video2" report the participant's response to a Likert scale (one to ten) on the questionnaire regarding the degree to which the participant found the videos funny. We also include a robust standard error term, [epsilon]. (19)

We provide two tables of summary statistics. Table 2 focuses on the variables related to our hypotheses and allows us to perform an uncontrolled analysis of differences in the group means on time spent working on the test, while Table 3 provides summary statistics for all of the control variables in our regression models. With regard to Table 2, it is notable that our participants spent a remarkable amount of time working on the puzzle, an average of nearly 23 minutes. In previous studies, researchers found only a small minority of their participants that were willing to stay the maximum time to work on the challenging puzzle. Baumeister et al. (1998) found that only 4 of their 67 participants (6.0%) stayed for the maximum time of 30 minutes. Nearly half of our participants (37 of 75) had to be stopped from working after 30 minutes. (20)

Comparing the means across categories, it is clear that athletes spend significantly more time--over 4 minutes--on the test than do nonathletes. Similarly, and as expected based on the literature on willpower drain, those in the treated group spend over 4 1/2 fewer minutes on the test than do those in the control group. Males and females spend nearly the exact same amount of time working on the test. (21)

Our hypothesis that the treated athlete would experience less willpower drain is also in evidence here as the treated athletes spent nearly 8 minutes longer working on the puzzle than did nonathletes in the treatment group. Athletes in the control group, however, did not work on the test for significantly longer than did nonathletes in the control group. The puzzle was difficult, but under normal conditions in which participants had full banks of willpower, one may expect that essentially all college-level students would be willing and able to work on a difficult task for 30 minutes. A typical college exam, for example, will take an hour to complete. The major differences in time spent on the puzzle occurred only after the willpower-draining task was administered. This supports previous findings in the willpower literature. Oaten and Cheng (2006a, 2006b) conclude that individuals can improve willpower stamina through activities such as exercise, but not necessarily increase their capacity for willpower.

Table 3 provides summary statistics for the dependent and independent variables in our regression analyses. Within the sample, there are 43 women and 32 men. Of the women in the sample, 20 are athletes while 26 men are athletes. In the sample, 27 individuals work an average of 12 hours per week. Of those that work in the sample, 15 are athletes who work slightly more than 12 hours per week. (22) On average, most of the participants in the study had eaten within 3 hours, with two outliers that had not eaten within 15 hours of the study. Half of the study participants were in their first year of college.

B. Results

Table 4 presents the results from our regression models that include controls for various exogenous factors that may affect time spent on the puzzle. Model 1 provides evidence of the ego depletion effect. Individuals in the treatment group spent significantly less time on the puzzle (almost 4 minutes less) as compared to those who were allowed to respond emotionally to the videos, controlling for all other relevant variables. No other variables had an effect significantly different from zero when predicting the amount of time that was spent on the puzzle, but the signs on coefficients were not unexpected. For example, the working dummy had a negative effect on time spent working on the puzzle. The length of time to the last meal had a negative effect on time spent. This may support (although not statistically significantly so) the glucose effect on willpower documented by Gailliot et al. (2007) and Gailliot (2012). The sign on the Video 1 coefficient was also negative, indicating the degree to which an individual found the video funny reduced the amount of time spent on the puzzle. This may reflect the intensity of self-control expended in the prior task, and its effect on willpower depletion. Intuitively, the more self-control required in the prior task may lead to quicker willpower depletion in follow-up tasks.

Model 2 included the athlete indicator but dropped the treatment indicator in order to test the effect of being an athlete by itself on time spent on the puzzle. The results indicate that athletes spend significantly more time than nonathletes on the puzzle (over 5 minutes more), after controlling for the exogenous factors. This model does not, however, take into consideration the treatment effect, so while it does establish a significant difference between athletes and nonathletes on time spent on the puzzle, it tells us little about differences between athletes and nonathletes in ego depletion. Videol is negative and significant in this model. This means that, controlling for the exogenous factors as well as being an athlete, watching Videol, whether the participant was in the treatment or the control group, resulted in less time spent on the task. This warrants further exploration in Model 3 to see if the effects are driven by the treatment group's attempt at controlling their emotion to the video.

While Models 1 and 2 illustrate differences in the group means, we are more focused on analyzing the interaction between athletes and the treatment effect. Specifically, are there differences in time spent on the task between athletes and nonathletes within the treatment and control conditions? Models 1 and 2 suggest that a better model would include a difference-in-difference (DID) term to be able to compare the effect of the treatment on athletes versus the treatment on nonathletes and to be able to test our hypothesis that athletes will experience less ego depletion than nonathletes.

Model 3 performs the DID analysis by adding the interaction term. Treatment X Athlete. With the interaction variable, the marginal effects of Athlete on Time now becomes,

(4) [partial derivative]Time/[partial derivative]Athlete = [[beta].sub.2] + [[beta].sub.3] X Treatment.

We hypothesized [[beta].sub.3] would be positive. If [[beta].sub.3] > 0, athletes spent more time on the puzzle in the treatment condition. That is, athletes would have been more resistant to the willpowerdraining task of restraining emotions while watching the funny video.

The final column of Table 4 presents the results from Model 3. The coefficient on the athlete-treatment interaction term is positive and statistically significant. This finding supports our hypothesis that athletes would be more resistant to the willpower-draining task. Athletes stayed for over 7 more minutes (435 s) longer, on average, in the treatment condition than nonathletes in the treatment condition.

The coefficient on Videol was negative and significant, indicating that as one's reaction to Videol increased, time spent on the subsequent willpower task fell. Perhaps, any reaction to a funny video--whether that be laughing hysterically, or attempting to stifle one's laugher--would have a willpower-draining effect. The willpower literature suggests an intuitive explanation that as one expends more energy on a self-control task, less willpower remains to perform a follow-up task. It is unclear why there was no effect from Video2, particularly because the mean evaluation for Video2 is higher than Videol, indicating that participants found it to be funnier. (23) A robustness check was run to test this hypothesis. Table SI in the Supporting Information provides results from two regression models, A1 and A2. The first model built from Model 3 above and included an interaction term between Video 1 and Treatment. Including this interaction variable, Treatment became insignificant, but Video 1 and the DID (the interaction between Treatment and Athlete) remained significant. The interaction term, however, was not significant. This indicates that it is not the link between watching the video and the treatment condition that is prompting the effect. In Model A2, we incorporated an interaction between Athlete and Video 1. In this model, we return to our original results in which the Treatment effect as well as the DID are significant. (24)

C. Discussion

Our findings support our hypothesis that athletes deplete their willpower reserves less quickly than nonathletes on tasks subsequent to prior task requiring expending willpower or self-control. Because the experimental design allowed us to randomly assign individuals to the treatment group, we can also infer a causal relationship between being an athlete, and a smaller willpower drain in the treatment group as compared to nonathletes in the treatment group. We cannot, however, make conclusions as to why this is the case. Athletes were not randomly selected into being athletes in our sample. Do athletes differ in some systematic way from nonathletes that induced selection into athletics or did participation in athletics itself foster skills that would aid in resisting willpower drain? It is quite possible that only individuals with greater reserves of willpower join and remain in athletics. As athletics becomes more serious, difficult, and selective, only the participants with the most willpower may continue playing. Thus, athletics may act as a filtering mechanism. (25)

Despite the fact that we cannot provide evidence that participation in athletic causes greater levels of willpower, our results are an important contribution to the literature on the athletic earnings premium. Willpower has been documented to be an important noncognitive skill that is associated with success in the labor market. Our research illustrated that athletes are better able to withstand willpower depletion than nonathletes.

Regardless of the direction of causality between athletic participation and willpower development, persistent athletic participation can provide a signal to employers that athletes possess strong noncognitive skills. This cannot fully identify the source of the labor market premiums rewarded to athletes. Athletes are likely in better health and as such, more attached to the labor market with fewer absences. Athletes may also possess other skills rewarded in labor markets and may develop better job market skills through athletic participation. Several explanations for the labor market premium exist, however it is clear that athletes are less likely to experience a willpower drain after prior expenditure of selfcontrol. Such noncognitive skills may translate directly or indirectly into higher earnings as athletes may be better able to persevere through work tasks.

D. Limitations and Extensions

We recognize the limitations with our study. First, the external validity of our findings is limited due to our sample population. We recruited student participants from a single public university in the Midwest of the United States, so we cannot claim to generalize our results to the general population. This, however, is the nature of experiments. They are costly and tend to have small, nongeneralizable samples.

Second, we are marginally underpowered with our sample size. The primary consequence of being underpowered is an increase in the likelihood of type II error that it is more difficult to observe an effect when one is actually present. Therefore, we may be missing a significant effect present in our independent variables, but because we obtain statistical significance on our primary variable of interest--the interaction between athlete and treatment--we are more convinced we can reject the null.

Third, we do not have a direct control measure for intelligence or other measurable cognitive skills. While much of the psychology literature from which we based our procedure does not control for intelligence, intelligence may be related to performance in the puzzle creating bias in our results. Most notably, an intelligent participant may have realized that the puzzle was unsolvable and stopped working.

To this concern, we offer two comments. Consider two college success measures: GPA and graduation rate. A relative consensus in the literature identifies that college athletes have higher graduation rates than nonathletes. While the average Division-I scholarship athlete's GPA is moderately lower than nonathletes, that gap shrinks when comparing walk-ons to nonathletes and no study has found a significant difference among Division-Ill athletes. All athletes in our study were non-scholarship athletes because UW-L participates in Division-Ill athletics.

Fourth, our results would be further validated with a measure of the intensity in which the athlete participates in a given sports activity. Kosteas (2012) finds that simply engaging in regular exercise results in a 6%--10% wage premium and more frequent exercise generates an even larger impact. A follow-up study would question participants (athletes and nonathletes alike) on their leisure exercise activities as well as the intensity of such activities as well as any formal participation in sports.

In terms of extensions, our hypothesis relies on the general understanding that being an elite athlete requires countless hours of training and determination. We hypothesized that such training would develop in the athlete many noncognitive skills such as determination, endurance, and willpower that would lead to success in the labor market. Athletic training is not the only avenue to such noncognitive skill development. Other extracurricular activities such as becoming an accomplished musician may have a similar effect. Future studies might include participants with other skills that required many years of dedication to develop.


Noncognitive skills such as willpower and self-control have been found to be at least equally important predictors of life and career outcomes as traditional "hard skills," including academic performance and IQ (Duckworth and Seligman 2005; Heckman, Stixrud. and Urzua 2006; and Mischel, Shoda. and Peake 1988). Such skills may be the key factor in explaining the earnings premium among athletes (Stevenson 2010). In this study, we sought to identify differences in willpower depletion among athletes and nonathletes.

We hypothesized that athletes who were able to persist through years development and training would have developed deeper reserves of willpower and/or would experience less willpower depletion when put through a task designed to drain willpower. To test this hypothesis, we conducted an experiment, closely mirroring the procedure of Muraven, Tice, Baumeister (1998). Utilizing a DID regression analysis, we found that athletes were significantly less affected by the willpower-draining task of refraining from laughing during the funny videos. We found no statistical difference in performance between athletes and nonathletes in the control group, but athletes in the treatment group worked on the puzzle for 7 more minutes than nonathletes. We conclude that while athletic participation does not increase an individual's willpower capacity, it does improve willpower stamina.

doi: 10.1111/coep.12150


Barron, J. M., B. T. Ewing, and G. R. Waddell. "The Effects of High School Athletic Participation on Education and Labor Market Outcomes." Review of Economic Statistics, 82(3), 2000, 409-21.

Baumeister, R. F., T. F. Heatherton, and D. M. Tice. Losing Control: How and Why People Fail at Self-Regulation. San Diego, CA: Academic Press, 1994.

Baumeister, R. F., E. Bratslavsky, M. Muraven, and D. Tice. "Ego Depletion: Is the Active Self a Limited Resource?" Journal of Personality and Social Psychology, 74(5), 1998, 1252-65.

Baumeister, R. F, M. Gailliot, C. N. De Wall, and M. Oaten. "Self-Regulation and Personality: How Interventions Increase Regulatory Success, and How Depletion Moderates the Effects of Traits on Behavior." Journal of Personality, 74, 2006, 1773-801.

Baumeister, R. F, K. D. Vohs, and D. M. Tice. "The Strength Model of Self-Control." Current Directions in Psychological Science, 16(6), 2007, 351-55.

Baumeister, R. F, and J. Tierney. Willpower: Rediscovering the Greatest Human Strength. New York: Penguin Books, 2011.

Cabane, C., and M. Lechner. "Physical Activity of Adults: A Survey of Correlates, Determinants, and Effects." Economics Working Paper Series 1428, University of St. Gallen, School of Economics and Political Science, 2014.

Costa, P. T. Jr., and R. McCrae. Revised NEO Personality Inventory (NEO-PI-R) and NEO Five-Factor Inventory (NEO-FFE) Professional Manual Odessa, FL: Psychological Assessment Resources, 1992.

Duckworth, A. L., and M. Seligman. "Self-Discipline Outdoes IQ in Predicting Academic Performance of Adolescents." Psychological Science, 16(12), 2005, 939-44.

Erdfelder, E., F. Faul, and A. Buchner. "GPOWER: A General Power Analysis Program." Behavior Research Methods, Instruments, & Computers, 28(1), 1996, 1-11.

Ewing, B. T. "Athletes and Work." Economics Letters, 59, 1998, 113-17.

--. "The Labor Market Effects of High School Athletic Participation: Evidence from Wage and Fringe Benefit Differentials." Journal of Sports Economics, 8, 2007, 255-65.

Feather, N. T. "The Relationship of Persistence at a Task to Expectation of Success and Achievement Related Motives." Journal of Abnormal and Social Psychology, 63, 1961, 552-61.

Gailliot, M. T. "Improved Self-Control Associated with Using Relatively Large Amounts of Glucose: Learning Self-Control Is Metabolically Expensive." Psychology, 3(11), 2012, 987-90.

Gailliot, M. T., R. F. Baumeister, C. N. DeWall, J. K. Maner, E. A. Plant, D. M. Tice, L. E. Brewer, and B. J. Schmeichel. "Self-Control Relies on Glucose as a Limited Energy Source: Willpower Is More Than a Metaphor." Journal of Personality and Social Psychology, 92(2), 2007, 325-36.

Glass, D. C, J. E. Singer, and L. N. Friedman. "Psychic Cost of Adaptation to an Environmental Stressor." Journal of Personality and Social Psychology, 12, 1969, 200-10.

Gosling, S. D., P. J. Rentfrow, and W. B. Swann Jr. "A Very Brief Measure of the Big Five Personality Domains." Journal of Research in Personality, 37, 2003, 504-28.

Heckman, J. J., and Y. Rubinstein. "Importance of Noncognitive Skills: Lessons from the GED Testing Program." American Economic Review, 91(2), 2001, 145-9.

Heckman, J. J., J. Stixrud, and S. Urzua. "The Effects of Cognitive and Noncognitive Abilities on Labor Market Outcomes and Social Behavior." Journal of Labor Economics, 24(3), 2006, 411-82.

Heckman, J. J., and T. Kautz. "Hard Evidence on Soft Skills." Labour Economics, 19(4), 2012, 451-64.

Hyytinen, A., and J. Lahtonen. "The Effect of Physical Activity on Long-Term Income." Social Science & Medicine, 96, 2013, 129-37.

Koebler, J. "High School Sports Participation Increases for 22nd Straight Year." U.S. News & World Report, 2011. Accessed September 2, 2011. http://www.usnews. com/education/blogs/high- school- notes/2011 /09/02/ high-school-sports-participation-increases-for-22ndstraight-year.

Kosteas, V. "The Effect of Exercise on Earnings: Evidence from the NLSY." Journal of Labor Research, 33, 2012, 225-50.

Lechner, M. "Long-Run Labor Market and Health Effects of Individual Sports Activities." Journal of Health Economics, 28, 2009, 839-54.

Lechner, M., and P. Downward. "Heterogeneous Sports Participation and Labour Market Outcomes in England." Discussion Paper No. 2013-25, University of St. Gallen, 2013. Accessed May 21, 2014. http://wwwl.vwa.umsg. ch/RePEc/usg/econwp/EWP- 1325.pdf

Lechner, M., and N. Sari. "Labor Market Effects of Sports and Exercise: Evidence from Canadian Panel Data." IZA DP No. 7931, 2014. Accessed May 21.2014. http://ftp.iza. org/dp7931 .pdf

Long, J. E., and S. B. Caudill. "The Impact of Participation in Intercollegiate Athletics on Income and Graduation." Review of Economics and Statistics, 73, 1991,525 -31.

McCormick, R. E., and M. Tinsley. "Athletics Versus Academics? Evidence from SAT Scores." Journal of Political Economy, 95(5), 1987, 1103-16.

Mischel, W. "Preference for Delayed Reinforcement: An Experimental Study of a Cultural Observation." The Journal of Abnormal and Social Psychology, 56, 1958, 57-61.

--. "Processes in Delay of Gratification," in Advances in Experimental Social Psychology, Vol. 7, edited by L. Berkowitz. New York: Academic Press, 1974, 249-92.

Mischel, W., Y. Shoda, and P. K. Peake. "The Nature of Adolescent Competencies Predicted by Preschool Delay of Gratification." Journal of Personality and Social Psychology, 54, 19X8, 687-96,

Muraven, M., R. F. Baumeister, and D. M. Tice. "Longitudinal Improvement of Self-Regulation Through Practice: Building Self-Control Through Repeated Exercise." Journal of Social Psychology, 139, 1999,446-57.

Muraven. M.. D. M. Tice, and R. F. Baumeister. "Self-Control as Limited Resource: Regulatory Depletion Patterns." Journal of Personality and Social Psychology, 74, 1998, 774-89.

Muraven, M., and R. F Baumeister. "Self-Regulation and Depletion of Limited Resources: Does Self-Control Resemble a Muscle?" Psychological Bulletin, 126(2), 2000, 247-59.

Oaten, M., and K. Cheng. "Longitudinal Gains in SelfRegulation from Regular Physical Exercise." British Journal of Health Psychology, 11(4), 2006a, 717-33.

--. "Improvements in Self-Control from Financial Monitoring." Journal of Economic Psychology, 28(4), 2006b, 487-501.

Pocheptsova, A., O. Amir, R. Dhar, and R. F. Baumeister. "Deciding without Resources: Resource Depletion and Choice in Context." Journal of Marketing Research, 46(3), 2009, 344-55.

Rehberg, R., and W. E. Schafer. "Participation in Interscholastic Athletics and College Expectations." American Journal of Sociology, 73(6), 1968, 732-40.

Rooth, D. "Work Out or Out of Work--The Labor Market Return to Physical Fitness and Leisure Sports Activities." Labour Economics, 18(3), 2011, 399-409.

Schmeichel, B. J.. K. D. Vohs, and R. F. Baumeister. "Intellectual Performance and Ego Depletion: Role of The Self in Logical Reasoning and Other Information Procession." Journal of Personality and Social Psychology, 85, 2003, 33-46.

Shoda, Y, W. Mischel, and P. K. Peake. "Predicting Adolescent Cognitive and Self-Regulatory Competencies from Preschool Delay of Gratification." Developmental Psychology, 26, 1990, 978-86.

Spreitzer, E., and M. Pugh. "Interscholastic Athletics and Educational Expectations." Sociology of Education, 46(2), 1973, 171-82.

Stevenson. B. A. "Beyond the Classroom: Using Title IX to Measure the Return to High School Sports." Review of Economics and Statistics, 92(2), 2010, 284-301.

Tangney, J. P. R. F. Baumeister, and A. L. Boone. "High Self-Control Predicts Good Adjustment, Less Pathology, Better Grades, and Interpersonal Success." Journal of Personality, 72, 2004, 271-322.

Vohs, K. D., and T. F. Heatherton. "Self-Regulatory Failure: A Resource-Depletion Approach." Psychological Science, 11,2000,249-54.

Vohs, K. D., B. D. Glass, W. T. Maddox, and A. B. Markman. "Ego Depletion Is Not Just Fatigue: Evidence from a Total Sleep Deprivation Experiment." Social Psychological and Personality Science, 18(2), 2011, 166-73.

Wegner, D. M., R. Erber, and S. Zanakos. "Ironic Processes in the Mental Control of Mood and Mood Related Thought." Journal of Personality and Social Psychology, 65, 1993, 1093-107.


Additional Supporting Information may be found in the online version of this article:

Table S1. Videol interactions

Table S2. Female interactions


Hoffer: Assistant Professor, Department of Economics, University of Wisconsin-La Crosse, La Crosse, WI 54601. Phone 608-785-5293, E-mail

Giddings: Associate Professor, Department of Economics, University of Wisconsin-La Crosse, La Crosse, Wl 54601. Phone 608-785-5297, E-mail


ACT: American College Testing

DID: Difference-in-Difference

GPA: Grade Point Average

NCAA: National College Athletic Association

UW-L: University of Wisconsin-La Crosse

(1.) Willpower, along with other "non-cognitive" skills such as tenacity, resilience, teamwork, and self-control, is proving to be at least as important as cognitive skills in predicting positive outcomes in life (Heckman and Kautz 2012; Heckman and Rubinstein 2001; Heckman, Stixrud, and Urzua 2006).

(2.) For the purpose of our study, we will use the terms willpower and self-control interchangeably. The definition of willpower, according to Merriam-Webster is "the ability to control yourself."

(3.) Title IX is a law enacted in 1972 which states in part that "No person in the United States shall, on the basis of sex, be excluded from participation in, be denied the benefits of, or be subjected to discrimination under any education program or activity receiving federal financial assistance."

(4.) Since Mischel's work, self-control (or the lack thereof) has been linked to a broad spectrum of behaviors. High scores on self-control have been correlated with higher grade point averages, higher self-esteem and fewer reports of psychopathology, less binge eating and drinking, better relationships and personal skills, secure attachments, and more optimal emotional responses (Mischel etal. 1988; Shoda, Mischel, and Peake 1990; Tangney, Baumeister, and Boone 2004). Adults with more willpower are less likely to abuse drugs and are even liked more by their peers (Baumeister and Tierney 2011).

(5.) All experimental documents, including recruiting fliers, questionnaires, and assessment tasks are provided in the experimental materials in the Supporting Information.

(6.) Prior research has established that attempting to regulate one's emotions is effortful and "requires exertion to succeed" (Wegner, Erber, and Zanakos 1993).

(7.) Participants were informed that the puzzle would be challenging, but they were not informed that the puzzle would be unsolvable.

(8.) The $ 10 payment provided no incentive to work harder or participate longer on the task. Because intercollegiate athletes at UW-L do not receive scholarships, we can assume that there is no a priori difference in opportunity cost between athletes and nonathletes. See Table 3 for summary statistics on the variables in the model. Athletes do not work for pay significantly more hours than nonathletes in our sample. On average, athletes work a little over 12 hours per week while nonathletes work slightly over 11 hours per week. Similarly, there is no significant difference in hours worked per week between the treatment and control groups.

(9.) Relying on undergraduates as study participants is common in experiments in both the psychology and economic literature. The four experiments documented in Muraven, Tice, and Baumeister (1998) drew on data collected from between 60 and 86 undergraduate students in introductory psychology courses at Case Western Reserve University. Oaten and Cheng (2006a. 2006b) collected data from 24 undergraduates across Macquarie University.

(10.) We determined the number of subjects using an a priori power analysis, conducted with the G*Power program (Erdfelder, Faul, and Buchner 1996) (alpha = .05, Power (1-Beta) = .80). We designed this study to allow for an easier future longitudinal study. We plan to revisit this study when the participants are in their final year of college to evaluate if their willpower reserves change over their years in college. Therefore, we selected only freshmen and sophomore participants to allow a minimum of 2 years (ideally 3-4 years) between initial and final testing for a follow-up study.

(11.) The ACT test is a standardized college readiness exam in the United States used by many college admissions offices. Although statistics vary from year to year, the national average hovers around 20 with an SD of 6.

(12.) The National College Athletic Association (NCAA) consists of three divisions based both on the size of the school and the priority placed on athletics, Division-I. DivisionII, and Division-Ill. Division-I is the most elite in terms of the level and intensity of competition. For Division-Ill, the division in which UW-L competes, academics are the primary focus for student-athletes, but they still compete with other colleges and universities. NCAA regulations make Division-Ill athletics less strenuous and binding as compared to Division-1. For example, a 2008 NCAA survey found that Division-I athletes devoted 44.8 hours per week to athletic responsibilities and approximately 40 hours per week to academic life. The intercollegiate athletes in our sample reported spending approximately 12 hours per week on athletics. One of the main differences between the Division-I and DivisionIII athlete is that athletics-based financial aid is prohibited in Division-111.

(13.) Intramural sports are structured leagues, tournaments, and activities specifically designed for recreational and competitive opportunities for students at the university. Teams compete with other teams within the university. Over 20 different sports are offered at UW-L.

(14.) For example, in the 2013-2014 school year there were 35,393 boys playing high school hockey, 4,360 men playing in college, or 12%. On average, across all sports during that year, nearly 4.5 million boys played a sport in high school, whereas 341,530 played in college, or 7.6%. Only 5% of UW-L students are intercollegiate athletes.

(15.) Note that almost by definition, an intercollegiate athlete would have played varsity sports in high school. In our sample, all of the UW-L intercollegiate athletes were also varsity players in their senior year in high school. See Table 1.

(16.) Note that the main difference between an individual who played a varsity sport during their senior year in high school and an individual who went on to the intercollegiate Division-Ill level is ability or talent. We can assume that intercollegiate athletes in their freshman year have faced a similar level of intensity over the course of their athletic career as those that competed in their senior year in high school. As intercollegiate participation goes on into the years beyond their freshman year, however, we would expect the high school athlete and the college athlete to diverge dramatically in terms of the level of intensity in which they participate in the sport. To mitigate this, we included a dummy variable in the empirical analysis for "non-freshmen."

(17.) Our study is limited in its measure of intensity of athletic participation. Hours spent at UW-L sports ranges from 4 to 25 hours per week.

(18.) Gailliot et al. (2007) found that acts of self-control cause a reduction in blood-glucose levels. If one consumes sugary drinks after a task involving willpower, they are less likely to do poorly on subsequent tasks requiring selfcontrol. Gailliot (2012) showed that the greater the selfcontrol required for a task, the greater the body metabolized glucose. This body of research shows that replacing the body's stores of glucose (i.e., getting a snack) can improve one's subsequent willpower.

(19.) It is important to note that we do not have controls for socioeconomic status in our study, however, athletes at UW-L are not paid, and do not receive scholarships. As will be shown in Table 3, athletes are just as likely to be working as nonathletes, and even tend to work more hours per week. We can assume that there is no systematic difference in opportunity cost faced by athletes versus nonathletes and those in the treatment versus control groups. Furthermore, we have little difference in our sample along the lines of race and ethnicity. Additionally, there is only one non-Caucasian individual in the sample.

(20.) Note that 70%, or 26 of those that spent over 28 minutes on the test were athletes.

(21.) Although not presented here, it did not matter whether males and females were in the treatment or control group, they spent the same amount of time working on the puzzle.

(22.) Note that working athletes work slightly more than working nonathletes (12.63 hours versus 11.21 hours per week).

(23.) The mean for Video! was 6.69 while the mean for Video2 was 7.07. A /-test of their means suggests that participants found them to be equally funny.

(24.) We ran similar robustness checks on the female variable despite our results that females do not differ from males in any significant manner, and the results are presented for Models A3 and A4 in Table S2. An interaction term between Female and the Treatment resulted in nothing significant in the model. An interaction term between Female and Athlete, however, returned our original results with the Treatment effect being significant as well as the DID being significant.

(25.) We have gathered data on whether participants were athletes at any point in high school, athletes through their senior year of high school, and whether participants plan on participating in college. In future research, we hope to recruit these same participants during their final year of college, reexamine willpower, and analyze continued athletic participation with willpower measures at different time points.

Participant Athletic Participation

Athletic Participation                       N    N Males    N Females

Total number of participants                75       32          43
Participated in athletics during senior     46       26          20
   year in high school
Participated in athletics in high school    62       30          32
   at all
Never participated in athletics             13       2           11
Participates in intercollegiate athletics   16       9           7
   at UW-L


Mean Time Worked on Puzzle by Participant Group

Category                   N        Mean (SD)          SE        t

All participants           75   1377.81 (523.74)
Athlete status
 Athlete **                46   1482.283 (475.86)   104.1653   -2.23
 Nonathlete                29    1212.10(560.95)     70.16
Treatment group
 Treated **                41   1251.02 (585.60)     91.46     2.37
 Control                   34   1530.71 (394.18)     67.60
 Female                    43     1379.00 (540)      82.35     -0.02
 Male                      32   1376.22 (509.63)     90.09
Treatment and athlete
 Treated athlete **        25   1443.28 (527.98)     105.60    -2.85
 Treated nonathlete        16   950.625 (557.32)     139.33
Control and athlete
 Controlled athlete        21   1528.71 (413.29)     90.19     0.037
 Controlled nonathlete     13   1533.92 (377.61)     104.73
Within athlete
 Treated athlete           25   1443.28 (527.98)     105.60    0.602
 Controlled athlete        21   1528.71 (413.29)     90.19
Within nonathlete
 Treated nonathlete **     16    950.63 (557.32)     139.33    3.216
 Controlled nonathlete     13   1533.92 (377.61)     104.73

Note: **A statistically significant difference at the 0.05 level in the
mean duration spent on solving the puzzle between the two groups.


Summary Statistics

                     All    Athlete   Nonathlete   Treatment   Control

Time              1377.81   1482.28     1212.10     1251.02    1530.71
                  (523)     (475)     (560)        (5850)      (394)
Female              0.57      0.43         0.79        0.56      0.59
                  (0.49)    (0.50)       (0.41)      (0.50)    (0.50)
Work                0.36      0.33         0.41        0.39      0.32
                  (0.48)    (0.47)       (0.50)      (0.49)    (0.47)
Hours worked          12     12.63        11.21       12.41     11.41
    per week
   (conditional   (5.71)    (4.06)       (7.41)      (5.69)    (5.97)
   on working)
Hours UWL           2.73      4.46            0           3      2.41
   sports         (6.28)    (7.56)                   (6.39)    (6.24)
Last meal           3.52      3.34         3.82        4.29      2.60
                  (3.71)    (3.64)       (3.86)      (4.22)    (2.76)
Non-Freshman        0.49      0.50         0.48        0.63      0.33
                  (0.50)    (0.50)       (0.51)      (0.49)    (0.47)
Video 1             6.69      6.78         6.55        6.83      6.53
                  (1.74)     (1.6)       (1.97)       (L72)    (1.78)
Video2              7.07      7.33         6.66        6.93      7.24
                  (1.90)    (1.76)       (2.06)      (2.07)    (1.69)
N                     75        46           29          41        34

Note: SDs are in parentheses.

"Note that the hours spent on athletics at UW-L is fairly small among
all athletes. If we look at only those individuals participating in
intercollegiate sports at UW-L, the average number of hours per week
on sports increases to 12.8 with a SD of 7.55.


Regression Results

                    Model 1     Model 2     Model 3

Treatment          -225.89 *               -492.85 **
                   (123.04)                 (186.61)

Athlete                        334.32 **     103.00
                               (143.98)     (145.61)

Treatment x                                 434.63 *
   Athlete                                  (242.41)

Female              -46.38       91.52       92.44
                   (140.16)    (154.97)     (141.50)

Work                -48.60       -9.93       -35.30
                   (225.35)    (236.03)     (206.93)

Hours worked         11.93       2.10         1.24
   per week         (16.96)     (16.34)     (13.06)

Hours UWL sports     4.26        -6.43       -7.25
                    (10.54)     (10.73)     (10.52)

Last meal           -18.23      -15.86       -8.10
                    (18.29)     (17.35)     (16.43)

Non-Freshman        -289.67     -396.31     -289.77
                   (422.27)    (352.77)     (344.09)

Video 1             -52.95     -74.28 **    -58.27 *
                    (33.32)     (33.62)     (31.40)

Video2               23.49       2.94        -6.37
                    (35.58)     (36.14)     (34.81)

R2                   .1364       .1584       .2373
Observations          75          75           75

Note: Robust standard errors in parentheses. p< .10, * p < .05,
** p < .01 ***.
COPYRIGHT 2016 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2016 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Hoffer, Adam; Giddings, Lisa
Publication:Contemporary Economic Policy
Article Type:Report
Geographic Code:1USA
Date:Jul 1, 2016
Previous Article:Willingness-to-pay for sporting success of football Bundesliga teams.
Next Article:The (aggregate) demand for state-lottery tickets: what have we really learned?

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters