research paper on sports betting

• Follow us on Facebook
• Follow us on Twitter
• Criminal Justice
• Environment
• Politics & Government
• Race & Gender

Expert Commentary

Sports betting in the US: A research roundup and explainer

We look at the landscape of legal sports betting in America, explain what the research says about how legalization affects tax revenues, and provide a brief history of the activity.

Republish this article

This work is licensed under a Creative Commons Attribution-NoDerivatives 4.0 International License .

by Clark Merrefield, The Journalist's Resource October 25, 2022

This <a target="_blank" href="https://journalistsresource.org/economics/sports-betting-research-roundup-explainer/">article</a> first appeared on <a target="_blank" href="https://journalistsresource.org">The Journalist's Resource</a> and is republished here under a Creative Commons license.<img src="https://journalistsresource.org/wp-content/uploads/2020/11/cropped-jr-favicon-150x150.png" style="width:1em;height:1em;margin-left:10px;">

On Nov. 8, Californians will vote on two ballot measures that would allow for different forms of sports betting in the state.

Proposition 26 would allow sports betting at licensed casinos and horse tracks on tribal lands and run by federally recognized Native American tribes.

Prop. 27 would allow tribes licensed to offer gambling and major gaming companies to offer online sports betting. These companies include FanDuel and DraftKings, which together make up roughly two-thirds of the U.S. online sports betting market.

“If both pass, they might both go into effect or the result could be decided in court, depending on which one gets more yes votes,” writes CalMatters economics reporter Grace Gedye in an article from June.

Although California is the only state with sports betting on the midterm ballot , it’s not the only state where sports betting is a topic of political discussion — and relevant for journalists across beats to understand. For example, Georgia gubernatorial candidate Stacey Abrams recently expressed support for legalized sports betting in her state. Abrams’ opponent, Gov. Brian Kemp, is opposed.

In Missouri, state lawmakers from both parties support legalizing sports betting , but Gov. Mike Parson is hesitant. Vermont lawmakers are considering taking up a sports betting bill during the next legislative session. Gubernatorial candidates in South Carolina and Texas support legal sports betting. In Florida, there is an ongoing lawsuit over whether the state should be allowed to give the Seminole Tribe the exclusive right to run online sports betting there.

Legal sports wagering in the U.S. has grown vertically in recent years — from less than $5 billion worth of bets placed in 2018 to $57 billion in 2021 — despite sports betting remaining illegal in nearly half of states. Sportsbooks, the entities that take sports bets, bring in about $4 billion yearly after wagers are settled.

The reason for this growth: a May 2018 Supreme Court ruling. Justice Samuel Alito, in delivering the 6-3 decision , reasoned that 1992 federal legislation banning states from allowing sports betting was unconstitutional.

Under the 1992 law, the Professional and Amateur Sports Protection Act, the federal government did not “make sports gambling itself federal crime,” Alito writes in the 2018 decision. “Instead, it allows the [U.S.] Attorney General, as well as professional and amateur sports organizations, to bring civil actions to enjoin violations.” Other than legislative powers the Constitution grants Congress, the federal government cannot “issue direct orders to state legislatures,” Alito writes. The majority interpreted the Professional and Amateur Sports Protection Act as doing so.

Here is how John Holden , an assistant professor of business at Oklahoma State University who has written extensively on sports gambling, explains the 2018 ruling:

“If the federal government wants to make sports betting illegal, they’re free to do so, but they can use the [Federal Bureau of Investigation] and the [Department of Justice] to enforce that,” Holden says. “They can’t tell a state legislature that you need to keep that law on the books and use your state police to go out and bust up gambling rings.”

A fundraising breakdown from the Los Angeles Times shows about $132 million has been raised to support Prop. 26, with about $43 million in opposition funding. Top backers include the Federated Indians of Graton Rancheria, the Pechanga Band of Indians and the Yocha Dehe Wintun Nation. Non-Native American casino and gaming interests are largely opposed — conversely, they have backed Prop. 27, which would open up sports betting to all gambling interests, not just Native American-run casinos.

Tribal gaming brings in nearly $40 billion a year across all tribes that operate gambling enterprises, according to the National Indian Gaming Commission . “Gaming operations have had a far-reaching and transformative effect on American Indian reservations and their economies,” write the authors of a 2015 paper about how the act affected tribal economic development. “Specifically, Indian gaming has allowed marked improvements in several important dimensions of reservation life.”

The landscape of legal sports betting

If California legalizes sports betting, it would represent a major coup for gaming interests in the state. In California, the most populous state, horse racing is the only legal form of sports betting.

Sports wagering is legal in 28 states plus the District of Columbia, according to a recent Washington Post analysis. Seven states prohibit online sports betting and only allow in-person wagers at licensed locations, such as casinos: Delaware, Maryland, Mississippi, Montana, New Mexico, North Carolina and North Dakota. Sports betting is legal but pending rollout in four states: Maine, Massachusetts, Nebraska and Ohio. Kansas is the most recent state to implement legal in-person and online sports gambling, as of Sept. 1.

States place a range of licensing fees on operators and tax rates on sports betting revenue, from a low of 6.75% in Nevada and Iowa to a high of 51% in New Hampshire and New York. States use tax revenues for a variety of purposes . Some, like Delaware, put sports wagering taxes toward their general fund. Colorado uses sports betting taxes to pay for its statewide water plan, Illinois funds transportation infrastructure and New York funds education programs.

In states where sports betting is legal, bettors can wager on nearly any major sporting event, both professional and amateur. For example, bettors can wager on the outcome of a baseball game, as well as events within the game, such as whether a particular player will hit a home run.

Polling indicates California may be unlikely to join the legal betting club. CalMatters reported earlier this month that despite various campaigns raising more than $440 million in marketing related to Props. 26 and 27, each measure is garnering support from less than a third of likely voters, according to October polling from the Institute of Governmental Studies at the University of California, Berkeley.

Below, we explore recent research on sports betting. Among the findings of the seven studies featured here:

• Sports bettors are more likely to be white, male, and exhibit psychological traits consistent with narcissism.
• Tax revenue from sports betting may appear substantial in raw numbers, but the impact on tax coffers is muted when compared with income and sales taxes, or tax revenue from other gambling offerings.
• Evidence is mixed as to whether introducing sports betting cannibalizes — eats away at — revenue from other types of gambling.
• Some college football referees may more heavily penalize betting favorites.

The nonprofit National Council on Problem Gambling estimates as many as 8 million adults in the U.S. may have a mild, moderate or severe gambling problem. However, there is a lack of comprehensive, recent academic research on the extent of gambling addiction in the U.S., and the societal costs.

If you feel you may have a problem with gambling you can get help from the National Council on Problem Gambling by call or text at 1-800-522-4700, or online chat at ncpgambling.org/chat .

Research roundup

The Income Elasticity of Gross Sports Betting Revenues in Nevada: Short-Run and Long-Run Estimates Ege Can and Mark Nichols. Journal of Sports Economics, October 2021.

The study: The authors analyze quarterly sports betting data from Nevada covering 1990 to 2019, to explore whether sports betting might be a viable tax revenue stream for other states. Sports betting has been legal in Nevada for decades, so it is the only state with long-run data that can potentially provide insight on the tax base future in states that have legalized sports betting since 2018. The authors note that Nevada is a “mature” market for sports betting, meaning industry growth is relatively stable year to year. A state that newly legalizes sports gambling is likely to see an immediate jump in sports betting revenue, with industry growth levelling off over time.

The findings: In the short-run, quarter-to-quarter, the rise and fall of sports betting revenue in Nevada is most closely tied to changing sports seasons. The authors suggest this is due to differences in how much bettors wager on various sports — the NFL, for example, is “the most popular sport to place wagers on,” with revenues rising and falling as an NFL season begins and ends. In the long run, taxable income in the state and sports betting revenues tend to grow at similar rates. Sports betting revenue in Nevada is a small fraction of revenues from other sources.

The authors write: “Total sports betting revenue in Nevada, the amount kept by the casinos, was $329 million in 2019, implying $22.2 million in tax revenue for the state. In contrast, casino gambling in Nevada in 2019 was $12 billion, generating $810 million in tax revenue. Sports betting is a gambling activity where the amount retained by the casino, and consequently retained by the state, is relatively small as most of the money from losing bets is transferred to those with winning bets. Therefore, sports gambling is a smaller contributor to tax coffers compared to more traditional tax sources such as income and taxable sales or, if applicable, casino revenue.”

A Comparative Analysis of Sports Gambling in the United States Brendan Dwyer, Ted Hayduk III and Joris Drayer. International Journal of Sports Marketing and Sponsorship, August 2022.

The study: The authors explore whether there are psychological differences between bettors and those who do not bet, as well as differences in how closely bettors identify with social institutions, such as religious organizations and far-right or far-left politics.

The authors surveyed 377 bettors and 402 non-betting sports fans from 47 states and explored differences between bettors and non-bettors in states with legal gambling and states where gambling is banned. They also asked about narcissism, which past research has found “is associated to gambling behavior especially as it relates to risky behavior such as participating in illegal gambling,” the authors write. Bettors in the sample were 81% male, compared with 69% of non-bettors. Among bettors, 64% were white and 27% were Black, while 77% of non-bettors were white and 17% were Black.

The findings: In legal gambling states, bettors felt more self-worth than non-bettors, though in states where gambling is illegal the difference in self-worth was almost nil. In legal gambling states, bettors reported a stronger personal identity, “or the importance with which an individual identifies with their relationship and career,” than non-bettors. This relationship flipped in illegal gambling states, with non-bettors showing a stronger personal identity than bettors. In both illegal and legal gambling states, bettors reported slightly higher levels of social uselessness — “an individual’s perceived lack of worth related to social institutions” — than non-bettors, though the gap was wider in illegal gambling states.

The authors write: “Bettors look different and come from different backgrounds and locations. Psychographically, they were clearly more narcissistic. They also indicated a higher social identity and self-worth, yet perceived themselves as less worthy members of important social institutions. In general, sports bettors out consumed non-bettors as it relates sports spectatorship.”

Game Changing Innovation or Bad Beat? How Sports Betting Can Reduce Fan Engagement Ashley Stadler Blank, Katherine Loveland, David Houghton. Journal of Business Research, June 2021.

The study: Legal sports betting means more than $4 billion in additional yearly revenue across the four major sports leagues, according to research the authors cite from the American Gaming Association. At the same time, there may be drawbacks that come with the financial windfall. The authors conduct two studies to explore how sports betting affects fan engagement — the emotional connection fans have with their favorite teams.

The first study included 325 people recruited from Mechanical Turk and focused on betting on a team to win, also called moneyline betting. The second was among 167 Mechanical Turk-recruited participants and focused on prop, or “proposition” bets. Prop bets are bets made on the outcome of some action during the game — whether the next foul ball is caught, missed or goes into the stands, for example. The study is among the first to explore whether there are negative emotional responses from fans related to sports betting.

Participants read a scenario — they were to imagine watching a Major League Baseball game, then randomly they were told they placed either no bet on the game or one of several types of bets. These bets included a $20 bet for the home team to win, along with prop bets. Gaming experts, according to the authors, contend that prop bets can potentially keep fans engaged even if the outcome of the game is obvious — if a team is up by 10 runs by the middle innings, for example. In each study, the participants were asked questions to gauge their emotional investment before and after being told the outcome of the game and their bets. Questions broadly asked about team loyalty, feelings of connectedness to the team and the likelihood participants would watch the team or attend a game, along with other measures of fan engagement.

The findings: In the short run, immediately after a game, the study indicates that betting and losing can decrease fan engagement. Participants who placed no bet were more likely to exhibit loyalty and purchase team-branded merchandise when the team lost, compared with those who placed a moneyline bet. Those who won a prop bet were slightly more likely to be engaged with the team than those who did not bet — but those who lost a prop bet were much less engaged than those who did not bet.

The authors write: “Although industry experts expect sports betting to increase fan engagement, results from two studies do not support this expectation. Instead, we find that when fans lose a bet, positive emotions and subsequent fan engagement decrease.”

College Football Referee Bias and Sports Betting Impact Rhett Brymer, Ryan Rodenberg, Huimiao Zheng and Tim Holcomb. Eastern Economic Journal, January 2021.

The study: The authors explore whether betting lines are related to bias in officiating in the six major Division I college football conferences across 6,598 games from 2005 to 2012. Betting lines indicate whether a sportsbook thinks a game will be close, will favor one team or the other, or be a blowout. The authors note that “college football and basketball are the only major U.S. sports in which conferences have primary managerial responsibility for officials.” If there is a game late in the season with an undefeated team playing a middling team, the conference will benefit financially if the undefeated team wins and goes on to play in a high-profile bowl game. “Referees, as employees of the conferences, are theoretically more likely to have implicit bias favoring the team with higher revenue potential,” the authors write. They use penalty yards per game as a proxy for whether an officiating crew exhibits bias toward one team or the other.

The findings: The authors find signs of bias during in-conference games in two conferences: the Atlantic Coast Conference and the Big East, which reorganized in 2013 and no longer sponsors football. In-conference games are those where two teams from the same conference play each other. In those ACC games where there one team was favored to win by three touchdowns or more, the authors find officials call 6.5 more penalty yards per game against the favorite. In the Big East, the penalty yards increase to 5.7 for the heavy favorite. Further, ACC officials appeared to flag fewer penalty yards against teams that had been in the league longer and enjoyed historic success, rather than newer teams enjoying more recent success. The authors found no officiating bias when an out-of-conference opponent was heavily favored.

The authors write: “… with increasing state regulation, there will likely be more scrutiny of officiating given that a wider spectrum of consumers will have a financial interest in game outcomes. Increased fan engagement via legal sports wagering highlights the importance of pinpointing evidence of bias and undertaking measures to ensure unbiased officiating and game integrity.”

Legalized Sports Betting, VLT Gambling, and State Gambling Revenues: Evidence from West Virginia Brad Humphreys. Eastern Economic Journal, January 2021.

The study: In one of the only studies to examine state-level sports betting revenue after the 2018 Supreme Court ruling, Humphreys looks at sports betting tax revenues in West Virginia and whether gamblers shifted their wagering from video lottery terminal games in casinos to sports betting.

The findings: From September to December 2018, casinos in West Virginia introduced five new sportsbooks, one at each of its licensed casinos. The year before, the state saw a windfall of $253 million in tax revenue from video lottery games. In the year after sports betting was introduced, the author estimates $45.4 million in lost video lottery revenue, with new sports betting revenue pegged at only $2.6 million. The state taxes video lottery revenues at 53.5%, while sports betting revenues are taxed at 10%.

The author writes: “These results should give state policy makers considering legalization of sports betting pause. While new revenue streams from legalized sports betting appear attractive on the surface, states already generate substantial tax revenues from gambling, and the introduction of sports betting to this mix does not leave spending on other forms of gambling untouched.”

Sports Betting’s Impact on Casino Gambling: Cannibalization or Expansion? Ernest Goss and Peyton Miller. University of Illinois Law Review, October 2021.

The study: Another one of few papers to examine how tax revenues and the games bettors played changed after the 2018 ruling, the authors analyze what happened after Iowa allowed sports gambling after August 2019. Iowa casinos that offer sportsbooks pay 6.75% of their sports betting revenue to the state, “a rate tied with Nevada for the lowest nationally,” the authors write. Like in West Virginia, taxes on all other forms of casino gambling are much higher — 22% on revenue over $3 million. The authors do not look at the specific effects of sports betting on other types of gambling, but rather whether there were any changes in overall revenues after August 2019.

The findings: Mobile sports betting and sports betting in casinos did not affect statewide gambling revenues from August 2019 to March 2020. After March, casinos shuttered due to the COVID-19 pandemic.

The authors write: “While these results do not indicate cannibalization within the Iowa gambling market, there are relevant implications for both casinos and the state. Conditions within the state of Iowa may limit the applicability to other states. For example, the varying tax brackets across gambling forms differ from casino taxing in other states.”

Frameworks of Gambling Harms: A Comparative Review and Synthesis Virve Marionneau, Michael Egerer and Susanna Raisamo. Addiction Research and Theory, August 2022.

The study: The authors gather and analyze “harm frameworks” related to problem gambling. A framework in this context refers to a way of categorizing and thinking about an issue with an ultimate goal of understanding the issue in a comprehensive way and finding solutions. A harm is simply an outcome that most of society would classify as negative — losing one’s house, for example, because of gambling-related losses.

While not specifically related to sports betting, the frameworks explored in the paper are useful for those who want to better understand what can happen to individuals and families affected by problem gambling. After searching major academic research databases, the authors settled on seven papers published between 2000 and 2021 that developed an original harm framework related to problem gambling — four of the papers focused on developing the same framework, leaving four frameworks total. The authors, while applauding the research that has already been done, note that further research is needed.

The findings: Two of the frameworks discussed problem gambling harms related to the workplace and personal relationships. One framework separated psychological and cultural harms, and harms related to crime. Another framework mostly focused on risk factors related to problem gambling, risks which “also occur on the individual, familial, community, and societal levels,” the authors write. They note none of the models explore the degree to which problem gambling harms individuals, families, communities and society — all the harms or risks were “treated as somewhat equal,” they write. Financial harms, they argue, might be a relatively worse harm since they “can be seen to precede or even cause many of the other harms, including criminal acts or emotional suffering.” The authors argue for more research on social harms, where, on the whole, the existence of high levels of problem gambling, “can cause harms irrespective of individual participation, including corruption, economic substitution, match fixing, environmental damage related to tourism, or animal suffering.”

The authors write: “We have found that while existing conceptualizations include a wide definition of harms, most harm items are still seen to stem from individual engagement with gambling. Further incorporation of social and societal harms is still needed to conceptualize and operationalize gambling as a public health issue. This includes the development of societal-level harm measurement and harm minimization.”

A brief history of U.S. sports betting

Americans have a long tradition of gambling. College of the Holy Cross economist Victor Matheson recounts in a January 2021 article in the Eastern Economics Journal:

“Lotteries funded activities such as the original European settlement at Jamestown, the operations of prestigious universities such as Harvard and Princeton, and construction of historic Faneuil Hall in Boston … In the sports realm, by 1900 betting on horse races was made illegal except in Kentucky and Maryland, states that to this day host two of the three Triple Crown events in American horseracing, the Kentucky Derby and the Preakness Stakes. States began to relegalize gambling on horse racing in the 1930s as a method of economic stimulus during the Great Depression.”

By the early 1960s, illegal gambling enterprises run by organized crime groups were worth a combined $7 billion . For more than 30 years, the Wire Act , enacted in September 1961, was the only federal law that addressed sports gambling. The law prohibits the use of a wire — a phone, or, more recently, the internet — to transmit information about placing sports bets across state lines.

The Indian Gaming Regulatory Act, which became federal law in 1988, allowed federally recognized Native American tribes to operate casinos on their land. Sports betting in tribal-run casinos, however, was not allowed unless a tribal-state compact was signed. This is the root of the current legal dispute in Florida. Such compacts were in effect in 22 states as of June 2021, according to the International Center for Gaming Regulation at the University of Nevada, Las Vegas.

By the early 1990s, federal legislators were expressing moral panic over the possibility of states allowing sports betting within their borders, to take advantage of billions being wagered illegally.

Illegal transactions are, by nature, difficult to track. People who bet illegally and their bookies do not typically share receipts with the government or trade groups, so it is difficult to say with precision how big the illegal gambling market was before 2018.

Noting that caveat, the American Gaming Association estimated illegal sports betting as a $150 billion-a-year business before the 2018 Supreme Court ruling. It is an oft-repeated figure in news stories and on websites devoted to sports betting.

In 1991, when overall illegal sports bets were estimated in the tens of billions, a Senate report declared sports gambling a “national problem.” The report continued:

“The harms it inflicts are felt beyond the borders of those states that sanction it. The moral erosion it produces cannot be limited geographically. Once a state legalizes sports gambling, it will be extremely difficult for other States to resist the lure. The current pressures in such places as New Jersey and Florida to institute casino-style sports gambling illustrate the point. Without federal legislation, sports gambling is likely to spread on a piecemeal basis and ultimately develop an irreversible momentum.”

Professional sports league commissioners and former athletes were publicly and adamantly against legal sports betting. They objected that sports integrity would be irreparably harmed, including the possibility of fixed games.

Gary Bettman, the top lawyer for the National Basketball Association, made clear to federal lawmakers in 1990 that sports betting was at odds with the league’s profit motive: “Bettors do not care about the win of their team,” Bettman said during a Senate committee hearing. “They only care about the spread being covered and winning their bets. That is not our product. That is not the product we are selling.”

President George H. Bush signed the Professional and Amateur Sports Protection Act on Oct. 28, 1992. It went into effect on Jan. 1, 1993. Bettman became commissioner of the National Hockey League a month later — and, eventually, a fan of sports betting.

“What we’ve learned is that [sports gambling] is another point of engagement for the fans,” Bettman said during a 2019 American Gaming Association conference. “Ultimately, I think if you’re interested in sports betting, you’re going to have an increased opportunity to engage with the game.”

After 1992, some limited sports betting was grandfathered in Delaware, Oregon and Montana. Delaware, for example, allowed a certain type of bets on National Football League games. States were given one year to legalize casino sports betting after the federal law went into effect, but none did. Nevada was the only grandfathered state that fully allowed sports betting. For nearly three decades, the 1992 federal legislation enshrined Las Vegas as the U.S. sports betting hub.

“[Legal] betting was pretty much happening in Nevada, in a regulated market, and you’d have to be at casino to do it,” says Holden, who wrote a comprehensive overview of the rise of legal sports betting published February 2019 in the Georgia State University Law Review. “That gave us sort of the image of sportsbook-style betting that you would see in the movies, you would see on TV. You go to the counter, you place a bet and you watch the game on 50 different screens.”

Constitutional cracks emerged in the 2010s. In 2014, New Jersey legislators voted to reverse their law banning sports betting there. The National Collegiate Athletic Association brought the state to court. This was the case the Supreme Court heard years later, leading to the fall of the Professional and Amateur Sports Protection Act.

Major sports leagues today are on board with gambling. It is impossible to watch professional sports without encountering advertisements encouraging betting. Online sportsbooks spend $154 million yearly in local TV spots, according to Nielsen. Aside from accepting ad dollars from sportsbooks, every major sports league and numerous individual teams have lucrative partnership deals with sportsbooks.

About The Author

Clark Merrefield

The Gambling Behaviour and Attitudes to Sports Betting of Sports Fans

• Original Paper
• Open access
• Published: 01 February 2022
• Volume 38 , pages 1371–1403, ( 2022 )

Cite this article

You have full access to this open access article

• Emma Seal   ORCID: orcid.org/0000-0001-8476-9201 1 ,
• Buly A. Cardak 2 ,
• Matthew Nicholson 3 , 4 ,
• Alex Donaldson 4 ,
• Paul O’Halloran 4 ,
• Erica Randle 4 &
• Kiera Staley 4  

18k Accesses

9 Citations

31 Altmetric

Explore all metrics

Survey responses from a sample of nearly 15,000 Australian sports fans were used to study the determinants of: (i) gambling behaviour, including if a person does gamble and the type of gambling engaged with; (ii) the number of sports and non-sports bets made over a 12-month period; and (iii) attitudes towards betting on sports. The probability of betting on sports decreased with increasing age and was lower for women and people with a university education. This gender difference varied with age, with the greatest difference found among the young. Similar effects were observed for the number of sports bets made, which declined with age. The gender difference in the number of sports bets also varied with age with the greatest difference found among the young arising from the high propensity of young men to bet on sports. Attitudes to sports betting were also analysed, with a key finding that, within friendship circles, the views that sports betting is perceived as harmless, common and very much a part of enjoying sports were stronger among young men. These permissive attitudes were stronger among people who bet on sports and those who bet on sports more frequently. The analysis of sports fans provides insights into the characteristics of the target market most likely to bet on sports, which can be used to inform public health initiatives and harm reduction campaigns.

Similar content being viewed by others

Demographic, Behavioural and Normative Risk Factors for Gambling Problems Amongst Sports Bettors

Young People’s Perceptions of the Effects and Value of Sports Betting Inducements

Gambling risk groups are not all the same: risk factors amongst sports bettors.

Avoid common mistakes on your manuscript.

Introduction

Harm from gambling is a significant global public health issue, with negative impacts on the health and wellbeing of individuals, families and communities (Gainsbury et al., 2014 ). Researchers have argued the harm to health and wellbeing caused by gambling is equivalent to that associated with major depressive disorders, and substance misuse and dependence (Browne et al., 2016 ). There is an array of research linking harmful gambling to health and social issues, including an individual’s health and wellbeing (Rockloff et al., 2020 ; Suomi et al., 2014 ), impacts on families and relationships (Dowling, 2014 ), and an association with intimate partner and family violence (Dowling et al., 2019 ). Generally, harms related to gambling reflect social and health inequalities, with negative effects unequally skewed towards economically and socially disadvantaged groups (Cowlishaw et al., 2016 ; Raybould et al., 2021 ; Wardle et al., 2018 ). Further, Deans et al., ( 2017a , 2017b ) argued that older adults, young men, and children are most vulnerable to harm from gambling.

In this paper, we explore the gambling choices of a diverse group of sports fans from Victoria, Australia, aged 18 and over, based on data from a survey of almost 15,000 members and fans of elite sporting clubs. In doing so, we investigate the relationship between individual demographic characteristics, the gambling behaviour of these sports fans and differences in attitudes to sports betting. Australia is recognised globally as having one of the most accessible and liberalised gambling environments, with policy and regulation, online platforms and the diversification of gambling products all increasing the availability and uptake of different gambling opportunities (Deans et al., 2016a ; Hing et al., 2017 ; Pitt et al., 2017a ). However, this trend is reflected elsewhere, with similar issues reported in the United Kingdom (McGee, 2020 ), Spain (Lopez-Gonzalez, et al., 2020 ) and Ireland (Fulton, 2017 ), implying our analysis is of international importance for those seeking to understand gambling choices and attitudes, and mitigate harm through appropriate policies and programs.

Recent evidence from the Household, Income and Labour Dynamics in Australia survey (a nationally representative longitudinal survey) demonstrated that there were 6.8 million regular gamblers in 2015, of whom an estimated 1.1 million were at risk of harm from gambling-related problems (Armstrong & Carroll, 2017 ). The National Australian Gambling Statistics Report highlighted that total gambling losses rose 5% between 2017 and 2018 to $24.89 billion. These statistics demonstrate gambling is an ongoing and increasing threat to individual and public health. Not only are individuals at risk of harm from gambling, for one person with problematic behaviour, an estimated five to ten people are adversely affected (Productivity Commission, 1999 ), implying widespread economic and social costs of gambling (Wardle et al., 2018 ).

Rise and Normalisation of Sports Betting

This research focuses on sports betting, a rapidly emerging sector of the gambling industry. Its impact on normalising gambling, especially among the young, has been of increasing concern over the last decade in countries like Australia and the United Kingdom (Purves et al., 2020 ). Sports betting is one of the few forms of gambling that has shown a substantial increase in participation in recent years (Hare, 2015 ). In Australia, sports betting resulted in the largest year-on-year percentage increase (16.3%) in gambling losses during 2017–2018. The relationship between sports and gambling is increasingly symbiotic, with teams from Australia’s two major professional sports, the Australian Football League (AFL) and National Rugby League (NRL), significantly involved in the ownership and promotion of gambling products and services. Activities include formal sports partnerships, uniform naming rights, stadium signage and the promotion of odds during televised broadcasts. This general trend has been termed the ‘gamblification’ of sports by McGee ( 2020 ) and has become ubiquitous across a variety of sports settings, from elite to community level.

As a consequence of the pervasiveness of sports betting, researchers have increasingly sought to identify and describe the ‘normalisation’ effect of sports betting and its acceptance as part of peer-based socialisation and general sports fandom (Bunn et al., 2019 ; Raymen and Smith, 2017 ). A growing body of evidence has started to address the factors that lead to sports betting being perceived as an everyday part of sports, fostering its uptake. This is aligned with an increasing focus within broader gambling research on the influence of the environment and social determinants on people’s behaviour, as opposed to concentrating on a problem or pathology within the individual (Johnston and Regan, 2020 ).

There has been a strong research focus on the rise and prominence of sports betting marketing, quantifying how prevalent gambling promotions are during sports broadcasting (Milner et al., 2013 ), on social media platforms (Thomas et al., 2018 ), in live events within stadia (Thomas et al., 2012 ), and exploring how online platforms have been harnessed by wagering companies to encourage consumption (Deans et al., 2016a ). Thomas et al. ( 2012 ) highlighted there were very few visible or audible messages to counter overwhelmingly positive messages about sports betting during matches. Their research also addressed how sports betting advertising and associated strategies affect the attitudes of specific community sub-groups, including young people, parents, and young males. Pitt et al. ( 2016a ) found children could recall sports betting brand names, places they had seen betting advertising and associated plot details of advertisements. Deans et al. ( 2017b ) conducted similar work with young men and demonstrated sports betting marketing influences their betting behaviour.

Research has also focused on people’s attitudes to sports betting advertising, to improve understanding of community sentiment. Generally, this has shown that both parents and young people disagree with the increase in sports betting advertising and have concerns about how these messages promote a seemingly natural affinity between gambling and sports (Nyemcsok et al., 2021 ; Pitt et al., 2016b ). However, Pitt et al. ( 2016b ) reported that young people’s discourses about sports increasingly involve discussions about gambling ‘odds’ and that some young people believe that gambling is a usual and valued consumption activity during sports. Alternative evidence suggests young men feel particularly overwhelmed and bombarded by sports betting advertising (Thomas et al., 2012 ). This is unsurprising because this group is the target market for most Australian wagering operators. Hing et al. ( 2016 ) argued that such operators deliberately position sports betting as an activity engaged in by single, professional, upwardly mobile young men.

Other environmental or ‘normalisation’ issues investigated in the research literature include the availability and convenience of sports betting on mobile phone apps or online, with ease of access facilitating gambling (McGee, 2020 ), the socio-cultural alignment between sports betting and sports (Deans et al., 2017a ; Thomas, 2014 ), and how physical and online environmental factors influence the gambling risk behaviours of young men (Deans et al., 2016b ). For example, Deans et al. ( 2017a ) conducted semi-structured interviews with a convenience sample of 50 Australian men, aged 20–37, who were fans of and had bet on either NRL or AFL matches (games). These young male sports bettors reported their betting was normal and socially accepted, especially among sports fans, and ‘gambling-related language had become embedded in peer discussions about sport’ (p. 112). As such, Deans et al. concluded an exaggerated normalisation of wagering might exist in male sports fans’ peer groups. The young men in their study had established rituals (e.g. punters’ clubs) that reinforced their social connection to sports betting, but also enhanced the peer pressure to bet—an outcome that is perhaps inevitable given the role of social interaction in normalising behaviour (Russell et al., 2018 ).

Understanding ‘Sports Bettors’

A separate but connected branch of research literature has concentrated on profiling groups most at risk of experiencing harm from sports betting, particularly describing their attitudes and characteristics. As indicated previously, two groups of major concern are men and youth in general (both male and increasingly female). Studies and reviews have consistently found that young adult males are at greater risk of problem gambling (Hing et al., 2015 ; Williams et al., 2012 ). Recently, such research has also explored sports betting specifically. For example, Hing et al. ( 2016 ) in a quantitative study with a purposive sample of 639 Australian adults, identified key demographic risk factors for problem sports bettors included being male, younger, never married, and living either alone, in a one-parent family with children, or in a group household. Other risk factors included having a higher level of education and working or studying full-time. Numerous, frequent, and larger bets appeared to characterise high-risk sports bettors, as opposed to those deemed at low risk of experiencing harm. This is supported by recent research by Ayandele, Popoola and Obos ( 2019 ), who surveyed 749 Nigerian tertiary students aged 16–30 years to explore how socio-demographic factors, peer-based gambling and sports betting knowledge interact to shape young adults’ attitudes to sports betting. They found a favourable attitude towards sports betting was associated with being older and male, having a knowledge of sports betting and was positively related to the betting attitudes and behaviours of friends.

Whilst understanding normalisation factors and processes is important, alongside the attitudes and characteristics of different sub-groups, what is missing from current research is a large-scale examination of the attitudes of sports fans specifically, and the key demographic factors that are associated with their sports betting behaviour. This study addresses the gap. To the best of our knowledge, this research, conducted with almost 15,000 unique respondents, is the largest quantitative study operating at the nexus of sports fans and gambling behaviour. Arguably, the issues described above are more impactful in this cohort because they are highly engaged with sports and very exposed to marketing and the gambling economy. It is imperative to understand how such environmental and socio-cultural processes influence sports fans’ betting behaviour and to identify the sub-groups where these behaviours are most apparent.

Using quantitative research with a broad demographic group of sports fans (aged 18 and over), we aimed to compare attitudes between sports bettor, non-sports bettor and non-bettor cohorts, and examine the factors that would make it more likely for a sports fan to be a sports bettor. Broadly, we focused on attitudes to betting in sports, the risks associated with sports betting and perceptions about how much of a social norm this activity is. The research was guided by the following questions:

What demographic factors make it more or less likely for a sports fan to bet on sports? How does this correlate with the number of sports bets a person makes?

What is the impact of described normalisation processes on the attitudes of sports fans that bet on sports, compared with those that do not bet at all, and those that bet but not on sports?

Understanding the demographic profiles and risk factors for sports bettors and their attitudes is an increasingly important area of research. The results from this research could inform public health interventions and policy to help ensure they appropriately address areas of concern.

The research questions above are addressed as part of a broader project undertaken in partnership with the Victorian Responsible Gambling Foundation (VRGF). The project involved a survey that was distributed in collaboration with 17 professional sporting clubs from Australian Football, Basketball, Cricket, Soccer, Netball and Rugby Union in the state of Victoria, Australia. The survey was targeted at members, fans, and supporters of these clubs. In each instance, the survey was shared via social media channels (including Facebook and Twitter) and electronic direct marketing using email to each club’s membership base. Depending on the sporting code and relevant season (summer or winter), the survey was shared either between 30th October–3rd December 2020 or 25th February–18th March 2021. Data were collected in the context of the COVID-19 pandemic and research has demonstrated an increase in sports betting and a decrease in other types of gambling (e.g., casino, horse racing, pokies, etc.) during this period (Jenkinson et al., 2020 ). Whilst this could have impacted the behaviour and attitudes of respondents in this survey, the results still highlight the groups most at risk from engaging in sports betting and their associated attitudes. The survey took on average 15 min to complete and elicited a total of 17,228 responses. However, due to incomplete survey responses, the estimating sample is restricted to at most 14,950 observations in the analysis below. Three components of the survey were used in this study. First was data on gambling behaviour comprising responses about (i) whether an individual gambles or not and whether any gambling is sports betting, non-sports betting or both; and (ii) the number of bets in a given period.

Participants were asked about gambling activity in general first. They were advised: “Gambling includes activities in venues such as casino table games, pokies, TAB, Keno etc ., as well as raffles, lotteries and scratchies. It also includes gambling online or via apps such as sports betting, race betting and online pokies and casino games, where you bet with money.”

This was followed with the questions:

Thinking of all these types of gambling, in the past 12 months, have you spent any money on these gambling activities?

In the past 12 months, how often have you gambled? (with a number of times per week, month or year options available)

Participants were then asked about sports betting activity. They were advised: “ Sports betting refers to legal wagering with bookmakers on approved types of local, national or international sporting activities, (other than horse or greyhound racing both on or off the course) in person, via the telephone or via an app or online.”

Thinking of all these types of sports betting, in the past 12 months, have you spent any money on these sports betting activities?

In the past 12 months, how often have you taken part in sports betting? (with a number of times per week, month or year options available)

This data are summarised in the first six rows of Table 1 under the sub-headings ‘betting category’ and ‘number of bets’. For clarity, when referring to the specific participant groups involved in the research, we will use the terms sports bettors, non-sports bettors, or non-bettors to avoid referring to gambling and betting interchangeably. Table 1 shows that about 35% of the sample are non-bettors while another 35% are non-sports bettors (i.e. they bet on lotteries, raffles, poker or slot machines and casino gambling). The remaining 30% of the sample are sports bettors divided evenly between those that engage in sports betting only and those that engage in both sports and non-sports betting. The average number of bets in a year for the full sample is 19.6 for non-sports and 14.5 for sports bettors. Footnote 1 This data are also presented by betting category, showing that sports bettors, on average bet many more times per year than non-sports bettors. However, those that bet on both have a much higher betting frequency again, betting nearly twice as often as those who bet only on sports. The much higher standard deviation among sports bettors is also notable, suggesting there is much greater variation in betting frequency among sports bettors than non-sports bettors.

The second component of the data used in this study comprises demographic characteristics. This includes gender, age, location (metropolitan or regional), marital status, education, employment status, income categories, country of birth, parents’ country of birth, Aboriginal Torres Strait islander origin and health status. The data are summarised for the full sample and the different betting categories in Table 1 . Some key features of the data are that about 70% of the sample is male but nearly 80% of sports bettors are male and the average age of the sample is 50 years, but sports bettors have an average age of 43 years for sports bettors and 46 years for those who both sports and non-sports bet. Two other notable features are the much higher proportion of non-sports bettors who are retirees (0.24) compared to the proportion of sports bettors (0.07) and the high proportion of responders who did not report their income (0.20).

The third component of the data used here comprises responses to questions about attitudes to gambling. The questions are grouped into two categories including (i) general attitudes to gambling and sports betting; and (ii) perceptions of the attitudes and behaviours of others. Responses were elicited on a scale from 0 (totally disagree) to 10 (totally agree). The questions are presented in Table 2 , where sample means and standard deviations are presented for the full sample and the different betting categories.

An important research question here is whether people in different betting categories respond to each question differently. We conducted a one-way analysis of variance (ANOVA) for each question to test whether the mean responses of each group are statistically significantly different. The p -value of this joint F -test for each question is presented in the ‘Full Sample’ column in braces; all tests have p -value of [0.000] implying we reject the null hypothesis of equality between the mean response of each group to each question. We also test the differences between means for each group using t -tests with a Sidak correction to account for the possibility of a false positive finding given the large number of t- tests. The table reports results of tests of differences between the mean response of non-bettors and each type of betting group with statistically significant differences at the 10%, 5% and 1% levels denoted by *, ** and *** respectively. The results show average responses of almost all betting groups differ from those of non-betters at the 1% level of significance. Differences are most stark in relation to the notion that ‘sports betting should not be part of experiencing sport’ and items related to the social aspects of sports betting, particularly the place of sports betting in the person’s family and friendship groups. However, responses to (i) ‘most people in society think betting on sport is harmless’ were not statistically different between sports bettors and non-bettors; and (ii) ‘odds talk is common in discussions about sport with my friends and peers’ were not statistically different between non-bettors and non-sports bettors.

An important feature of the survey is that it reached beyond people who identify as gamblers, providing insights into a more diverse sample than many previous studies. However, a limitation is that the design is focused on members, fans or supporters of a group of elite clubs or teams, implying to some degree that respondents are likely to be more engaged with sport than the average member of the Victorian population. Therefore, the differences between bettors and non-bettors identified here are potentially lower bounds and analysis of a more representative sample might uncover even greater differences. Another important benefit of the sampling frame is that it is likely the target audience for sports betting advertisers. Therefore, the analysis offers insights into a group that is likely most targeted and affected by sports betting advertising and understanding this group provides valuable insights to harm minimization policies with respect to sports betting.

Empirical Methods

The empirical analysis can be divided into two broad approaches. The first is to analyse the determinants of the betting choices of survey respondents. The second is to analyse the responses to gambling attitude questions. The approaches to these analyses are described in turn below.

Who Bets and How Often?

Each survey respondent is assumed to choose between four types of betting activity: no betting, non-sports betting, sports betting or both sports and non-sports betting. We define the gambling choice of each survey respondent \(i\) as \({G}_{i}=j\in \left\{1, 2, 3, 4\right\}\) . We want to understand the relationships between different demographic characteristics’ and gambling choices. Given these four possible gambling choices or outcomes are unordered, we used a multinomial Probit specification to estimate the relationships. The probability that individual \(i\) makes gambling choice \({G}_{i}=j\) is given by

where \({X}_{i}\) is a vector of personal characteristics (including gender, age, location, marital status, education, employment status, income categories, country of birth, parents’ country of birth, Aboriginal Torres Strait islander origin and health status, as listed in Table 1 ), \(\Phi \left(.\right)\) is the cumulative density function of the Standard Normal distribution and \({\beta }_{j}\) provides the parameter estimates on \({X}_{i}\) for gambling choice \(j\) ; see Greene ( 2018 ) Chapter 18 for more details. This model allows us to understand the influence of each characteristic on the four possible gambling choices from a model that jointly estimates the probabilities of each alternative gambling choice.

As parameter estimates do not have a clear intuitive interpretation in such models, we compute marginal effects for each \({X}_{i}\) which are given by

where \(\overline{\beta }\) is the probability weighted average of the parameter estimate across the four different possible gambling choices. The multinomial Probit results reported in Table 3 below are average marginal effects which are computed as the average of for \({\delta }_{ij}\) across all \(i\) individuals. The interpretation of these marginal effects is that they tell us the impact of a unit increase in \({X}_{i}\) (for example female versus male or a one-year increase in age) on the probability of making gambling choice \(j\) ; that is, four marginal effects will be reported for each variable, one for each of no betting, non-sports betting, sports betting or both sports and non-sports betting.

Along with the choice of gambling type, individuals choose how many bets to place, and the factors that influence this number of bets are also of interest. Since 35% of survey respondents are non-bettors, the data on the number of bets comprises a large number of zeros. To accommodate this feature of the data, a Cragg hurdle model is adopted (Cragg, 1971 ). This model involves two parts: the first is a model of the decision to gamble (selection model), while the second is a model of the number of bets. As we have data on the number of sports bets and non-sports bets, we estimate this two-part model separately for each of these choices. The first part of the model is the selection decision, given by

where \({C}_{i}\) is individual \(i\) ’s choice of whether to bet on sports (1) or not (0) or alternatively to make non-sports bets (1) or not (0). The control variables \({X}_{i}\) are as defined above for the multinomial Probit model in Eq. ( 1 ), while \(\alpha\) is a vector of parameters capturing the influence of each control variable on the decision to place bets or not and \({\epsilon }_{i}\) is a mean zero, constant variance normally distributed disturbance term. The second part of the model estimates the number of observed bets. This continuous variable is given by

an exponential specification of the Cragg hurdle model where \({B}_{i}\) is the number of bets made per year by individual \(i\) , \(\theta\) is the set of parameters reflecting the impact of variables \({X}_{i}\) , again as defined above and \({u}_{i}\) is a mean zero, constant variance disturbance term.

The key idea of this model is that the decision to gamble is modelled separately from the decision of how many bets to place. Estimates of \(\alpha\) are important determinants of the decision not to gamble and therefore choose zero bets, whereas \(\alpha\) and \(\theta\) together determine the number of bets if a person chooses to gamble. The overall marginal effect of control variables \({X}_{i}\) , which in our specification are common to the selection and number of bets equations, are computed using the margins command in STATA and presented in Table 4 below. The detailed expressions for these marginal effects can be found in Burke ( 2009 ). The interpretation of these marginal effects is that they reflect the impact of a unit change in the value of a control variable \({X}_{i}\) (i.e. gender or age) on the average number of bets in a given period of time (a year in this instance).

Factors Affecting Attitudes Towards Sport Betting

Our analysis of responses to the 12 different questions about attitudes to sports betting listed in Table 2 comprises two key objectives. First, we are interested in the relationship between individual gambling choices and attitudes to sports betting—this involves the type of betting and the number of bets. It is anticipated that sports bettors will hold more positive attitudes towards sports betting than non-bettors or non-sports bettors. Second, we are also interested in the relationship between demographic characteristics and attitudes to sports betting. Responses to each question range on a scale from 0 to 10 but are standardized to have mean zero and standard deviation of one, to enable comparisons of effects of different variables across survey questions. These standardized responses are modelled using OLS. The model estimated is given by

where \({A}_{i}\) is the response of individual \(i\) to one of the 12 questions on their attitude to gambling listed in Table 2 . For each question, two versions of the model are estimated where \({Z}_{i}\) is a vector of control variables, including all variables in \({X}_{i}\) which is as defined above, along with either (i) gambling choice, \({G}_{i}\) , (no betting, non-sports betting, sports betting or both sports and non-sports betting); or (ii) the number of sports and non-sports bets, \({B}_{i}\) , included. Model parameters are given by \(\gamma\) while \({\varepsilon }_{i}\) is a disturbance term with zero mean and constant variance. As the dependant variable, \({A}_{i}\) , is a standardized measure of responses to attitude questions, \(\gamma\) should be interpreted as the average number of standard deviations of change in \({A}_{i}\) per unit change in \({Z}_{i}\) .

In all models, we allow for non-linear age effects by including quadratic and higher order age terms, along with a linear age term, testing their significance using a likelihood ratio (LR) test. We also allow for gender effects to vary with age by including age and gender interactions and testing for their significance, again using a LR test.

Factors Affecting Gambling Behaviour

The results of estimation of Eq. ( 1 ) are presented in Table 3 . The average marginal effects of the listed control variables on different gambling choices of no betting, non-sports betting, sports betting and both sports and non-sports betting are presented in first, second, third and fourth columns respectively. The model includes linear, quadratic and third order age terms. Footnote 2 Focusing on the cases of sports betting only (third column) and both types of betting (fourth column) and on results that are significant at the 1% level (denoted by ***), we found that relative to males, females are 9.6% less likely to bet on sport and 6.2% less likely to bet on both. To test the hypothesis that gender effects vary with age, all the included age terms are interacted with the gender indictor, with these interactions supported by a LR test = 62.63 ( p value = 0.00). The marginal effect for females relative to males is plotted for each betting category in panels (a)–(d) of Fig.  1 . The difference between men and women sports betting is greatest among the youngest in the sample and the difference decreases with age. The result suggests that young men are up to 25% more likely than women of the same age to bet on sports; the difference is less than 10% for people over 50 years. In addition, the average marginal effect of a 1-year increase in age is reported in Table 3 . Comparing otherwise identical individuals with a 10-year age difference, the older person is 5.0% less likely to sports bet only and 2.0% less likely to make both sports and non-sports bets relative to the younger person.

Marginal effect of gender on the probability of each betting category plotted over age. Panels ( a )–( d ) are based on estimates presented in columns (1)–(4) of Table 3 respectively. Point estimates are denoted by dots with 95% confidence intervals for each point estimate included

Relative to self-employed, which is the base category, students are 6.7% less likely to bet on sports, while those on home duties are 7.0% less likely to gamble on both sports and non-sports. Income results are all relative to the base range of “less than $10,000”, with little difference between different income ranges in the probability of sports betting only. However, respondents above $60,000 are between 5.6 and 8.0% more likely to bet on both sports and non-sports than those in the base income range. This suggests there is little effect on gambling probability of additional income as the marginal effects are similar for each category above $60,000. Finally, respondents whose parents were both born overseas were 2.7% less likely than the base category (both parents born in Australia) to bet on both sports and non-sports.

In Table 4 , we present estimates of the model of the number of bets specified in Eqs. ( 3 ) and ( 4 ). The marginal effects of the full set of control variables on the number of non-sports bets and sports bets are presented in first and second columns of Table 4 respectively. Age is included in these models through linear, quadratic and cubic terms. Footnote 3 As the number of bets is a continuous variable, the marginal effects are interpreted as the impact of a unit change in the control variable on the number of bets per year.

Focusing first on results significant at the 1% level (denoted by ***) for the number of non-sports bets, females make on average 9 fewer bets per year than males. Betting increases with age, with a 10-year older person making on average 5.29 more bets. The model also includes an interaction between all age terms and gender to test the hypothesis that gender effects vary with age — this interaction is supported relative to the model without the interactions, LR test = 28.41 ( p value = 0.00). The marginal effect of gender on the number of non-sports bets increases with age with women over 58 placing around 20 fewer bets than men of the same age: panel (a), Fig.  2 . Further results include that someone who is married makes 4.5 fewer bets than a single person and parents place 5.0 fewer bets than people with no children. Education reduces betting with those holding trade qualifications betting 6.3 times less than someone who did not complete high school while those with university qualifications betting 15.3 fewer times than someone who did not complete high school. Employment status and income are uncorrelated with the number of non-sports bets. Respondents whose parents were both born overseas place on average 4.4 fewer bets than people whose parents were both born in Australia.

Marginal effect of gender on the ( a ) number of non-sports bets and ( b ) number of sports bets, plotted over age. Panels ( a ) and ( b ) are based on estimates presented in columns (1) and (2) of Table 4 respectively. Point estimates are denoted by dots with 95% confidence intervals for each point estimate included

Again, focusing on results significant at the 1% level, results for the number of sports bets show that women place on average 18.2 fewer bets than men per year. In contrast to non-sports bets, the number of sports bets placed decreases with age; a person 10-years older bets on average 5.0 fewer times per year. The interaction between age and gender confirms that the gender effect on the number of sports bets does vary with age—supported relative to the model without the interactions, LR test = 37.52 ( p value = 0.00). However, the marginal effect of gender on the number of sports bets decreases with age, with 18-year-old women placing 50 fewer sports bets than 18-year-old men. This difference is as low as 10 fewer sports bets per year when comparing women and men aged over 60 years: panel (b), Fig.  2 . Education effects are similar to those for non-sports bets. People with trade qualifications bet on average 6.7 (10.0) times less than someone who did not complete high school. The only employment status category that is related to the number of sports bets is being a student, who place on average 10.3 fewer bets than the self-employed. The number of sports bets is related to income with most annual income categories above $40,000 betting on average between 8 and 11 more times per year, though some differences are significant only at the 5% level and others are smaller with 5 more bets per year and significant at the 10% level. Finally, respondents whose parents were both born overseas place 4.2 fewer sport bets on average than people whose parents were both born in Australia.

Attitudes to Sports Betting

Selected results of estimating the model presented in Eq. ( 5 ) using the responses to the first set of questions on general attitudes about sports betting summarized in Table 2 are presented in Tables 5 and 6 . Full results of these models are available in Online Resource 1 and Online Resource 2 of the Supplementary Materials, where estimates for all variables included in the models are presented. In all these models, an interaction between gender and all age terms (up to third order age terms are included in all models) is considered with the result of a LR test of the restriction that the coefficients on the interactions are zero presented in the last row of each column. In cases where the restriction is rejected and the interaction is non-zero ( p value < 0.05), the model presented includes the interactions.

Results in Table 5 are for the model estimated with all the demographic control variables in \({X}_{i}\) together with a set of indicators for each type of betting behaviour, \({G}_{i}\) , including non-sports betting, sports betting and both sports and non-sports betting, with no betting the omitted base category. Each column presents estimates for a model of standardized responses (mean zero and standard deviation of one) to a separate statement about sports betting. The statements upon which each dependent variable is based are listed in the table notes. Focusing on results significant at the 1% level (denoted by ***), the results show that after controlling for a large set of individual demographic characteristics, people who bet are on average less concerned about sport betting issues than non-bettors. This is evident for all 5 statements modelled. We can see in column (1), for example, responses to the statement ‘sports betting should not be part of experiencing sport’, relative to non-bettors, the average response of people who bet on non-sport only is 0.20 standard deviations lower, while people who bet on sport (only or both sport and non-sport) have an average response that is 0.71 standard deviations lower. Other key results from these models are that females are more concerned about sports betting than males, except for in their responses to the statement ‘people who bet regularly on sport are at risk of harm from gambling’, where there is no difference between men and women. The relationship with age is statistically significant for the statements ‘sports betting should not be part of experiencing sport’ and ‘people who bet regularly on sport are at risk of harm from gambling’ but the effects are small, with a 10-year older person having on average a 0.06 standard deviation higher response to the former question and a 0.03 standard deviation lower response to the latter. People from regional locations are on average more concerned about sports betting than people from metropolitan locations; however, this concern is not evident in response to ‘sports betting should not be part of experiencing sport’. The results on education show that relative to the base case of ‘did not complete high school’, those with trade qualifications are on average more concerned about sports betting and in turn, people with university education are even more concerned with even greater differences evident than for those with trade qualifications.

The results presented in Table 6 are for models of the same questions with the same demographic controls included but with betting behaviour replaced by the number of sports and non-sports bets, \({B}_{i}/100.\) Once again, each column presents estimates for a model of standardized responses (mean zero and standard deviation of one) to a separate statement about sports betting. The statements upon which each dependent variable is based are listed in the notes to the table. The impacts of the demographic characteristics in Table 6 are qualitatively similar to those found in Table 5 . The key difference between Tables 5 and 6 is that gambling categories are replaced with the number of sports bets and the number of non-sports bets. On average, people who bet more often are less concerned about sports betting. However, 100 more sports bets per year (approximately 2 bets per week) has nearly double the impact on responses of 100 more non-sports bets. For example, responses to ‘sports betting should not be part of experiencing sport’ are on average 0.14 standard deviations lower for every additional 100 non-sports bets but are 0.30 standard deviations lower for every additional 100 sports bets.

The above analysis is repeated for the second set of statements summarized in Table 2 which focus on perceptions of the sports betting attitudes and behaviours of others. Selected results of this analysis with betting categories included are presented in Table 7 and with the number of bets included are presented in Table 8 . Each column presents estimates for a model of responses to a separate statement about sports betting. Survey responses used to estimate each model have been standardized to have mean zero and standard deviation of one. The statements upon which each dependent variable is based are listed in the notes to each table. Full results of these models are available in Online Resource 3 and Online Resource 4 of the Supplementary Materials.

Focusing on significance at the 1% level (denoted by ***), Table 7 shows, except for the first two statements that focus on attitudes in society, bettors have families and friendship groups where gambling is common and perceived as harmless. However, the difference between sports bettors and non-sports bettors is stark. For example, in response to the question ‘most people in my friendship group bet on sport’, the average response of non-sports bettors is 0.16 standard deviations higher than non-bettors whereas the average response of sports bettors is up to 0.50 standard deviations higher, with both significant at 1%. The difference between non-bettors and sports bettors (0.49 standard deviations) is nearly 10 times as large as the difference between non-bettors and non-sports bettors (0.05 standard deviations) in response to the question ‘odds talk is common in discussions about sport with my friends and peers’. The largest impact of age is on the peer and friendship group statements, columns (5)–(7), with a response of a person 10-years younger on average 0.15 standard deviations higher. The interaction between gender and age is illustrated for the results reported in column (6) which is based on the statement “most people in my friendship group bet on sport” in panel (a) of Fig.  3 . The difference between men and women is greatest at younger ages with women in the 18–45 year range responding on average 0.5 standard deviations lower than men, suggesting young men have a much stronger belief than young women that their friends are involved in sports betting.

Marginal effect of gender on response to the question “Most people in my friendship group bet on sport”, plotted over age. Panel ( a ) is based on model in column (6) from Table 7 which includes betting categories and panel ( b ) is based on model in column (6) from Table 8 which includes number of bets. Point estimates are denoted by dots with 95% confidence intervals for each point estimate included

The results presented in Table 8 are for models of the same questions analyzed in Table 7 , with the same demographic controls included but with betting behaviour replaced by the number of sports and non-sports bets, \({B}_{i}/100.\) The relationships with the demographic characteristics are qualitatively similar to the results presented in Table 7 and discussed above. The more a respondent bets, the greater their agreement with all but the first two statements that focus on attitudes in society. The relationship between responses and number of sports bets is up to 4 times as large as the relationship with the number of non-sports bets. For example, responses to the question ‘most people in my friendship group bet on sport’ on average increase by 0.10 standard deviations for people who place 100 more non-sports bets per year (2 more bets per week) but for an otherwise identical person who places an additional 100 sports bets, their response is on average 0.38 standard deviations higher. These sorts of differences are evident across all questions about family and friendship groups, with a greater number of bets associated with responses that show sports betting is believed to be more common and perceived as less harmful in these circles. It is also found the more sports bets a person places, the stronger their agreement with the statement ‘most people in society bet on sport’, though the relationship with the number of non-sports bets is statistically insignificant. The gender and age interaction for the model in column (6) in Table 8 is presented in panel (b) of Fig.  3 . The figure shows the difference between men and women is greatest among 18–28-year-olds, with the responses of women in this age range on average 0.70 standard deviations lower. This compares with differences of less than 0.30 standard deviations for those over 60 years. The results are consistent with those in panel (a) of Fig.  3 , suggesting that gender age differences are robust whether we control for the betting category or the number of bets per year.

This study builds on existing literature at the intersection of sports betting and sports, providing a comprehensive analysis of the sports betting behaviour of sports fans, including many people who choose not to gamble at all. Survey respondents’ attitudes to sports betting were analysed using betting behaviour and a wide range of demographic characteristics. The approach differs from many previous studies as we targeted a broader demographic of sports fans, rather than focusing only on those engaged in gambling (sports or non-sports), which is a strength. We did not measure whether a person’s gambling behaviour is deemed ‘problematic’, but previous research has demonstrated a connection between frequency of sports betting and problematic gambling behaviour (Hing et al., 2016 ). Therefore, our analysis of the number of sports bets provides a useful proxy to identify those most at risk of experiencing gambling harm. The research was guided by two overarching questions addressed in turn through the following discussion.

Demographic Profile of Sports Bettors

The dominant theme emerging from our analysis is the importance of gender, age and their interaction. The gender difference in the probability of sports betting is wider among the youngest in the sample; 18-year-old men are about 25 percentage points more likely than their female counterparts to bet on sports, whereas this difference is less than 5 percentage points for those over 60 years. Similar patterns are evident for the number of sports bets placed, with younger men placing more bets than similar aged women and fewer bets being placed with each additional year of age. Consequently, young men are most at risk based on their sports betting engagement and number of bets placed. This aligns with previous studies (Hing et al., 2016 ; Williams et al., 2012 ), but widens our understanding of the sports betting behaviour of sports fans. Moreover, even though this has been described in smaller-scale qualitative research studies (Deans et al., 2017b ; Waitt et al., 2020 ) our results are based on empirical analytic techniques applied to a larger and more diverse sample. The results comprehensively demonstrate sports betting is predominantly pursued by young men, in sharp contrast to other forms of gambling. Given the recent growth of sports betting, its marketing, and increasing contribution to problem gambling (Hing et al., 2019 ), as well as the need for appropriately tailored prevention and early intervention public health initiatives, this finding is significant for highlighting the distinctive sports betting behaviour of young men aged 18–35. Recent research has started to examine young women aged 18–35 as an emerging gambling cohort (see McCarthy et al., 2020 ), but our results demonstrate no significant gender or age effects for women’s sports and non-sports betting behaviour.

Other important demographic factors included education level, relationship status, and employment status. People who are widowed or separated were less likely to bet on sports, but no other relationship types were significant at the 1% level. University educated individuals were less likely to bet on sports than those who did not complete high school. Employment status did not exhibit a strong relationship with sports betting, except students and unemployed were less likely than self-employed to bet on sports. Surprisingly, income did seemingly not influence whether people engaged in sports betting only. This was more important in the context of making both sports and non-sports bets—those reporting higher levels of income were more likely to engage in these gambling types.

Whilst other studies have reported various demographic risk factors for sports betting and gambling, our results contribute by clearly demonstrating the significant interaction between age and gender. The importance of our study for public health policy and harm reduction campaign strategies is twofold. First, our sampling frame is likely the target audience of sports betting marketers, providing strong evidence upon which to base public health policy and harm reduction campaigns. Second, such campaigns should be aimed specifically at young men to help counteract the increasing environmental and social normalisation of sports betting. The next section focuses on the key attitudinal differences that emerged from the results to answer our second research question.

Attitudes Associated with Sports Betting

Our results demonstrate there are significant differences between the attitudes of sports bettors (either sports betting or sports betting combined with non-sports betting), non-sport bettors and non-bettors. Not only do sports bettors feel more strongly that sports betting has a place in sport, they are also less concerned about the risks and harms of sports betting. These results help to demonstrate the effects of the normalisation processes outlined in previous studies. Whilst existing literature has documented how sports and sports betting have become synonymous (Milner et al., 2013 ; Nyemcosk et al., 2021 ; Pitt et al., 2016a ; Thomas, 2018 ), the attitudinal differences we identified highlight how the ‘gamblification’ (McGee, 2020 ) of sports has penetrated individual perceptions about sports betting as an activity and influenced behaviour. Moreover, the differences between sports bettors and non-sports bettors suggest something unique is happening for this group; it is not necessarily related to the act of gambling, but potentially broader environmental and socio-cultural influences. Gender and age effects are also apparent, with women less likely to agree that sports betting should be part of experiencing sports and more likely to agree that sports betting can place people at higher risk of other harms. A similar pattern is evident for age, with younger people generally being more permissive of sports betting.

The influence of the social aspects of sports betting, namely the characteristics of sports bettors’ social networks, is also a strong emerging theme, underpinned by several attitudinal measures. Sports bettors are more likely to have family and friendship groups where gambling is common and perceived as relatively harmless. Additionally, they are more likely to agree that discussions about odds and the placing of bets is the norm amongst peers. In this context, there are significant differences observed between both sports bettors and non-bettors, and sports bettors and non-sports bettors. This again highlights that there are potentially distinct socialisation processes specifically influencing the attitudes and behaviours of sports bettors. Whilst this has previously been described on a relatively smaller scale (Thomas, 2017a ), our research demonstrates that these interactional and socialisation factors are highly meaningful in an extensive cohort of sports fans.

Age and gender are the key demographic factors related to responses in a similar way to that described in the previous section. Whilst women are more likely to agree that sports betting is common in broader society and amongst family members, men are more likely to indicate it is common within their peer groups. Men are also more likely to state that ‘odds talk’ is prevalent when socialising with their peers. In combination with the demographic risk factors outlined in the previous section, it is apparent that men also have different interactions with sports betting. They are more likely to agree it has a place in sports, less likely to think it is risky and can lead to other harms, more likely to have friends and peers who bet on sports, and more likely to have dialogue that supports and endorses the normalisation of sports betting. These attitudes combined suggest it is imperative public health prevention measures and harm reduction interventions target young men. The impact of peer socialisation processes and hegemonic masculine norms around sports betting have been described in previous studies (Ayandele et al., 2019 ; Bunn et al., 2019 ; Deans et al., 2017a ). Sports betting has also been related to the development of socially valorised identities for young men (Lamont and Hing, 2021 ). Our research supports and builds on these previous findings by demonstrating in a large sample that it has become a more prevalent part of sports fandom for younger adult men.

On a large and unique scale, we have demonstrated fundamental attitudinal and behavioural differences, and distinct and concerning trends, among those who engage in sports betting, thereby offering important insights about those most at risk. Most previous research has been qualitative or focused on those identified as having problematic gambling behaviours. By contrast, the scale and type of results generated from this study have afforded the ability to compare differences between non-bettors and bettors, providing compelling evidence of current issues amongst a general cohort of sports fans. Importantly, this study provides data about an emerging public health crisis in which younger men are most at risk because they are more exposed to sports betting normalisation processes, show greater engagement with sports betting and express more permissive attitudes. As such, the results of this study provide a foundation for public health interventions and programs.

One participant reported 52,000 bets per year while some others reported 5,200 and 2,600 bets per year. Such observations were treated as outliers and excluded from the analysis. This amounted to 14 observations being excluded from the analysis.

Higher order age terms are not supported by a likelihood ratio (LR) test of a model with a fourth order age term with a test statistic of 4.63 ( p -value = 0.20).

Higher order age terms are not supported by a likelihood ratio (LR) test of a model with a fourth order age term with a test statistic of 3.33 ( p -value = 0.19) for the model of the number of non-sports bets and 2.74 ( p -value = 0.25) for the model of the number of sports bets.

Armstrong, A., & Carroll, M. (2017). Gambling activity in Australia—findings from wave 15 of the HILDA survey . Australian Gambling Research Centre, Australian Institute of Family Studies.

Google Scholar  

Ayandele, O., Popoola, O., & Obosi, A. (2019). Influence of demographic and psychological factors on attitudes toward sport betting among young adults in Southwest Nigeria. Journal of Gambling Studies, 35 , 343–354. https://doi.org/10.1007/s10899-019-09882-9

Article   Google Scholar  

Browne, M., Langham, E., Rawat, V., Greer, N., Li, E., Rose, J., Rockloff, M., Donaldson, P., Thorne, H., Goodwin, B., Bryden, G., & Best, T. (2016). Assessing gambling-related harm in Victoria: A public health perspective . Victorian Responsible Gambling Foundation.

Bunn, C., Ireland, R., Minton, J., Holman, D., Philpott, M., & Chambers, S. (2019). Shirt sponsorship by gambling companies in the English and Scottish Premier Leagues: Global reach and public health concerns. Soccer and Society, 20 (6), 824–835. https://doi.org/10.1080/14660970.2018.1425682

Article   PubMed   Google Scholar  

Burke, W. J. (2009). Fitting and interpreting Cragg's tobit alternative using Stata. The Stata Journal , 9 (4), 584–592.

Cowlishaw, S., & Kessler, D. (2016). Problem gambling in the UK. Implications for health, pyschosocial adjustment and healthcare utilisation. European Addiction Research, 22 , 90–98. https://doi.org/10.1159/000437260pmid:26343859

Cragg, J. G. (1971). Some statistical models for limited dependent variables with application to the demand for durable goods. Econometrica, 39 (5), 829–844.

Deans, E. G., Thomas, S. L., Daube, M., & Derevensky, J. (2016b). “I can sit on the beach and punt through my mobile phone”: The influence of physical and online environments on the gambling risk behaviours of young men. Social Science and Medicine, 166 , 110–119.

Deans, E. G., Thomas, S. L., Daube, M., & Derevensky, J. (2017a). The role of peer influences on the normalisation of sports wagering: A qualitative study of Australian men. Addiction Research and Theory, 25 (2), 1–11.

Deans, E. G., Thomas, S. L., Daube, M., Derevensky, J., & Gordon, R. (2016a). Creating symbolic cultures of consumption: An analysis of the content of sports wagering in advertisements in Australia. BMC Public Health, 16 , 208–215.

Article   PubMed   PubMed Central   Google Scholar  

Deans, E. G., Thomas, S. L., Derevensky, J., & Daube, M. (2017b). The influence of marketing on the sports betting attitudes and consumption behaviours of young men: Implications for harm reduction and prevention strategies. Harm Reduction Journal, 14 (1), 1–12.

Dowling, N. (2014).  The impact of gambling problems on families  (AGRC Discussion Paper No. 1). Australian Gambling Research Centre, Melbourne.

Dowling, N., Oldenhof, E., Cockman, S., Suomi, A., Merkouris, S., & Jackson, A. (2019). Problem gambling and family violence: Factors associated with family violence victimization and perpetration in treatment-seeking gamblers. Journal of Interpersonal Violence, 36 (15–16), 7654–7669.

Fulton, C. (2017). Developments in the gambling area: Emerging trends and issues supporting the development of policy and legislation in Ireland. Department of Justice and Equality Report . Accessed 10 August 2021. http://hdl.handle.net/10197/8612

Gainsbury, S. M., Hing, N., Delfabbro, P., Dewar, G., & King, D. (2014). An exploratory study of interrelationships between social casino gaming, gambling, and problem gambling. International Journal of Mental Health and Addiction, 13 (1), 136–153. https://doi.org/10.1007/s11469-014-9526-x

Greene, W. H. (2018). Econometric analysis (8th ed.). Pearson Education.

Hare, S. (2015). Study of gambling and health in Victoria: Findings from the Victorian prevalence study 2014 . Victorian Responsible Gambling Foundation.

Hing, N., Lamont, M., Vitartas, P., & Fink, E. (2015). Sports-embedded gambling promotions: A study of exposure, sports betting intention and problem gambling amongst adults. International Journal of Mental Health and Addiction, 13 (1), 115–135. https://doi.org/10.1007/s11469-014-9519-9

Hing, N., Russell, A., Lamont, M., & Vitartas, P. (2017). Bet anywhere, anytime: An analysis of Internet sports bettors’ responses to gambling promotions during sports broadcasts by problem gambling severity. Journal of Gambling Studies, 33 , 1051–1065.

Hing, N., Russell, A., Thomas, A., & Jenkinson, R. (2019). Wagering advertisements and inducements: Exposure and perceived influence on betting behaviour. Journal of Gambling Studies, 35 , 793–811.

Hing, N., Russell, A., Vitartas, P., & Lamont, M. (2016). Demographic, behavioural and normative risk factors for gambling problems amongst sports bettors. Journal of Gambling Studies, 32 , 625–641.

Jenkinson, R., Sakata, R., Khokhar, T., Tajin, R. & Jatkar, U. (2020). Gambling in Australia during COVID-19 . Australian Institute of Family Studies Report, Australia.

Johnstone, P., & Regan, M. (2020). Gambling harm is everybody’s business: A public health approach and call to action. Public Health, 184 , 63–66. https://doi.org/10.1016/j.puhe.2020.06.010

Article   CAS   PubMed   Google Scholar  

Lamont, M., & Hing, N. (2021). Sports betting motivations among young men: An adaptive theory analysis. Leisure Sciences, 42 (2), 185–204. https://doi.org/10.1080/01490400.2018.1483852

Lopez-Gonzalez, H., Russell, A. M. T., Hing, N., Estévez, A., & Giffiths, M. D. (2020). A cross-cultural study of weekly sports bettors in Australia and Spain. Journal of Gambling Studies, 36 , 937–955.

McCarthy, S., Thomas, S. L., Pitt, H., Daube, M., & Cassidy, R. (2020). It’s tradition to go down the pokies on your eighteenth birthday—the normalisation of gambling for young women in Australia. Australia and New Zealand Journal of Public Health, 44 , 476–381.

McGee, D. (2020). On the normalisation of online sports gambling among young adult men in the UK: A public health perspective. Public Health, 184 , 89–94.

Milner, L., Hing, N., Vitartas, P., & Lamont, N. (2013). An exploratory study of embedded gambling promotion in Australian football television broadcasts. Communication, Politics, and Culture, 46 , 177–198.

Nyemcsok, C., Thomas, S. L., Pitt, H., Pettigrew, S., Cassidy, R., & Daube, M. (2021). Young people’s reflections on the factors contributing to the normalisation of gambling in Australia. Australia and New Zealand Journal of Public Health, 45 , 165–170.

Pitt, H., Thomas, S. L., & Bestman, A. (2016a). Initiation, influence, and impact: Adolescents and parents discuss the marketing of gambling products during Australian sporting matches. BMC Public Health, 16 , 967–979. https://doi.org/10.1186/s12889-016-3610-z

Pitt, H., Thomas, S. L., Bestman, A., Daube, M., & Derevensky, J. (2017). Factors that influence children’s gambling attitudes and consumption intentions: Lessons for gambling harm prevention research, policies and advocacy strategies. Harm Reduction Journal, 14 (11), 1–12.

Productivity Commission (1999). Australia’s gambling industries . Report no. 10. Australian government.

Pitt, H., Thomas, S. L., Bestman, A., Stoneham, M., & Daube, M. (2016b). “It’s just everywhere!” Children and parents discuss the marketing of sports wagering in Australia. Australian and New Zealand Journal of Public Health, 40 (5), 480–486.

Purves, R. J., Critchlow, N., Morgan, A., Stead, M., & Dobbie, F. (2020). Examining the frequency and nature of gambling marketing in televised boradcasts of professional sporting events in the United Kingdom. Public Health, 184 , 71–78.

Raybould, J. N., Larkin, M., & Tunney, R. J. (2021). Is there a health inequality in gambling related harms? A systematic review. BMC Public Health, 21 , 305. https://doi.org/10.1186/s12889-021-10337-3

Raymen, T., & Smith, O. (2017). Lifestyle gambling, indebtedness and anxiety: A deviant leisure perspective. Journal of Consumer Culture , 20 (4), 381–399.

Rockloff, M., Browne, M., Hing, N., Thorne, H., Russell, A., Greer, N., Tran, K., Brook, K., & Sproston, K. (2020). Victorian population gambling and health study 2018–2019 . Victorian Responsible Gambling Foundation.

Russell, A., Langham, E., & Hing, N. (2018). Social influences normalize gambling-related harm among higher risk gamblers. Journal of Behavioral Addictions, 7 (4), 1100–1111. https://doi.org/10.1556/2006.7.2018.139

Suomi, A., Dowling, N. A., & Jackson, A. C. (2014). Problem gambling subtypes based on psychological distress, alcohol abuse and impulsivity. Addiction Behaviour, 39 , 1741–1745. https://doi.org/10.1016/j.addbeh.2014.07.023

Thomas, S. L. (2014). Parents and adolescents discuss gambling advertising: A qualitative study . Victorian Responsible Gambling Foundation, Melbourne. Available from: http://www.responsiblegambling.vic.gov.au/__data/assets/pdf_file/0006/14676/Parents-andadolescents-discuss-gambling-advertising-a-qualitative-study.pdf

Thomas, S. L., Bestman, A., Pitt, H., Cassidy, R., McCarthy, S., Nyemcsok, C., & Daube, M. (2018). Young people’s awareness of the timing and placement of gambling advertising on traditional and social media platforms: A study of 11–16-year-olds in Australia. Harm Reduction Journal, 15 (1), 1–13.

Thomas, S., Lewis, S., Duong, J., & McLeod, C. (2012). Sports betting marketing during sporting events: A stadium and broadcast census of Australian Football League matches. Australian and New Zealand Journal of Public Health, 36 (2), 145–152.

Thomas, S. L., Randle, M., Bestman, A., Pitt, H., Bowe, S. J., Cowlishaw, S., & Daube, M. (2017). Public attitudes towards gambling product harm and harm reduction strategies: An online study of 16–88 year olds in Victoria, Australia. Harm Reduction Journal, 14 (1), 1–11.

Waitt, G., Hayden, C., & Gordon, R. (2020). Young men’s sports betting assemblages: Masculinities, homosociality and risky places. Social and Cultural Geography, Online First. https://doi.org/10.1080/14649365.2020.1757139

Wardle, H., Reith, G., Best, D., McDaid, D., & Platt, S. (2018). Measuring gambling-related harms: A framework for action . Gambling Commission.

Williams, R. J., Volberg, R. A., & Stevens, R. M. (2012). The population prevalence of problem gambling: Methodological influences, standardized rates, jurisdictional differences, and worldwide trends . Problem Gambling Research Centre.

Download references

Acknowledgements

We would like to acknowledge the Victorian Responsible Gambling Foundation for supporting and funding this research. We would also like to acknowledge and thank Joe Vecci and Roger Wilkins for helpful comments and suggestions.

Open Access funding enabled and organized by CAUL and its Member Institutions. This research received funding from the Victorian Responsible Gambling Foundation.

Author information

Authors and affiliations.

Social and Global Studies Centre, RMIT University, 360 Swanston Street, Melbourne, VIC, 3000, Australia

School of Business, La Trobe University, Melbourne, VIC, Australia

Buly A. Cardak

Monash University Malaysia, Kuala Lumpur, Malaysia

Matthew Nicholson

Centre for Sport and Social Impact, La Trobe University, Melbourne, VIC, Australia

Matthew Nicholson, Alex Donaldson, Paul O’Halloran, Erica Randle & Kiera Staley

You can also search for this author in PubMed   Google Scholar

Contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Emma Seal, Buly Cardak and Matthew Nicholson. The first draft of the manuscript was written by Emma Seal and Buly Cardak with guidance from Matthew Nicholson and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Emma Seal .

Ethics declarations

Conflict of interest.

The authors have no competing interests to declare that are relevant to the content of this article.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary file1 (DOCX 90 KB)

Rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Seal, E., Cardak, B.A., Nicholson, M. et al. The Gambling Behaviour and Attitudes to Sports Betting of Sports Fans. J Gambl Stud 38 , 1371–1403 (2022). https://doi.org/10.1007/s10899-021-10101-7

Download citation

Accepted : 19 December 2021

Published : 01 February 2022

Issue Date : December 2022

DOI : https://doi.org/10.1007/s10899-021-10101-7

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Sports betting
• Fan behaviour
• Sports fans
• Gambling harm
• Normalisation
• Find a journal
• Publish with us
• Track your research

Sports Betting

This article provides insights into the increasing interest in sports betting and its connection to consumer behavior, demographics, and personality traits. Understanding these dynamics is essential for businesses seeking to engage with this growing market.

• The increasing popularity of sports betting, driven by prominent platforms like FanDuel and DraftKings, is explored through data from the 2023 Prosper Media Behavior and Influence (MBI) study, encompassing 17,159 adults aged 18+.
•  Two key questions were asked in the study: “Do you gamble on sports?” and “Do you play fantasy sports?” Responses were categorized as regularly, occasionally, or never.
• 31.7% of respondents engage in sports gambling (regularly or occasionally), while 27.2% participate in fantasy sports, with a substantial overlap of 22.7% doing both.
• Within these categories, 71.7% of sports gamblers also play fantasy sports, and 83.4% of fantasy sports participants also gamble on sports.
• Approximately 40% of the 2023 MBI study population resides in states where online sports betting is illegal, despite substantial legal concerns, especially in large states like California and Texas.
• Historical data reveals a steady increase in sports gambling, particularly in recent years, while fantasy sports participation has remained relatively stable.
• In 2023, 42.1% of males reported gambling on sports compared to 21.5% of females, with both genders showing growth over the past three years, albeit at different rates.
• Demographic characteristics reveal higher sports gambling rates among Gen-Z, Millennials, and males, with connections to various occupations and a slight advantage for those with higher education.
• A classification regression tree (CRT) analysis identifies segments with high sports gambling rates, with age, gender, income, marital status, and ethnicity being significant predictors.
• Viewership patterns indicate that gambling on sports is related to specific sports categories, with minor sports and women’s sports exhibiting greater relative interest among gamblers. Additionally, sports bettors are more likely to engage in sports-related activities, favor alcoholic beverages, and display specific personality traits, including high extraversion and lower grit, demonstrating a distinct consumer profile.

Forward By Dr. Larry DeGaris, Executive Director, Medill Spiegel Research Center, Northwestern University

“Legalized sports gambling is the endgame. One-day fantasy delivers a similar fan experience to gambling, so I expect the current database of customers would provide a good foundation for sports gamblers.” I said this in 2015 when I was interviewed for the Bloomberg article “Daily Fantasy Sites Seen Positioned for Jump to Sports Gambling.”

This interview came on the heels of NBA commissioner Adam Silvers’s 2014 NYT op-ed, “Legalize and Regulate Sports Betting,” advocating legalizing sports betting; I knew it was just a matter of time before sports betting was legalized nationwide and DraftKings and FanDuel were planting the seeds.

Fast forward to 2023, and DraftKings and FanDuel are dominating the sports betting market while big sports brands like Fox and big gaming brands like MGM struggle or fail. I thought it was a good strategy. I didn’t realize how good.

The sports betting market highlights the value of customer data. DraftKings and FanDuel’s market domination is based on their customer data strategy. They bet that fantasy players, especially daily fantasy players, would be the most promising market for sports betting. As sports betting became legal in more states, they were ready. It was a big bet and a big win.

Similarly, brands are placing big bets on Retail Media Networks. The strategy is the same. As we enter the cookie-less world, customer data will become more valuable. That’s no fantasy.

More recently, Fanatics is getting into the sports betting game with its acquisition of Pointsbet, betting that its database of fans will help them compete where others have fallen short. I’m not sure I like the odds. Let’s reverse engineer DraftKings’s and FanDuel’s winning strategy. Martin Block’s nifty little sports betting paper shows us the way.

The relationship between betting on sports and playing fantasy sports is very high. The two groups are almost identical. Unlike the years it took Amazon to build a customer database, DraftKings and FanDuel had theirs built before sports betting became legal. As more states legalized sports betting, they were ready. That initial advantage has been difficult for other entrants to overcome.

Having a database of sports fans doesn’t measure up to a targeted list of fantasy sports players. Similarly, I reckon Fanatics has a great database of fans passionate about their teams, which likewise wouldn’t translate well to betting.

Digging deeper, we see the underlying mechanisms of sports betting and outline a path forward. DraftKings and FanDuel won the first round, but it ain’t over ‘til it’s over.

The women’s market in sports tends to be underserved in general because fan populations are more likely to be male. The difference in degree mistakenly is equated with a categorical difference. Sports bettors are more likely to be male, but there are a lot of women betting on sports.

Sports betting companies have been active as sponsors of sports properties. But sports betting is a fan activity and, as Martin points out, there are many product categories that index high for sports bettors. Creating a brand partnership strategy using sports betting as a platform could bring another revenue stream.

Sports bettors are highly social and competitive. That’s the underlying link between fantasy and betting. Fantasy was a natural connection for sports betting in that respect, but if sportsbooks can innovate ways to connect fans who weren’t playing fantasy, they can gain an entry point. What do you talk about when your team is having a bad year? This week’s parlays. Sports betting is a social experience for fans. Sports betting companies can create a social infrastructure for fans to interact, as they’ve done with fantasy leagues.

The story highlights both the importance and limitations of customer data. Customer data tells you who. Traditional market research can help understand the why.

We should keep this in mind as we watch retail media grow. First-party customer data is tremendously valuable but limited without a deeper understanding of shopper experiences and identities.

By Dr. Martin Block, Professor Emeritus, Northwestern University, Retail Analytics Council

Interest in sports betting is increasing with the publicity that FanDuel and DraftKings have recently enjoyed. To better understand sports betting data from the annual Prosper Media Behavior and Influence (MBI) study collected in January of 2023 (n=17,159) of adults 18+ is analyzed. Two questions were asked: “Do you gamble on sports?” and “Do you play fantasy sports?” Both questions were supplied with answer options of regularly, occasionally, and never. Table 1 shows the proportion of the sample that gamble on sports combining regularly and occasionally at 31.7%, and those playing fantasy sports at 27.2%. There is substantial overlap between gambling and fantasy sports, with 22.7% reporting both. Of those who gamble, 71.7% say they also play fantasy sports. Of those who play fantasy sports, 83.4% say they also gamble. Not all respondents say they gamble and play exclusively online. Of those who say they gamble on sports, 66.0% say they also gamble online. Of those who play fantasy sports, 84.1% say they also play fantasy sports online.

The legality of online sports betting has also been a topic of interest. Using FanDuel’s map of states where the service is legal, Table 2 shows that approximately 40% of the population as represented by the 2023 MBI are in states where it is not legal. Two big states, California and Texas are among the states where it is not legal. It should be noted that in states where it is not legal, the proportion is slightly higher than in states where it is legal. The difference, despite the very large sample sizes, is not statistically different. Legalizing online sports betting appears to have little influence.

The gambling and fantasy sports questions have been asked in the MBI for several years. Table 3 shows the trends over the last eight years for gambling, playing fantasy sports, and playing video games. Playing video games has slowly increased over the years but is showing a slight decline in 2023. Gambling on sports has steadily increased, especially in recent years. Playing fantasy sports has remained almost perfectly flat.

Considering just the last three years, as shown in Table 4 illustrates the recent growth, especially for 2023. It also shows a comparison of gambling by gender. In 2023, 42.1% of males reported gambling on sports compared to 21.5% of females. The average annual growth rate for males over the three years is 3.9%, compared to 4.0% for females.

Table 5 shows some differences by various demographic characteristics and the breakdown by regularly and occasionally gambling on sports. Looking at the characteristic variables by themselves, it appears that sports gambling is highest among Gen-Z, Millennials, and males. There seems to be strong connections to occupation, especially business owners, professionals, salespersons, workers, students, and military. All these occupations have a social component. There is also a slight advantage for higher or more education.

Identifying the Sports Bettor

Using a classification regression tree (CRT) to predict gambling on sports allows for the identification of consumer segments that have high rates of gambling on sports. Table 6 shows the results of such analysis and the important predictor variables. The most important predictor, by far, is age. This is followed by gender, income, family size, marital status, ethnicity, and population density. The model itself is reasonably strong, predicting a bettor 73.7% of the time using 10-fold cross-validation.

The tree itself identifies the segments as terminal nodes, as shown in Table 7. The highest node, those under 44, male, incomes over $106.5K, and married report 81.2% betting on sports and comprise 5.3% of the total population of adults. Slight older, ranging from 44 to 51, report 67.7% betting. Under 51, males, over $1065k   and are single (never married) report 60.9% and comprise 1.6% of adults. Males under the age of 51 with incomes between $53k and $106K living in an urban area report 64.0%, and males in suburban and rural areas report 50.6%. Males under the age of 51 with incomes less than $53K, who are non-white report 51.1% gambling on sports and comprise 6.3% of the total population. These are all the male segments above 40% of sports gambling.

The female side of the tree shows non-white with a family size of four or more at 45.3%, comprising 5.4% of the total sample. Non-white females, with a family size of less than four, and younger than 36 report 40.0% gambling and comprise 4.0% of the total sample. Among the older segments are males earning over $53K, aged 52 to 55, reporting 49.6% gambling, and African American males earning under $53K, aged 52 to 63, reporting 44.9% gambling.

Sports Program Viewing

Respondents were asked if they regularly viewed 12 different sporting events. To establish a pattern, the 12 were factor-analyzed (principal component) as shown in Table 8. The viewing patterns are arranged in three groups labeled “major,” which includes the NFL, college football, and other team sports of broad interest. The analysis shows that those who report viewing the NFL are also likely to watch college football. The second group, labeled “minor,” includes professional tennis and golf and have a narrower level of interest and are not thought of as team sports. The third group is labelled “women’s” for woman’s sports..”

Relating the viewing to gambling on sports is shown in Table 9. Overall, 73.2% of adults report viewing at least one of the categories of sports programming. The NFL is viewed by 38.4% of adults, followed by college football, major league baseball, and the NBA. Sports betting is certainly related to viewing, as in the case of the NFL, where 46.1% of the viewers compared to 34.7% of the viewers do not gamble, providing a gambling index of 120.1. The major sports have an average index of 150.7, minor sports have an average index of 176.9, and women’s sports have the highest average index of 189.8. Minor, less team-oriented sports, and especially women’s sports, are of greater relative interest to gamblers.

Weighting sports betting by regular viewing demonstrates the relative interest in the sports category, as shown in Table 11. The NFL overall leads sports betting activity, followed by the NBA, college football, major league baseball, and college basketball. All these categories are team-oriented and of broad interest.

Those who report gambling on sports are more likely to engage in a variety of sports-related activities, as shown in Table 12. Engaging in online fantasy sports and online gambling index at almost double. Sports on the radio and smartphones are higher than regularly watching sports on TV. Fox Sports indexes higher than ESPN.

Respondents were also asked about 31 different leisure activities, which are compared to reported gambling on sports, as shown in Table 13. The highest indexing activity is playing team sports. Also high are very active categories such as tennis and snow skiing. Social activities, such as tailgating and going to gambling resorts, bars, and sporting events are also high. The image of sports betting is being social, interested in competition, and physically active. On the other hand, those indexing low is generally passive and perhaps a bit more solitary, reporting activities such as watching TV, surfing the internet, crafting hobbies, and reading books. Those activities are indexed at nearly the same (not shown) as going shopping, online communities, and travel.

Product Consumption

Consumption of alcoholic drinks is commonly associated with viewing and gambling on sports. Table 14 shows this to be true as most categories of drinks index at almost double among those that gamble. The lowest index is wine.

Average monthly product spending is collected monthly, with the categories rotating across different months. The data from the monthly surveys (average n=7,500) needs to be integrated with the annual MBI, which is done using decision tree equations from available demographic variables common to both the monthly and MBI studies. Shown in Table 15 are the average reported household monthly spending across a variety of product categories. The months are taken from 2022 prior to the 2023 MBI. The averages include zero if the respondent reports no spending in that category. The highest indexing category is sporting goods, followed by men’s clothing and both fast food and full-service restaurants. It is worth noting the home improvement spending indexes highly note the characteristics of the sports gambler. Smartphone and grocery spending is almost the same. Overall, those who gamble on sports spend more on average than those who do not.

Personality

Several personality measures were also collected in the MBI including the five-factor personality inventory commonly referred to as OCEAN, grit, and impulsiveness. The personality inventory used by Prosper in the MBI is taken from the short-form version developed by the Gosling Laboratory at the University of Texas. The five factors include:

• Extraversion: Extraversion is characterized by excitability, sociability, talkativeness, assertiveness, and high amounts of emotional expressiveness. People who are high in extroversion are outgoing and tend to gain energy in social situations. People who are low in extroversion (or introverted) tend to be more reserved and must expend energy in social settings.
• Agreeableness: This personality dimension includes attributes such as trust, altruism, kindness, affection, and other prosocial behaviors. People who are high in agreeableness tend to be more cooperative, while those low in this trait tend to be more competitive and even manipulative.
• Conscientiousness: Standard features of this dimension include high levels of thoughtfulness, with good impulse control and goal-directed behaviors. Those high on conscientiousness tend to be organized and mindful of details.
• Neuroticism: Neuroticism is a trait characterized by sadness, moodiness, and emotional instability. Individuals who are high in this trait tend to experience mood swings, anxiety, moodiness, irritability, and sadness. Those low in this trait tend to be more stable and emotionally resilient. This trait is sometimes inversely reported as “emotional stability.”  The Prosper database uses the “emotional stability” terminology.
• Openness: This trait features characteristics such as imagination and insight, and those high in this trait also tend to have a broad range of interests. People who are high in this trait tend to be more adventurous and creative. People low in this trait are often much more traditional and may struggle with abstract thinking.

Table 16 shows a comparison of sports gamblers by the five personality traits. All five are statistically different. The biggest differences are gamblers are more competitive and more extravagant with less impulse control. Gamblers also have higher sociability, are more traditional, and are less likely to think abstractly. They are also more emotionally stable and resilient.

A non-cognitive predictor of success in life, such as grit, like having a clear inner compass that guides all your decisions and actions. Grit is defined as the disposition to demonstrate perseverance and passion for long-term goals (Angela Duckworth, Grit: The Power of Passion and Perseverance , Scribner, 2016). Passion is a deep, enduring knowledge of what you want. Perseverance is hard work and resilience. All three predictors are expressed as a five-point scale, with higher values being stronger on the characteristics developed from a series of five-points. Prosper used the short-scale version.

Again, shown in Table 16 is the comparison of sports gamblers. All the differences are statistically different. Gamblers have lower overall grit, meaning they are less oriented toward long-term goals. The biggest difference for gamblers is lower passion, that is knowledge of what they want. Gamblers are higher in perseverance.

Impulsivity, as measured by agreement with the statement “Live for today because tomorrow is so uncertain,” also demonstrates a difference among gamblers, as shown in Table 17. Gamblers are much more likely to strongly agree with the statement and, conversely, strongly disagree. This reinforces the lack of long-term orientation among gamblers.

The Prosper Media Behavior and Influence study from January 2023 unveils the rising interest in sports betting, emphasizing the convergence of sports gambling and fantasy sports. The analysis explores demographic predictors, revealing age, gender, income, and marital status as key factors influencing sports betting engagement. Furthermore, the study uncovers correlations between sports program viewing, personality traits, leisure activities, and product consumption, painting a comprehensive picture of the diverse profile of sports gamblers.

Help | Advanced Search

Quantitative Finance > Portfolio Management

Title: optimal sports betting strategies in practice: an experimental review.

Abstract: We investigate the most popular approaches to the problem of sports betting investment based on modern portfolio theory and the Kelly criterion. We define the problem setting, the formal investment strategies, and review their common modifications used in practice. The underlying purpose of the reviewed modifications is to mitigate the additional risk stemming from the unrealistic mathematical assumptions of the formal strategies. We test the resulting methods using a unified evaluation protocol for three sports: horse racing, basketball and soccer. The results show the practical necessity of the additional risk-control methods and demonstrate their individual benefits. Particularly, we show that an adaptive variant of the popular ``fractional Kelly'' method is a very suitable choice across a wide range of settings.

Submission history

Access paper:.

• Other Formats

References & Citations

• Google Scholar
• Semantic Scholar

BibTeX formatted citation

Bibliographic and Citation Tools

Code, data and media associated with this article, recommenders and search tools.

• Institution

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs .

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here .

Loading metrics

Open Access

Peer-reviewed

Research Article

The negative consequences of sports betting opportunities on human capital formation: Evidence from Spain

Contributed equally to this work with: Mar Espadafor, Sergi Martínez

Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review & editing

* E-mail: [email protected] (ME); [email protected] (SM)

Affiliation Department of Social and Political Sciences, European University Institute, Fiesole, Italy

Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Validation, Visualization, Writing – original draft, Writing – review & editing

• Mar Espadafor, 
• Sergi Martínez

• Published: October 27, 2021
• https://doi.org/10.1371/journal.pone.0258857
• Reader Comments

The proliferation of on-site betting shops has received enormous public attention, becoming one of the most alarming health policy issues in contemporary cities. However, there is little evidence on whether its growing presence nearby vulnerable populations produce social harm beyond its known adverse individual effects. This study provides new evidence on the negative societal effects of betting houses. Our research design takes advantage of a new wave of openings in Madrid (Spain), which created a sudden increase in the supply of on-site gambling. Using a differences-in-differences design, we find that new betting houses decline nearby high schools’ educational performance, especially in public schools in less advantaged areas. This effect is neither trivial nor diminishing with time. This evidence suggests that betting houses increase inequality of educational opportunities. The ubiquity of betting houses around vulnerable populations in multiple regions drives us to think that these findings have relevant policy implications for many countries currently designing policies tackling the increase of problem gambling.

Citation: Espadafor M, Martínez S (2021) The negative consequences of sports betting opportunities on human capital formation: Evidence from Spain. PLoS ONE 16(10): e0258857. https://doi.org/10.1371/journal.pone.0258857

Editor: Gabriel A. Picone, University of South Florida, UNITED STATES

Received: May 6, 2021; Accepted: October 6, 2021; Published: October 27, 2021

Copyright: © 2021 Espadafor, Martínez. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data sources are described within the manuscript and its Supporting information files. They are publicly available online. The data and code are held in a public repository for replication purposes. See: https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/9XQQ9S .

Funding: The authors are extremely thankful to Professor Elias Dinas, the Swiss Chair in Federalism, Democracy, and International Governance at the Social and Political Science Department of the European University Institute, for funding this project. Nonetheless, the content of this paper is solely the responsibility of the authors and does not necessarily represent the official views of the EUI nor the Swiss Chair in Federalism, Democracy, and International Governance.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Gambling is increasingly recognized as a social disease [ 1 ]. It is becoming one of the most relevant present-day addictions: the World Health Organization states that 350 million gamblers display problematic patterns each year [ 2 , 3 ]. In their recent cross-national analysis, Calado and coauthors estimate that the problem gambling affects up to 12.3% of the OECD Countries’ population [ 4 ]. For instance, 5% of the US population has a gambling problem, despite the banning of gambling in half of the states [ 5 ]. Problem gambling is considered an addiction that, in most cases, starts during the early stages of the socialization process. One evidence of this is that, for example, the UK Gambling Commission reported that 41.4% of British young adults (15–24 years old) gambled during 2016 [ 6 ].

On the potential drivers of gambling, new evidence suggests that the contextual sphere (institutions and organizations) largely explains more variation than individual characteristics [ 7 ]. Following the Conceptual Framework of Harmful Gambling, [ 1 , 8 ], the availability of different gambling opportunities is a critical factor for activating people to gamble despite its known harmfulness [ 9 ]. Accordingly, the exposure hypothesis [ 10 , 11 ] points out to the current expansion of betting houses and betting shops across cities as one of the primary triggers of the growing gambling problem [ 12 – 16 ]. This paper further approaches the consequences of gambling opportunities’ expansion to its possible societal consequences. We focus on its effects on human capital formation and educational performance. How does a change in the supply of betting houses and gambling opportunities affect young adults’ human capital formation process? Despite the importance of this question, there is little empirical evidence of gambling’s impact on societal-level outcomes. This paper enters the fray by documenting the harmful effects of betting houses on high schools’ educational performance.

Extant literature on gambling splits between the causes or individual-level consequences of problem gambling. On the one hand, a growing literature analyzes the individual characteristics that draw people to addictive behaviors [ 7 , 17 – 20 ]. It turns out that, although gambling is illegal under age 18, evidence suggests that age is an important predictor for the development of problem gambling behavior. Teenagers have access to gambling venues, and they gamble even higher proportions than adults [ 20 , 21 ]. A new high school survey on drug use among adolescents confirms previous findings for the Spanish case [ 2 ]. This unexpected accessibility to gambling for minors might be explained by the fact that it is relatively easy to impersonate identity using an adult ID for online gambling and that only very few on-site betting houses in a limited number of countries effectively ask for ID cards on entry or reward cash out [ 2 , 4 , 22 ]. In addition, adolescents tend to have more addictive behaviors due to a weaker understanding of odds and probabilities [ 23 ], and a lack of perception of gambling as a risky activity [ 24 ]. These properties highlight adolescents’ propensity and vulnerability to an increase in the supply of gambling opportunities.

On the other hand, scholarship has paid increasing attention to the consequences of gambling in adolescence since the early 2000s due to its critical role during the socialization process (e.g., [ 12 , 25 , 26 ]). Most of this research strand examines the individual-level impact of gambling on adolescents and young adults, focusing on gamblers’ social integration [ 20 ], and the mental disorders derived from gambling such as other undesirable addictions like drug consumption [ 27 – 29 ]. Authors in this field have recently looked beyond the availability of gambling facilities, to its accessibility by minors as a contributing factor of problem gambling among adolescents [ 30 , 31 ]. These studies largely focus on problem gamblers and contribute to our understanding of gambling’s underlying factors, along with the consequences of gambling on this selected population. By strictly focusing on problem gamblers, however, the societal implications remain, at large, unclear.

We believe two principal biases prevent us from extrapolating these studies’ conclusions to a broader population: the self-selection of gamblers and the accessibility to gambling facilities. First, self-selection of problem gambler’s biases, due to this group’s predisposition to gambling, any comparison or extrapolation of their behaviors to non-gamblers on any outcome. Second, and in the same vein, gambling facilities are often located in areas with more potential gamblers. Thus, previous research cannot conclude whether differences between problem gamblers and non-gamblers are due to accessibility or selection as the two groups differ in many critical confounding characteristics.

This study proposes a solution to these limitations. First, we compare the evolution in academic performance of geographically close and similar high schools, which differed in one thing—one of them was recently exposed to a new betting house while the other was not. Secondly, using a case where vulnerable populations were not more likely to be exposed to betting houses. We do so by looking at the case of Madrid (Spain), where rich available data enables us to estimate the effect of gambling opportunities on educational performance in less wealthy areas (public high schools) and neighborhoods with a higher income level (charter high schools). This choice is also motivated by the particularly intensive spread of new betting houses experienced in Madrid between 2015 and 2017—see Fig 1 , the orange-underscored area.

• PPT PowerPoint slide
• PNG larger image
• TIFF original image

Opening year of ongoing betting houses in Madrid from 1990 to 2019. Note: Census data gathered from the Madrid City Council open data portal.

https://doi.org/10.1371/journal.pone.0258857.g001

Spain decriminalized gambling during the first post-authoritarian administration, in 1977 (Decree 16/1977) but did not regulate the gambling industry until the 1980s—the first regulatory attempt coming from the region of Catalonia in 1984, followed by Galicia in 1985 [ 32 ]. Madrid passed its first gambling law in 2001, which became outdated after the internet revolution of the 2000s. For this reason, the Spanish parliament passed the first national-level law regulating gambling in 2011. Nevertheless, the somewhat vague Spanish Law 13/2011 was far from closing the debate about gambling companies’ taxation and its potential negative externalities. National authorities promoted this lacking legal definition to decentralize the responsibility (and the benefits) from gambling regulations to regional authorities. For instance, the national law specified that minors were banned from gambling facilities, but left the responsibility to regions to detail any limitation for new betting houses’ distance to high schools, hospitals, or civic centers. Surprisingly enough, the first regional gambling regulation updates were passed between 2018–2019, leaving a decade to expand and normalize such a business sector.

In any case, previous and current legal cross-regional differences resulted in variations in the restrictions imposed on gambling companies and protections offered to vulnerable groups, which partly account for the contrasting development of this sector in Madrid or Murcia compared to Barcelona or Valencia. Madrid authorities’ minor reaction to the initial spread of gambling facilities and its negative externalities allowed for short-term expansion. The government of the Comunitat Valenciana, for instance, legally restricted new betting shops openings at more than 850m from the educational center and closed existing betting houses at less than 500m from public facilities (Law DOGV no. 8834 passed on June 15th, 2020). In contrast, while Madrid’s reform also includes ID controls at betting houses’ entrances and imposes 9,000€ penalties, it only restricts new gambling licenses to requests located at more than 100m. from the closest educational center (Decree 42/2019 of Madrid’s region). Focusing on the case of Madrid, its first betting house opened in 1990, but 119 Most of these betting shops are devoted to sports betting, but also include other games of chance such as slot machine and electronic gaming machines. More have opened since then—a growth of stores that is far from linear. After the legalization of sports betting in 2006, fifteen betting houses opened their doors in Madrid before 2011. Nevertheless, the spread of gambling facilities peaked in 2015–2017, when more than twenty-five betting houses opened in the capital city. This expansion increased Madrid citizens’ exposure and accessibility to gambling facilities, and regardless of the ban on underage gambling, also its popularity among teenagers: adolescents’ gambling prevalence increased by 30% in Spain between 2016 and 2019 [ 2 , 33 ]. The ESTUDES report exalts that the share of adolescents that gambled during last year rose from 21% in 2016, to 30% in 2019 [ 2 , 33 ]. According to ESPAD’s estimation, 10% of Spanish adolescents gambled during 2019 [ 34 ].

This spread of betting shops, affecting socio-economically diverse areas, turned Madrid into an instrumental case to examine the (potentially heterogeneous) consequences of gambling on human capital formation. We leverage the shock of exposure to close-by new gambling facilities suffered by some high school students to assess its consequences on academic performance. Understanding high schools as aggregate units of individual-level behavior, which respond to contextual factors such as the supply of gambling facilities, we draw on administrative data to measure high schools’ educational performance. A more detailed explanation of the data and its sources can be found in the S1 Text . It is worth noticing that the Spanish educational system comprises three levels: compulsory education (primary and lower-secondary education), higher secondary education (academic and vocational training), and tertiary education. We selected high schools that offered the Academic Track (N = 277). This training lasts for two years and gives access to tertiary education after passing a standardized state-level exam. Therefore, our population of analysis is a selected share of high school students that seek to access university and, in most cases, are between sixteen and nineteen years old. S2 Fig the geographic distribution of betting houses (in red crosses), and high schools (public, in blue, and charter, in red circles) in Madrid by 2017.

Building upon extant literature exalting the role of gambling facilities’ availability or supply as a predictor of its consumption during adulthood and adolescence [ 34 – 38 ], we estimate the effect of exposure to new betting houses on high school educational performance using a difference-in-differences setup (DiD). In our study, the DiD estimator compares the evolution of high schools’ performance before and after some schools became exposed to new betting houses between 2015 and 2017. This design hinges on the assumption that treated and non-treated high schools followed a similar educational performance trend before the betting houses opened—also known as the parallel trends assumption. We provide evidence supporting the plausibility of this claim. Building on that, the differential evolution in the educational performance of high schools suddenly exposed to betting houses, compared to the trend followed by schools not exposed to such stores, can be attributed to the independent effect of an increase in the supply of gambling facilities.

We find that betting houses unevenly harm average grades of nearby high schools in state-level exams. This harm is only present among public high schools in low-income areas. Compared to other public high schools situated in low-income areas, we find that those high schools located less than 500m from a new betting house decrease their average grade by 0.6 points on a 0–10 scale (the average mark is 6.1). In line with the Compensatory Advantage theory, we do not find such an effect on charter schools or public high schools located in neighborhoods above the average income level [ 39 ]. According to this theory, we expect that students living in high-income areas or attending charter schools have a security net that prevents them from falling into certain behaviors such as gambling. Furthermore, we expect that gambling’s adverse effects are less detrimental for children from more advantageous families.

This study provides novel empirical evidence of the negative consequences of gambling on one outcome affecting human capital formation. Only the most vulnerable collectives suffered from the negative consequences of an increase in Madrid’s gambling supply. Unfortunately, the conditions for these results to hold—an increase in the supply of betting houses around vulnerable populations—are met in many Spanish and worldwide cities beyond Madrid. Hence, these findings have relevant policy implications for the ongoing debate on the regulation of gambling and betting houses. By filling the gap of measuring the effect of gambling on adolescents’ human capital formation process, this paper sets the basis for guiding and supporting the state or supranational-level regulation of betting houses, which, according to our results, are undermining equality of opportunities.

Materials and methods

We measure high schools’ educational performance using administrative data detailing the average grade obtained by each school at the standardized state-level exams that give access to the university. These state-level exams are comparable to, for example, the A-level exams in the United Kingdom or the SAT in the US. This exam is equal for all students in the whole region of Madrid, and it is only taken by those students that (1) successfully manage to finish the academic track and (2) wish to access university. We acknowledge that test performance is a function of both knowledge and motivation. Given that this standardized state-level exam is a “high stakes” exam, we ensure that we are taking into account “test-taking motivations”, avoiding concerns as to whether grades collected are (or not) a valid measurement of students’ performance [ 40 ]. Moreover, these exams are anonymous and marked by external evaluators, which ensures that teachers within schools are not grading on a relative curve according to average classroom performance [ 41 ]. Accessible data on outcomes restrict the temporal scope of the study to 2014–2017. Indeed, the authors only access aggregate data by neighborhood and high schools—i.e., no individual data was used. Moreover, being betting houses placement independent to high school’s grade evolution, one may consider that as a natural experiment in which we, as researchers, cannot control the treatment administration.

Identification strategy

We seek to identify should students’ performance decrease by opening a betting house from one course to the following one. This short-term effect permits us to rule out the existence of a compositional effect—i.e., students self-selecting into high schools farther away from betting houses. As the academic track lasts for two years, students start this program before public authorities issued the license for a betting house that they would be exposed to during their test year. To support this claim’s robustness, we observe a placebo outcome: the number of students that sit for the state-level exam.

The 2015–17 wave of new betting house openings in Madrid is a suitable case as it provides twofold motives for isolating gamblers’ selection from its consequences. First, prior research associates increase in gambling availability (supply) with the growth of gambling participation and its associated pathologies (demand) [ 34 – 38 ]. Building on these papers and prior reports stating the prevalence of problem gambling among adolescents, we expect an increase in gambling facilities will lead to a growth of problem gambling in those areas.

Second, comparing public (89 high schools) with charter high schools (188 high schools) would be imprecise given that the two types of schools have different schedules. While most public high school students attend classes from 8:00 to 14:30, students in a charter or private high school also have lessons during the afternoons. Furthermore, unlike public schools, charter schools are characterized by their extensive supply of extra-curricular activities right after the afternoon lessons. Therefore, exposure to leisure activities outside school is lower and more supervised among high school students attending charter high schools. For this reason, although public and charter schools are equally likely to be close to a betting house (see S4 Table ), we expect an uneven effect of betting houses across different types of schools. An increase in the gambling supply is likely not to imply an effective increase in the gambling demand among charter or private school students. Unfortunately, we do not have survey data to test such a claim. Indeed, this argument applies to public schools located in high-income areas where social models of leisure, commuting patterns, and a more monitoring parental style might prevent adolescents from bad habits like gambling [ 18 , 42 , 43 ]. To avoid comparing apples with oranges, this paper solves the other part of the selection problem by matching high schools by type and neighborhood income level.

Exposure to gambling

Using a difference-in-differences setup, we leverage the variation in high schools’ exposure to gambling facilities as the treatment. We use two different specifications capturing exposure to such treatment. First, we use individual-level survey data on commuting patterns among Madrid citizens to uncover that the average students’ distance to educational centers is 500m. See S1 Table We employ this radius to divide Madrid high schools using a dummy variable switching on those education centers exposed to a betting house at less than 500m. S4 Fig plots the evolution of treated and non-treated high schools’ density according to this criterion. Second, we employ the high school/year logged meters distance to its closest betting house as a continuous measure of exposure to gambling facilities. We utilize this specification because exposure may drop much precipitously at shorter than longer distances, where it matters less. The log transformation accordingly weights variations in small distances more than in higher numbers. S4 Table details the evolution of high schools’ average distance to the closest betting house. As expected, given the wave of new openings experienced between 2015 and 2017, high schools’ average distance to betting houses decreases by 10% during this period.

Our study is a paradigmatic example of staggered DiD—different units adopt the treatment status at different time-periods and remain exposed at least until the last time-period (see S4 Fig for illustration). In those cases, TWFE models may wrongly categorize treated observations at the control group due to not changing treatment status in t +1 or t +2 periods after being treated since t 0 . As a result, the estimand of standard two-way fixed effects models incorporates a bias, a weight that might offset the real treatment effect. We attempt to rule out this bias using the group-time average treatment effect estimator proposed by Callaway and Sant’Anna [ 44 ]. This methodology calculates and discounts this bias, controls potential heterogeneity in the treatment effect across time-periods and subgroups, and helps estimate reliable pre-treatment trends.

Indeed, the use of a Difference-in-Differences setup for assessing the effect of betting houses on academic performance may hinge on the parallel trends assumption. For control units to credibly represent a post-treatment benchmark for treated units in the absence of treatment, both groups should follow similar change rates before one group adopts the treatment. The results plotted in the first panel of Fig 2 and t −1 and t −2 of Fig 4, obtained using TWFE and Sant’Anna and Callaway’s estimators, respectively, suggest that parallel trends assumption is likely to hold in this case study. The overall average treatment effect (blue spike, first column in Fig 2 ) presents a precisely estimated zero-effect or null finding (b = 0.003, s.e. = 0.109, p = 0.978). In other words, gambling companies did not target areas or high schools with worsening academic performance.

Note: The figure shows the point estimates and robust 95% (thin) and 90% (thick) confidence intervals estimated using Eq 1 . All models include school-level and year-fixed effects. The outcome variable is the school-level average grade obtained by all the students that sit the state-level exam (mean, 6.26; SD, 0.7). The overall sample contains 235 high schools, 149 charter, and 86 public. Models on the effect of betting houses on charter schools in neighborhoods below income average use 75 high schools, and those in neighborhoods above the average income 75 schools. Models focusing on the effect on public high schools include 62 high schools (242 observations) when analyzing below-average-income areas and 25 schools (100 observations) when approaching the effect of gambling facilities on public high schools located in high-income neighborhoods of Madrid.

https://doi.org/10.1371/journal.pone.0258857.g002

Fig 2 shows the estimates for the treatment effect obtained employing Eq 1 ( Eq 1 ). This figure comprises three panels. Each panel presents the estimated treatment effect in every time-period: before betting houses opened (placebo), during the opening year, and the full year after the opening occurred. Moreover, each panel includes different specifications. Columns split the sample by types of high schools. The first column presents the results obtained using all high schools in Madrid, the overall average treatment effect. The second column compares treated to non-treated charter high schools. The third column shows the estimated effect only using public high schools. Each spike includes the point estimate and the 90% and 95% confidence intervals estimated when only using schools located in neighborhoods above (in red) or below (in black) Madrid’s average neighborhood average income level. Blue spikes plot the results of models using high schools located in both wealthier and more deprived areas (See Data description section in the S1 Text further details). There is not enough variation in the exposure to betting houses among charter high schools located in above-average to estimate its effect.

Main results

Fig 2 ’s central panel exhibits the main finding of this paper: the short-term effect of new betting house openings. The central panel’s first column shows the average treatment effect of betting houses: high schools’ performance declines by 0.25 points on a 0 to 10 scale when a betting house opens at less than 500m (b = -0.249, s.e. = 0.103, p<0.05). The second and third columns compare betting houses’ effect on charter (left) and public schools (right).

In line with our expectations, both point estimates in the second column indicate that betting houses had no immediate effect on charter schools’ performance. In contrast, betting house openings close to public high schools located in neighborhoods below the average income level decreased their average grade by 0.5 points (1 standard deviation, p<0.05) on a 0–10 scale, being more robust in public schools located in low-income neighborhoods (b = -0.578, s.e. = 0.198, p<0.05). We estimate the differential effect of novel exposure to betting houses on the public with respect to charter high schools by interacting the treatment with a dummy variable representing public high schools. S15 Fig presents the interaction term when using all schools and only those located in neighborhoods below the avg. income. The differential impact of betting houses in public compared to charter schools is of 0.4 points in a 0–10 scale.

The third panel presents the estimated effect for t +1 , the first full academic course after the betting house opened. Results essentially confirm and aggravate the short-term consequences of increasing the supply of gambling facilities. While charter schools remain entirely unaffected, betting houses occasion a 0.7 decline in poorer, public high schools academic performance (b = -0.697, s.e. = 0.215, p<0.01).

To further test our results, we also used distance as a continuous variable. The three panels in Fig 3 presents the association between high schools’ distance to betting houses and its academic performance. The two plots on the left side display descriptive figures. At the top-left corner of the Fig 3 , the first panel plots the association between high schools logged meters distance to the closest betting house (horizontal axis) and their average grade obtained at the state-level exam (vertical axis) using all high schools in Madrid between 2014 and 2017. The blue line represents the linear regression estimation, and the shaded area is the 95% confidence interval. Notice that smaller values in the horizontal axis (less distance to the closest betting house) capture higher exposure to gambling facilities, which should accordingly relate to lower educational performance.

Note: The figure includes three panels. The top-left panel shows the linear association between distance and academic performance. The bottom-left splits the top-left panel into four subgroups by neighborhoods’ income level and type of school. The right-hand side panel shows the TWFE models’ results (following Eq 1 ) but using a continuous variable of exposure to betting houses—logged meters distance.

https://doi.org/10.1371/journal.pone.0258857.g003

In this line, the linear prediction shows a positive association between distance to the closest betting house and academic success. The two panels below split the sample into four groups. While the left-side panel uses high schools located in neighborhoods below Madrid’s median income level, the right-side only uses high schools in richer areas. In each panel, high schools and the linear predicted effect of distance on grades are, in turn, divided by the type of high school: charter (in red) and public (in black) high schools.

In line with prior results, the negative association between exposure to betting houses and academic performance is mostly attributed to the most vulnerable groups—public high schools in poorer areas. This descriptive intuition is confirmed when replicating the difference-in-differences setup but using the continuous variable as the treatment. Results are shown in Fig 3 right side panel. The average short-term treatment effect of increasing distance to the closest betting house (first column) is positive but not statistically significant at any level of confidence. Distance to betting houses only shows a confidently positive and robust association with academic performance when restricting the sample to the most vulnerable populations: public high schools in poorer neighborhoods (third column, black spike; b = 0.508, s.e. = 0.303, p<0.10). Acknowledging the non-linear distribution of logs, the bottom line of these results is that decreasing vulnerable high schools’ distance to the closest betting house by 300m. decreases their academic performance by 0.5 points on a 0–10 scale.

All in all, DiD and TWFE estimators reflect that betting houses harm academic performance in public high schools located in low-income areas of Madrid. Indeed, new betting houses do not harm the academic performance of close-by charter schools. This finding can be attributed to the expected uneven effect of new betting houses. As predicted by the Compensatory Advantage theory, charter schools might prevent students from gambling by including alternative activities such as afternoon lectures and extra-curricular activities. That said, many family-level factors that in turn affect selection into schools might be the moderators causing this missing association—i.e., different social models of leisure, commuting patterns, and a more monitoring parental style [ 18 , 42 , 43 ].

Robustness checks

This section presents four different tests supporting the robustness of the main findings. The application of the Callaway-Sant’Anna estimator to correct for potential biases resulting from two-way fixed effects models in settings with dynamic adoption of the treatment; the consistency of exposure distinguishing different intensity levels—i.e., monotonicity; placebo tests using Starbucks; and finally, examining the plausibility of compositional changes as an alternative explanation using the number of students sitting in for the exam.

The Callaway-Sant’Anna estimator adapts TWFE estimators to settings with staggered treatment administration by discounting treated observations from the control group from t +1 onwards [ 44 ]. Fig 4 presents the output from using the group average treatment effect estimator proposed by these authors. The plot illustrates the estimated difference between high schools with a close-by betting house in every period—before (in red) and after public authorities issue licenses (in blue). Conclusions from previous analyses remain intact.

Note: Effect of proximity to betting houses on high school’s average mark in state-level exams. Point estimates and Confidence Intervals are obtained using a regression model that allows for effects before and after betting houses were opened.

https://doi.org/10.1371/journal.pone.0258857.g004

Suppose the argument made in this paper is valid. In that case, we should observe public high schools located closer than 500m., for instance, 200m., to show a more reliable and substantial academic decline. We test this hypothesis by dividing the treatment dummy into two: one for high schools at less than 500m. and another at less than 200m. S19 Table suggests that regardless the low number of observations falling in this category, the decrease in academic performance is slightly more pronounced in public high schools at less than 200m of new betting houses.

An alternative explanation for such a finding is that any new leisure setting may hinder high schools’ performance by distracting students. As prior research exalts, this is not the case, and betting houses may be particularly harmful, especially for vulnerable populations. We seek to confirm this expectation by replicating our DiD setup employing the variation generated by new Starbucks cafes in Madrid. S11 and S12 Figs Placebo test: Starbucks section presents the results. We find no evidence that Starbucks affected academic performance in either poorer nor richer high schools.

Another alternative explanation is that our results could be driven by changes in the composition of students and neighborhoods. For example, a potential mechanism explaining the decline in high school’s academic performance is that their most outstanding students might sort themselves into other high schools farther away from in-degradation contexts such as the one given by betting houses. Should betting houses affect educational performance but not the composition of students who take the exam, we must find the number of students who sit in for the state-level exam unrelated to having a new betting house nearby. Fig 5 shows that the association between new betting houses and the number of students is not only positive but small and indistinguishable from zero, suggesting that there is no sorting driven by betting houses. One might also think that betting houses may devalue neighborhoods in the short run, decreasing students’ academic performance. S10 Fig replicates our DiD setup using avg. parish rent price. We find no evidence that gambling companies target already impoverishing areas to open betting houses, nor that betting houses devalue an area’s rental prices in the short run.

Betting houses effect on the no. of students. Note: The figure shows a replication analysis of Figs 2 and 4 but replacing this paper’s primary outcome with the number of students sitting for the exam. Notice that an adverse effect would inform that students sort out from high schools with close-by betting houses—the negative effect on academic performance would come from a compositional change.

https://doi.org/10.1371/journal.pone.0258857.g005

This study examines the impact of sports betting houses on the educational performance of high schools. We exploit the quasi-random variation yield by high schools’ proximity to new betting houses opened in Madrid to assess whether an increase in gambling facilities’ supply affects adolescents’ performance at state-level exams. Classifying high schools by type and income level allows us to isolate the moderation or selection effect grasped by income on the impact of betting houses on educational performance. Using a Difference-in-Differences estimator, we found that, in Madrid, public high schools located in low-income areas decrease their educational performance when a new betting house opens at less than 500m. We conversely find no such effect when looking at charter high schools or other public schools located in areas above the average Madrid’s income level.

As stated above, the effects of gambling are not small nor diminishing with time. They represent meaningful changes in school performance, which generated greater inequality. Betting houses’ effect is almost double the baseline difference between public and charter schools. This confirms that gambling is most detrimental for vulnerable populations.

It is worth noticing that these effects occur despite our reliance on a positively selected share of students. First, these students have surpassed the first critical educational transition in Spain (from compulsory to post-secondary education), and most will pursue a university degree. The Spanish educational system is characterized by a mismatch between higher numbers of people with either primary or tertiary education and an insufficient population with upper secondary education [ 45 ]. Hence, most students who manage to finish Compulsory Education and enroll in higher secondary education acquire tertiary education. Second, the academic track is considered the most prestigious and demanding track within post-compulsory education. Therefore, these students are positively selected for their educational aspirations and motivations compared to students of the same age who attend vocational training or have compulsory education. That said, these findings are not exempt from limitations and assumptions. This paper essentially employs the reduced version of an instrumental design. In other words, we build on previous evidence supporting that gambling accessibility (Z), its density and distance, increases gambling consumption and problem gambling (X) [ 16 , 30 ] to assess the consequences of gambling accessibility (Z) on human capital formation (Y). However, there is no data actually to validate the instrument in our case study. Further research should attempt to fill this gap using individual or student-level data.

How do these results translate to other contexts? We expect these results to hold all over Spain, especially in big cities such as province capitals, as both scope conditions are met: vulnerable populations—public high schools in low-income areas, and handy betting houses. One of the most popular gambling companies maps all their more than 1K on-site betting houses in Spain Source: https://m.apuestas.codere.es/csbgonline/home/mapCond?=mad —Visited April 1st, 2021, and regions like Murcia, Andalucía, or Galicia seem as affected as Madrid. These scope conditions are unfortunately all too often found outside Spain as well. Italy, the UK, some US states, and other OECD countries such as Sweden are starting to recognize gambling as a public health issue [ 46 ]. In some of these countries, adolescents are one of the most affected collectives. For instance, in Italy, where betting houses are legal, more than 4% of 15–19 years old are considered “problem gamblers” with an addiction [ 47 ], and 37% of 11–16 year-olds in England and Scotland gambled in 2018 [ 48 ]. In fact, by the age of 14, more than 11% of children in the UK have gambled. Authors’ own calculations using wave 6 of The Millennium Cohort Study, which is a longitudinal study following 16,000 children born around the 2000s in England, Scotland, Northern Ireland and Wales. In Croatia, similar to many other countries in the Balkans, about 19.4% of high school students regularly bet on sports [ 26 ], and between 8 to 12% of high school students in Zagreb display risky gambling patterns [ 49 ]. All in all, evidence suggests that countries and cities which turned into a legislative liberalization of gambling and betting, such as Zagreb or Madrid, suffered from a sudden escalation of betting shops, in particular, sports betting, which increased gambling opportunities in these cities. As pointed by Ricijas [ 26 ], even though gambling is an activity legally intended only for adults, youth throughout Europe have access to some sort of game of chance before they turn 18 [ 18 , 50 , 51 ]. Beyond gambling availability, current regulations are also failing at protecting our youth from accessing gambling facilities.

These findings have critical implications for designing policies tackling the increase of unequal opportunities promoted by betting houses. Some Italian regions limit new betting houses to a minimum of 500m distance from schools. In Spain, several regions, like Murcia or Aragón, passed laws limiting the minimum distance between a betting shop and an educational center. Spain’s current debate proposes different limits for the distance from betting houses to schools, with proposed levels between 100m. and 500m. These findings back those on the most conservative side: strong abutments should be employed to prevent vulnerable populations from falling into addictive dynamics. That said, this work performs a conservative school-level estimation of these effects using Madrid’s case study, and further work is then needed to delve into the dynamics and mechanisms through which gambling supply diminishes educational performance. Examining the consequences of existing differences in advertising politics such as whistle-to-whistle or shirt sponsoring bans between Italian and British regions, as well as the (lack of) implementation of legal age for gambling could help us understand the mechanisms and help us accurately design policies approaching the spread of new addictions among young generations.

Supporting information

S1 table. madrid underage commuting summary statistics..

Note: Authors own elaboration. Data source: The 2018 Household Mobility Survey conducted by the Consorcio de Mobilidad de Madrid.

https://doi.org/10.1371/journal.pone.0258857.s001

S2 Table. Summary statistics.

Note: Data obtained from the Madrid City Council’s census and the education authorities of the Region of Madrid. The authors’ estimated high schools-betting houses yearly distances.

https://doi.org/10.1371/journal.pone.0258857.s002

S3 Table. High schools’ distance to the closest betting house.

Note: Authors’ own elaboration. Data obtained from the Madrid City Council’s census and the education authorities of the Region of Madrid. The authors’ estimated high schools-betting houses yearly distances.

https://doi.org/10.1371/journal.pone.0258857.s003

S4 Table. High schools’ likelihood of being exposed to betting houses at less than 500m.

Comparison of public and charter schools. Note: Authors’ own elaboration. Data employed originally comes from the Madrid City Council’s census and the education authorities of the Region of Madrid. The authors’ estimated high schools-betting houses yearly distances.

https://doi.org/10.1371/journal.pone.0258857.s004

S5 Table. Exposure to close-by betting houses on High schools’ average performance.

This table includes the pre-openings placebos and the actual average treatment effect. Note: This table includes the pre-openings placebos and the actual average treatment effect. Authors’ own elaboration. Data employed originally comes from the Madrid City Council’s census and the education authorities of the Region of Madrid. The authors’ estimated high schools-betting houses yearly distances.

https://doi.org/10.1371/journal.pone.0258857.s005

S6 Table. Effect of betting houses when setting the treatment one year after opening, in t +1 .

This table distinguishes schools by type and income level. Note: This table distinguishes schools by type and income level. Authors’ own elaboration. Data employed originally comes from the Madrid City Council’s census and the education authorities of the Region of Madrid. The authors’ estimated high schools-betting houses yearly distances.

https://doi.org/10.1371/journal.pone.0258857.s006

S7 Table. Effect of betting houses when setting the treatment in the BH opening year, in t 0 .

https://doi.org/10.1371/journal.pone.0258857.s007

S8 Table. Placebo test. Effect of betting houses before BH’s opening year, in t −1 .

https://doi.org/10.1371/journal.pone.0258857.s008

S9 Table. Placebo test. Effect of betting houses two years before BH’s opening year, in t −2 .

https://doi.org/10.1371/journal.pone.0258857.s009

S10 Table. TWFE models estimated effect of (logged) distance to the closest betting house on educational achievement.

This table distinguishes schools by type and income level. Note: This table distinguishes schools by type and income level. Authors’ own elaboration. Data employed originally comes from the Madrid City Council’s census and the education authorities of the Region of Madrid. The authors’ estimated high schools-betting houses yearly logged distances.

https://doi.org/10.1371/journal.pone.0258857.s010

S11 Table. Compositional change, alternative explanation: Sorting. Betting houses on the number of students sitting in for the exam.

Note: Authors’ own elaboration. Data employed originally comes from the Madrid City Council’s census and the education authorities of the Region of Madrid.

https://doi.org/10.1371/journal.pone.0258857.s011

S12 Table. Compositional change, alternative explanation: Betting houses on the no. of students one year after BH opening, in t +1 .

This table distinguishes schools by type and income level. Note: This table distinguishes schools by type and income level. Authors’ own elaboration. Data employed originally comes from the Madrid City Council’s census and the education authorities of the Region of Madrid.

https://doi.org/10.1371/journal.pone.0258857.s012

S13 Table. Compositional change, alternative explanation. Betting houses on the no. of students one year after BH opening, in t 0 .

https://doi.org/10.1371/journal.pone.0258857.s013

S14 Table. Compositional change, alternative explanation: Betting houses on the no. of students one year after BH opening, in t −1 .

https://doi.org/10.1371/journal.pone.0258857.s014

S15 Table. Compositional change. Betting houses on district rent prices in euros/m2, in t +1 .

This table distinguishes schools by type and income level. Note: Authors extracted the data on rent prices extracted from Madrid open access records, “Renta mensual de la vivienda en alquiler (€/m2 construido) por Distrito y por Trimestre”. Note: Authors’ own elaboration. Data employed originally comes from the Madrid City Council’s census and the education authorities of the Region of Madrid. The authors also extracted the data on rent prices from Madrid open access records, “Renta mensual de la vivienda en alquiler (€/m2 construido) por Distrito y por Trimestre”.

https://doi.org/10.1371/journal.pone.0258857.s015

S16 Table. Compositional change. Betting houses on district rent prices in euros/m2, in t 0 .

Note: This table distinguishes schools by type and income level. Authors’ own elaboration. Data employed originally comes from the Madrid City Council’s census and the education authorities of the Region of Madrid. The authors also extracted the data on rent prices from Madrid open access records, “Renta mensual de la vivienda en alquiler (€/m2 construido) por Distrito y por Trimestre”.

https://doi.org/10.1371/journal.pone.0258857.s016

S17 Table. Compositional change. Betting houses on district rent prices in euros/m2, in t −1 .

https://doi.org/10.1371/journal.pone.0258857.s017

S18 Table. Placebo test.

Starbucks’ openings at less than 500m. Note: Starbucks’ openings on high-schools’ educational achievement. Authors gathered the information about Starbucks openings and location from the Madrid City Council’s census. Accessible at datos.madrid.es . The authors estimated its distance to high schools.

https://doi.org/10.1371/journal.pone.0258857.s018

S19 Table. Effect of distance to the closest betting house on academic achievement, in two different levels.

https://doi.org/10.1371/journal.pone.0258857.s019

S1 Fig. Underage commuting patterns in Madrid (2018).

https://doi.org/10.1371/journal.pone.0258857.s020

S2 Fig. Geographic distribution of high schools and betting houses in Madrid (2017).

Note: This map was originally created by the authors using open geolocated data from the Madrid City Council and the education authorities of the Madrid Autonomous Community. Stamen Design, under CC BY 4.0, and OpenStreetMap are the sources of the map tiles employed.

https://doi.org/10.1371/journal.pone.0258857.s021

S3 Fig. Distribution of the high schools’ distance to the closest betting house.

https://doi.org/10.1371/journal.pone.0258857.s022

S4 Fig. Evolution of treated and control groups, when using the binary distinction—schools at less than 500m.

Plot elaborated using Kim, Rauh, Wang and Imai’s Panelmatch code. Note: Data employed originally comes from the Madrid City Council’s census and the education authorities of the Region of Madrid. The authors’ estimated high schools-betting houses yearly distances.

https://doi.org/10.1371/journal.pone.0258857.s023

S5 Fig. Results’ summary.

Note: Authors’ own elaboration. Data employed originally comes from the Madrid City Council’s census and the education authorities of the Region of Madrid. The authors’ estimated high schools-betting houses yearly logged distances.

https://doi.org/10.1371/journal.pone.0258857.s024

S6 Fig. Distance to the closest betting house and high-school’s educational performance.

Distance is computed in logs. Note: Authors’ own elaboration. Data employed originally comes from the Madrid City Council’s census and the education authorities of the Region of Madrid.

https://doi.org/10.1371/journal.pone.0258857.s025

S7 Fig. Distance to the closest betting house and high-school’s educational performance.

Analyses split by type of school and neighborhood’s average income level. Distance is computed in logs. Note: Analyses split by type of school and neighborhood’s average income level. Distance is computed in logs. Authors’ own elaboration. Data employed originally comes from the Madrid City Council’s census and the education authorities of the Region of Madrid.

https://doi.org/10.1371/journal.pone.0258857.s026

S8 Fig. Summary of the effect of increasing (reducing) distance to betting houses on educational achievement.

https://doi.org/10.1371/journal.pone.0258857.s027

S9 Fig. Summary of the compositional effect of betting houses.

https://doi.org/10.1371/journal.pone.0258857.s028

S10 Fig. Short-term effect of betting houses on rental prices.

https://doi.org/10.1371/journal.pone.0258857.s029

S11 Fig. Placebo test: Distance to Starbucks.

Association between distance to Starbucks coffee shops and educational achievement. Distance is computed in log meters. Note: Authors gathered the information about Starbucks openings and location from the Madrid City Council’s census. Accessible at datos.madrid.es . The authors estimated its distance to high schools.

https://doi.org/10.1371/journal.pone.0258857.s030

S12 Fig. Placebo test. Starbucks’ openings at less than 500m.

Effect of Starbucks’ openings on educational achievement Note: Authors gathered the information about Starbucks openings and location from the Madrid City Council’s census. Accessible at datos.madrid.es . The authors estimated its distance to high schools.

https://doi.org/10.1371/journal.pone.0258857.s031

S13 Fig. Main effect.

Effect of BH openings on HS average grade using the Callaway-Sant’Anna estimator. Note: Authors’ own elaboration. Data employed originally comes from the Madrid City Council’s census and the education authorities of the Region of Madrid.

https://doi.org/10.1371/journal.pone.0258857.s032

S14 Fig. Compositional change.

Effect of BH openings on the number of students using the Callaway-Sant’Anna estimator. Note: Authors’ own elaboration. Data employed originally comes from the Madrid City Council’s census and the education authorities of the Region of Madrid.

https://doi.org/10.1371/journal.pone.0258857.s033

S15 Fig. Differential effect of betting house on public compared to charter high schools.

https://doi.org/10.1371/journal.pone.0258857.s034

https://doi.org/10.1371/journal.pone.0258857.s035

https://doi.org/10.1371/journal.pone.0258857.s036

Acknowledgments

The authors are grateful for the advice and suggestions of professors Elias Dinas and Fabrizio Bernardi. We really appreciate Per Engzell for his excellent feedback and encouragement. We thank Vicente Valentim, Eleanor Woodhouse, Kasia Nawalejko, Nikolaj Broberg, Carlos Gil-Hernández, Guillermo Kreiman, Mauricio Bucca, Marco Cozzani, Nerea Gándara, and the participants of the SISEC 2020 Conference and the NAFS Working Group at the EUI for the valuable feedback. We are extremely indebted to Sylvie Armstrong and Daniel F. Banks for reading all the versions of this document. Finally, we are thankful to the anonymous reviewer whose comments really improved the paper.

• View Article
• PubMed/NCBI
• Google Scholar
• 2. Ministerio de Sanidad GdE. Encuesta sobre uso de drogas en enseñanzas secundarias en Espana; 2018.
• 3. Organization WH. Global Forum; 2017.
• 6. Commission G. Participation in gambling and rates of problem gambling; 2016.
• 12. Abbott MW, Volberg RA. Gambling and problem gambling in the community: An international overview and critique. Citeseer; 1999.
• 15. Williams RJ, Volberg RA, Stevens RM. The population prevalence of problem gambling: Methodological influences, standardized rates, jurisdictional differences, and worldwide trends. Ontario Problem Gambling Research Centre; 2012.
• 22. Mejías I, Orgaz C, MaCruz G, Amézaga A, Carrasco C. ¿Qué nos jugamos? Fundación de Ayuda contra la Drogadicción and Centro Reina Sofía; 2020.
• 33. Ministerio de Sanidad GdE. Encuesta sobre uso de drogas en enseñanzas secundarias en Espana; 2020.
• 47. Cerrai S, Resce G, Molinaro S. Rapporto di Ricerca sulla diffusione del gioco d’azzardo fra gli italiani attraverso gli studi IPSAD ed ESPAD Italia. 2017.
• 48. Motha H, Pye J. Young People and Gambling 2020 Technical Report; 2020.
• 50. Mokinaro S, Vincente J, Benedetti E, Cerrai S, Colasante E, Arpa S, et al. ESPAD Report 2019: Results From European School Survey Project on Alcohol and Other Drugs. 2020.
• 51. Kraus L, Nociar A. ESPAD report 2015: results from the European school survey project on alcohol and other drugs. European Monitoring Centre for Drugs and Drug Addiction; 2016.

Sports Betting Academic Research Articles

Academic Sports Betting Research is at the heart of what Sports Insights does. We are an internationally recognized leader in betting information services. Our goal is to educate sports bettors about using statistical analysis and sports betting research to help evaluate risk, when attempting to predict “winners” in sporting events. We don’t guarantee winners, or promote impossible winning percentages. Professional sports bettors operate within a winning percentage of 54-57%.

The key point to understand is that the difference between winning and losing over the long-term is measured by only a few percentage points. Anyone can go 10-0 one week, but very few sports bettors win above 54% of their bets over the course of entire season. At Sports Insights, we define success as consistently winning over the long term. This page offers sports bettors a list of published academic research articles to help you win.

Visit this page often for new betting research articles and ideas. The staff at Sports Insights is constantly researching the sports betting market, developing new and cutting edge sports betting systems . Solid betting research should always form the foundation of any serious sports betting system.

Academic Research Papers on Sports Betting

A sizable body of literature examining the “efficiencies of the sports betting marketplace” already exists. We examined some of the better known works in hopes of educating sports bettors. Note that all the authors, unlike touts or “scamdicappers”, do not guarantee winners. They also believe that the inefficiency they discover will fade out over time.

SportsInsights.com looks at the sports betting world as a unique marketplace. Unlike the stock market, the sports betting marketplace produces clear winners or losers based on measurable outcomes (sporting events). By applying some of the same economic tools and theories used in the financial world, we evaluate the sports betting marketplace for inefficiencies. By “inefficiencies” we simply mean measurable and predictable mispricing of games. Our research has uncovered profitable betting systems and strategies that exploit these measurable inefficiencies in the sports world. The most published and easily implemented of these betting systems is our signature “Bet Against the Public” or “Fade the Public” strategy. Stop buying snake oil from “scamdicappers” and start sports investing with a proven betting strategy. Take a moment to review our sports betting research and articles.

• International edition
• Australia edition
• Europe edition

The legalization of sports gambling in the US was a mistake

The Toronto Raptors’ Jontay Porter was just banned for life for violating betting rules. There will be much more of this to come

O n Wednesday, the NBA announced that Jontay Porter, a center for the Toronto Raptors, was banned from the league for life. An investigation found that the bench player disclosed confidential information to gamblers, exited a match early to influence an “over/under” wager on his stat line, and bet on games using a friend’s account.

Porter’s actions shouldn’t be trivialized. Sport is an important part of our culture – and fair competition and the integrity of results are essential to it. But the real threat to sports and the livelihoods of billions of fans lies with the leagues, special interests and media outlets integrating addictive gambling with the games we love. The profit-seeking corporate encouragement of this behavior needs to be countered with strict federal regulation before an emerging public health crisis gets even worse.

In the 2010s, the Democratic governor, Phil Murphy, and the state of New Jersey challenged the Professional and Amateur Sports Protection Act (Paspa), which prohibited new state-sanctioned sports gambling. Legal books were limited to a few grandfathered states, like Nevada. At the time, the scope of illegal sports gambling was unclear, with some putting the number at $50bn .

The US supreme court took on the case in 2018, ruling that Paspa was unconstitutional. Today, 38 states and the District of Columbia have made sports betting legal, with legislation pending in other areas. The dream of figures like the NBA’s Adam Silver, who immediately after becoming commissioner in 2014 published a New York Times op-ed advocating legalization, was fulfilled.

The early results have made billions for gambling companies, television networks, state governments, and players and owners alike. It’s been a nightmare, however, for millions of ordinary people.

When I bet on sports in high school, the process involved studying the Vegas lines in the Daily News and placing small bets with a local bookie. By college, it meant navigating to an offshore gambling site, possibly converting some money to bitcoin, and placing a wager before the start of a match. Today, technology has changed things radically: we can seamlessly place bets on our addictive smartphones and we don’t just bet before the games, we can bet on the outcome of every play, with AI models generating odds in real time.

Sports betting apps store dozens of data points on every customer: they know what you like to bet on, when to send you push notifications, and what offers can draw you back in if you haven’t gambled in a while. Like any drug, gambling activates the brain’s reward system. But most street-level dopamine-peddlers don’t have access to the power of big data.

Nor do they have marketing departments. If you’ve watched a sports game, there’s no doubt you’ve seen advertisements from FanDuel, Draft Kings, BetMGM or any number of legal sportsbooks. They feature celebrities and athletes – people like Kevin Garnett, Jamie Foxx, Kevin Hart, Patton Oswalt and the entire Manning football dynasty – encouraging you to sign up and risk your wages. Podcasts at media networks like the Ringer are dedicated entirely to betting. ESPN, owned by the conservative Disney corporation, has even gone to the extreme of hosting its own sportsbook, ESPN BET.

The efforts are paying off. Last year, Americans legally wagered $120bn on sports, up 27.5% compared with 2022. And billions more are probably still bet illegally.

This sharp increase is a reminder that legalization does not just bring black markets into the light of day – it serves to radically expand markets. In addition to the social stigma that surrounded it, the barrier to entry for sports gambling used to be knowing a bookie and being willing to wager in cash. Then it became being tech-savvy enough to navigate sketchy offshore sites. Now it’s just being 18 years old and having a smartphone and a credit card.

It’s no surprise that young people are suffering the most from legalization. According to a St Bonaventure/Siena Research survey , 39% of men and 20% of women aged 18 to 49 years old bet on sporting events. Among young men, 38% say they’re betting more than they should, 19% have lied about the extent of their betting, and 18% have bet and lost money meant for meeting their financial obligations.

Gambling helplines have naturally been flooded. A recent 60 Minutes program notes that in the five years since New Jersey legalized sports gambling, calls to the state’s service has tripled, with the largest caller demographic being between 25 and 34.

Yet there seems to be no real constituency for tight federal regulation, much less prohibition. A libertarian-influenced Republican party is happy to support free markets, no matter their corrosive social effects.

Many progressives are also more wary of punitive states than the actions of powerful corporations. Legalized gambling gives them more taxes to spend without trying to take it from the pockets of big business or wealthy individuals. Never mind that the state is forced to absorb the externalities created by legalization and that the push has facilitated a redistribution of wealth from the poor to the rich.

Advances in artificial intelligence will only make online betting more addictive in the years to come, and no doubt this environment will create more Jontay Porters in the future. Ultimately, Americans won’t be able to take on the powerful forces corrupting our culture until we decide we want to live in a society that celebrates earning money, not winning it.

Bhaskar Sunkara is the president of the Nation, founding editor of Jacobin, and author of The Socialist Manifesto: The Case for Radical Politics in an Era of Extreme Inequalities

• Sport betting
• Toronto Raptors

Most viewed

Sports betting is out of control

Gambling ads and apps are turbocharging demand for an addictive product. it’s time to reintroduce some friction into the process..

M y college roommate was a bookie.

Our sophomore year, he and another kid pooled a couple thousand dollars and took bets from our classmates using lines they pulled from ESPN.com. It was real “mom-and-pop-type stuff,” as my former roommate puts it now. Everybody knew everybody.

Only once was he ever stiffed by a customer. He knew that this client, whose dad worked at the “Worldwide Leader in Sports,” ESPN itself, could pay. But time passed and he didn’t pursue it. My roommate did well for himself, and the book closed after graduation.

We still see each other regularly.

I mention that to make clear up front that I’m not morally opposed to gambling.

Advertisement

His operation was fairly modest. The business only grew by word of mouth. And bettors were limited — by the paltry wages of on-campus gigs, by what remained of their summer internship money, and by whatever allowance mom and dad were willing to front. And, after all, there was a ceiling to the risk my buddy and his partner, hardly barons themselves, were willing to take on, lest the whole shop go bust on one bad break. The vigilance of on-campus authorities meant that everything had to be done discreetly.

But if those same bettors were in college today, they’d have much greater freedom to wager money on sports — and they’d make a much bigger target for the bookies.

In 2018, the Supreme Court’s 6-3 decision in Murphy v. National Collegiate Athletic Association struck down the 1992 congressional ban on sports betting championed by two-time NBA champion and then-US senator Bill Bradley.

What’s happened since then shouldn’t come as a surprise, because you’re living it.

Now legal in Massachusetts and 37 other states, sports betting has become deeply normalized. So has its ubiquity — unbidden odds, enticements, and celebrity-driven advertisements are plastered everywhere. And it’s not really a secret who these commercials are aimed at: men, mostly young and often suffering from an acute awareness of their own low status in society.

It’s normal for your 20s to be a period of painful growth, but these days, it seems that men are struggling more than usual. Falling rates of male post-secondary enrollment bear this out. Richard Reeves at the Brookings Institution has written a lot about how men, on average, have not responded well to changes in our economy: The outsourcing of heavy industry and increasing automation have eliminated much blue-collar work. And college-educated men, who may be finding the rat race less than it was cracked up to be, are hardly immune to feeling dead in the water. The near-term result is lots of directionless men with time and energy to spare.

Here’s where gambling comes in. And it gets its foot in the door by posing a very simple question: What if you could change your fortunes overnight?

“The truth is, you’ve won too much,” Vince Vaughn tells Tom Brady in a recent Super Bowl ad for BetMGM, before NHL legend Wayne Gretzky pops in, for good measure. “Let others have their turn.”

Similarly, a DraftKings spot featuring Kevin Hart that aired during the 2022 Super Bowl offers a guarantee: “The crown is yours.”

Neither commercial admits the possibility of failure.

Hold on a second, you might be thinking. Surely, bettors must know they’re taking a risk by definition? That’s the whole point, right? “The house always wins.”

True, except that to play in many cases is literally to win or at least not lose, because the first bets most users will place these days are genuinely risk-free promotional offers. There are also bonus bets, site credit giveaways, deposit matches, profit boosts, rewards programs, referral bonuses, and more. Every advertisement, by every company, makes such offers.

Why are the companies so generous? They consider the initial avalanche of free bets and boosted odds nothing more than what’s known as customer acquisition cost.

Inevitably, some players will take their free bets, pocket the money if they win, and quit on the spot. But many more will take their free bets and, no matter if they win or lose, find they’ve acquired the taste to play again. Then they’re off to the races.

Betting advertisements play expertly on the parts of ourselves that wish to be daring, spontaneous, and victorious. They remind me of nothing more than the 2021 cryptocurrency bonanza — also famously risk-free, until it wasn’t — and the wall-to-wall advertisements geared at getting you to dump your modest savings into $GORILLA .

“Fortune favors the brave,” declared Matt Damon in a much-maligned ad for Crypto.com, possibly intending to stir our dreams of Rome (for some reason) before speeding off in a flying saucer, or however it ends.

Anyway, if you bought crypto at that point you probably took a bath.

Gambling ads pull a lot of the same tricks. They traffic in grandeur and celebrity, and no one can dispute that they’re manipulative and expressly designed to be. It’s a little banal even to point it out, but they almost exclusively feature incredibly famous and good-looking men, mostly athletes and movie stars, who have reached the pinnacle of worldly success by having excelled at something that has nothing at all to do with wagering money on the outcome of sporting events.

The 2024 Super Bowl, both the first to be played in Las Vegas and the most-watched TV event in US history, featured only three betting ads — deliberately capped at that number by the NFL. This was clearly meant as a sop to the no-fun crowd, allowing the NFL to say “See? Gambling hasn’t taken over.”

It was also tacit recognition that gambling ads are bad or that the advertisements are beginning to annoy people and it’s desirable to have fewer of them. (Of course, the tens of millions of children watching at home were subjected to zero ads for, say, flavored nicotine or tobacco products.) Americans went ahead and wagered a record $23 billion on the outcome anyway.

Betting advertisements rely on a timeworn playbook, a simplistic and reassuring story that the world comprises thinkers and doers, losers and winners, and that the doers are the winners. With the advent of legal mobile betting, ubiquitous and never farther away than one’s fingertips, the opportunity to take large risks for dubious rewards has never been easier to come by.

As it is, men in their early 20s appear to be uniquely predisposed to taking big risks, prone to overestimating their odds of success and underestimating the costs of failure. Researchers have called it the “ young male syndrome .”

Since the act of placing a bet has now become frictionless, those who are addicted need only the smallest nudge to get back in the game.

Calls to the New Jersey problem gambling helpline have nearly tripled since 2018, and the largest demographic happens to be men aged 25 to 34. Research suggests that it may take years for a problem gambler to seek treatment, which means many of those callers have had their habits for years.

Incidentally, the industry seeks men out when they are several years younger than 25 to 34. In 2022, The New York Times reported that sportsbooks have partnered with universities including Syracuse, Michigan State, and Louisiana State to invade email listservs and inundate students with betting enticements. True to form, the industry professes safety, but many of the students receiving these offers are underage .

The 2018 Supreme Court decision was made in a vacuum. Its effect has been to allow betting companies to do far more than merely bring little bookmaking operations like my college roommate’s into the light. Mobile sports betting and constant promotions have induced new demand, completely transforming many spectators’ relationship to sports.

One of the 38 states that have legalized sports betting in some form — Kansas — has even rolled out extensive subsidies for sportsbooks, at the industry’s behest. More states seem poised to legalize betting. It’s possible that the level of betting has yet to reach its peak.

Here’s how to restore some balance

It’s easy to look at the dramatic changes that have taken place since 2018 and to try to conjure up equally dramatic solutions, like a new federal betting ban along the lines of the 1992 law.

That’s extremely unlikely to happen — state governments are not eager to forgo the billions in annual gambling tax revenue they now receive. Lawmakers should, however, stop to consider that they’ve made a trade-off, that there are social costs attached and that they’re not negligible.

What we should instead aim for is a golden mean of targeted regulation, somewhere between clamping down entirely and letting it rip as we’re doing now.

For bettors who simply can’t help themselves, public health-minded lawmakers ought to add friction to the process. Jay Zagorsky at the Boston University Questrom School of Business suggests that players no longer be allowed to fund their betting accounts instantly over Venmo and other online payment services. Instead, they’d make deposits into the betting apps only with cash that they put up at a bank or a retail seller of lottery tickets. That would allow bettors to continue playing as long as they’re in the black, but if they bottom out, they’ll have to get off the dopamine roller coaster and engage in a brick-and-mortar transaction.

In the UK, “whistle-to-whistle” bans on betting ads during games have shown some encouraging results in reducing the number of children exposed to these enticements. And in Australia, a national self-exclusion register — which stops bettors from placing wagers and prohibits companies from enticing them — has seen 18,000 sign-ups in its first six months.

It’s past time to begin experimenting with approaches that have shown results elsewhere.

The betting industry has seen exponential growth since 2018, and because gambling addiction isn’t visible in the way that drug overdoses are, there will surely be some lag before we know the true scale of harm. It’s also possible that people can suffer financial ruin even without being addicted to gambling in the clinical sense.

Doing anything at all is superior to the current approach. If we can’t return to the way things were, then we can at least make it easier for problem gamblers to help themselves and give others the chance to avoid being pulled in.

Brendan Ruberry is a journalist in New York who serves as production editor and podcast producer at Persuasion. He adapted this article from a piece that appeared on his Substack newsletter, McBrodie .

• Share full article

Advertisement

Supported by

In Latest Gambling Scandal, Some See Glimpse of Sports’ Future

The N.B.A. banned a player for life for betting on games, a practice some worry could become more prevalent with the rise of wagering on sports.

By Kevin Draper and Tania Ganguli

Bill Bradley, the basketball Hall of Famer and former United States senator known as a staunch opponent of legalized sports betting, was speaking about the topic back in January. But he might as well have been predicting the future.

“Well there hasn’t been a scandal, yet,” he said, discussing how professional sports have become ever more entwined with the gambling industry in recent years. “So the worst has been avoided, but all of the conditions are there for the untoward to occur.”

On Wednesday, the National Basketball Association confirmed the untoward had occurred, issuing a lifetime ban to Jontay Porter, a seldom-used backup forward for the Toronto Raptors. The league said Mr. Porter wagered money on his own team to lose, pretended to be hurt for betting purposes and shared confidential information with gamblers.

“There is nothing more important than protecting the integrity of N.B.A. competition for our fans, our teams and everyone associated with our sport,” Adam Silver , the league’s commissioner, said in announcing Porter’s punishment.

There are those who worry that Porter is just the tip of the iceberg across American sports, and that unless everyone — leagues, players, unions, politicians, betting companies — gets together to prevent further betting scandals, the very viability of professional sports is at risk. The Porter case was all the more unsettling because it came just weeks after baseball’s biggest star, Shohei Ohtani , was connected to a gambling scandal when his longtime interpreter was accused of stealing millions of dollars from him to pay an illegal bookmaker.

“When sports lose the perception that they’re honest, their sport dies,” said Fay Vincent, the former Major League Baseball commissioner who played a key role in barring Pete Rose, the career hits leader, from the sport for life in the 1980s because he bet on his own team’s games.

Sports leagues and gambling companies argue that betting will take place whether or not the law allows it, so legalizing and regulating it protects the games by making it much easier to identify suspicious wagers. (Gambling on sports is now legal in 38 states.) That is what the N.B.A. said happened with Porter. Suspicious wagers on a game involving Porter were brought to the N.B.A.’s attention, according to the league, “by licensed sports betting operators and an organization that monitors legal betting markets.” A few weeks later he was gone from the sport.

Porter’s agent did not respond to a request for comment.

However, if not for the significant size of the bet, it is not clear that any actions by Porter would have been detected.

About 15 people in the N.B.A.’s offices and four or five lawyers are involved in the league’s efforts to educate players about its gambling policies, and to monitor and enforce those policies. The league has relationships with private organizations that monitor gambling, such as U.S. Integrity and Sportradar, as well as state gambling regulators and betting operators, all of whom can alert the league to suspicious activity that might involve players or other league or team personnel.

The N.B.A.’s investigation found that somebody associated with Porter bet $80,000 that, essentially, he would perform poorly in a game on March 20. These kind of wagers, known as prop bets, are not directly related to the outcome of the game. Instead they are wagers on specific in-game possibilities, like whether a player will score a certain number of points. Prop bets are often combined into a single wager called a parlay. Such bets have extremely low odds, but give high payoffs if successful.

Against the Sacramento Kings on March 20, Porter played just three minutes before leaving with what the team said was an illness. The $80,000 bet on his performance by his associate would have resulted in a $1.1 million payout if the suspicious activity hadn’t been detected, the league said.

There are few sportsbooks in the country that would even take an $80,000 bet on a prop parlay, let alone one involving a player like Porter.

The N.B.A. said its investigation also found that, from January through March, Porter placed “at least 13 bets on N.B.A. games using an associate’s online betting account.” Three of the bets were multigame parlays that involved Raptors games — he did not play in any of those games — and all were bets that the Raptors would lose.

Porter was a marginal player in the N.B.A., not necessarily the type who could be guaranteed to affect whether his team won or lost. But the individualized nature of many prop bets means more players are able to have a more direct impact on whether a wager is successful. The president of the N.C.A.A. has said that he would like to ban prop bets involving college athletes.

Mr. Vincent said he was not particularly confident that the current legal apparatus around sports gambling — consisting of different league regulations and varied state laws — combined with a public mostly excited to pull out their phones and bet $10 on a game, was an effective system to prevent or catch all problematic wagers. The N.B.A., like most professional leagues, has pushed for a federal law that would regulate all sports gambling in the United States, though that does not seem likely in the near term.

“I’m 85 years old so I won’t be around, but I don’t think the next 20 or 30 years is going to be a pretty story about gambling in the sports world because the money is going to be so enormous, and wherever the money is enormous the corruption follows,” he said.

The N.B.A. spends a lot of time educating its players on the rules around betting, especially the prohibition against wagering on basketball. The league does not allow the gambling companies it partners with to offer bets on its development league, the G-League, because it does not want to open up the possibility of players making less money than those in the N.B.A. being tempted to wager on their own sport. Last year, players in the National Football L eague and the National Hockey L eague were suspended for violating betting rules.

And yet Porter, who received these trainings and earned around $2.7 million in his N.B.A. career, which began in 2019 — and whose brother, Denver Nuggets wing Michael Porter Jr., will earn $33 million this season — still risked banishment.

Jontay Porter posted often on social media about trading stock options and cryptocurrencies, and co-founded a company to teach others to do the same. Devin Mills, a professor in Texas Tech University’s Department of Community, Family and Addiction Sciences, said it was not uncommon to see those interests overlapping with sports betting.

Mills said sports betting, similar to trading stock options and cryptocurrencies, was associated “with this kind of this characteristic where individuals study and really think they can beat the system because they know the game, they know the players, and there is some sort of trend analysis.”

Trading options, speculating on cryptocurrencies and betting on sports are all activities that can now be accomplished with a few clicks on a smartphone. They have all exploded in popularity in the last few years, especially with younger men who spend a lot of time online.

It is the crux of the problem for professional sports leagues, which encourage their fans to bet at the same time that they warn their players away from it.

“Do we have to help them identify an alternative activity to stimulate their mind and emotions, so they aren’t seeking the rush through betting?” Mills said.

Jenny Vrentas contributed reporting.

Kevin Draper writes about money, power and influence in sports, focusing on a range of topics, including workplace harassment and discrimination, sexual misconduct and doping. He can be reached at [email protected] or [email protected] . More about Kevin Draper

Tania Ganguli writes about money, power and influence in sports and how it impacts the broader culture. More about Tania Ganguli

Inside the World of Sports

Dive deeper into the people, issues and trends shaping professional, collegiate and amateur athletics..

Women’s Pro Hockey League: The fledgling league is booming — except in New York, where the team is in last place . But the players haven’t given up.

Aaron Rodgers’s Achilles’ Heel: The N.F.L. great was supposed to be the Jets’ savior. But since arriving in New York, he has spent more time voicing conspiracy theories  than playing quarterback.

A Key to Knicks’ Season: Jalen Brunson, Josh Hart and Donte DiVincenzo have been buddies since college , a situation that those who study the workplace say can foster success.

The Future of College Sports: A   National Labor Relations Board testimony, now in the hands of a judge, could have wide-ranging consequences  — positive and negative — for athletes and their institutions.

Voice of Problem Gambling: Craig Carton, the bombastic sports broadcaster, shows a different side on a weekly show  that focuses on the stories of gambling  addicts like himself.

American Pizazz Meets Sumo: At Madison Square Garden, New Yorkers got a rare look at an ancient Japanese sport , cheering and booing as though they were watching a Yankees game.

Alabama committee reaches deal on lottery and electronic gaming, but no sports betting

A LABAMA (WHNT) — Bills that would create a lottery in Alabama and allow for other limited forms of gambling passed out of committee Tuesday in the State Legislature.

A conference committee in the Alabama Legislature approved two bills, HB 151 and HB152, on Tuesday that could bring a lottery and other forms of gambling to the state.

If approved by the legislature, a special election will be held Aug. 20 for voters to decide if they want a lottery and other forms of gaming.

HB151 is an amendment to the Alabama Constitution allowing for The Alabama Education Lottery to benefit education. The amendment would also allow for electronic games of chance, traditional raffles, and traditional paper bingo. The amendment bans all other forms of gambling, including sports betting, which had been the topic of heated discussion.

Table games, such as poker and blackjack would also be banned. That ban also covers all card games, dice games and games with a dealer.

HB151 would also allow for electronic games at racetrack locations in Green County, Jefferson County, Macon County and Mobile County, along with bingo halls in Green County, Houston County and Lowndes County.

The amendment would require Governor Kay Ivey to enter into negotiations with the Poarch Band of Creek Indians for a compact limited to in-person activities on land held in trust before February 6, 2024.

HB152 would set up an Alabama Gambling Commission to regulate gaming in the state. The bill will use all of the enforcement structure previously passed by the Alabama Senate but will add a lottery corporation to oversee the lottery.

The bill also sets requirements for getting a license to operate the electronic gaming facilities allowed by HB152.

All lottery proceeds will be earmarked for education purposes including scholarships, research and bonuses for retired teachers. All gambling proceeds will go to an annual supplemental appropriation in the State General Fund.

The committee, made up of Senators Greg Albritton (R-Atmore), Garlan Gudger (R-Cullman), and Bobby Singleton (D-Greensboro) and Representatives Chris Blackshear (R-Phenix City), Andy Whitt (R-Madison), and Sam Jones (D-Mobile), approved the bill on a 6-0 margin Tuesday afternoon.

Although the gambling compromise has passed out of committee, it will still need to be approved by both the House and Senate, then signed by Governor Kay Ivey.

For the latest news, weather, sports, and streaming video, head to WRBL.