A comparison of gambling behaviors and beliefs in rural and urban populations in Pennsylvania

In recent decades, gambling has greatly expanded globally, in particular to online markets. This has been of concern due to the potential risks of gambling, such as problem or pathological gambling, as well as links to harms such as intimate partner violence, financial or career difficulties, association with mental health problems, and substance use^1-4. Much of the literature has focused on gambling within populations generally, with very little research examining differences in rural versus urban areas. (For a limited narrative review see Tolchard (2015)⁵.)

A significant body of literature explores the contrasting life experiences of those residing in rural versus urban areas, revealing differential effects on population health. These disparities are regarding issues including smoking behaviors⁶, obesity rates^7,8, access to medical and therapeutic specialists^9-11, social connectedness¹², religiosity¹³, government aid¹⁴, levels of air pollution¹⁵, physical activity among adolescents¹⁶, out-of-school programs for children¹⁷, limited broadband internet access^18,19, transportation options^20,21, access to health-related information^22,23, and poverty rates²⁴. These differences impact outcomes, such as varying COVID-19 survival rates²⁵, mental health outcomes²⁶, mortality rates^27-30, and even homicide rates³¹. Similarly, there is reason to believe there may be differential behaviors and impacts of gambling on urban and rural populations.

Overall, there is a dearth of research examining potential differences in gambling behaviors in rural versus urban areas, and many findings are contradictory. Internationally, the results have been mixed on whether participation and problem gambling are higher in rural or urban areas. Even within individual countries, results have been mixed, such as in Australia where there have been indications of higher rates of participation and problems in rural communities by some³²and lower rates, or equal rates, by others³³. Several European countries have noted few differences in prevalence between rural and urban areas such as in Hungary³⁴, or more outside of major urban areas such as in Sweden and Norway^35,36. In the Canadian context, several provinces have demonstrated increased rates of gambling participation and problem gambling in rural areas^37,38. Looking to the US, the results have also been shown to be mixed, with some more remote rural areas demonstrating lower prevalence rates³⁹, while some rural areas have higher rates of prevalence that may reflect a greater density of gambling opportunities in these areas^40,41.

One weakness of many of these studies is that they were conducted prior to the widespread offering of online gambling and the legalization of gambling in the global landscape. In particular, an important consideration to address is that while some rural areas may have limited internet access, more recent technological advances in online gambling have expanded access to online gambling to mobile devices, making gambling participation in rural areas more accessible. Research has linked online gambling on mobile and other devices (eg laptops and smart televisions) to engagement in a greater number of different gambling activities as well as greater problem gambling⁴².

Looking to the factors that shape regional differences in gambling behaviors specifically, Vasiliadis et al⁴³suggested that the physical accessibility of gambling venues plays a complex role in gambling involvement and problem gambling. Indeed, proximity to and density of gambling opportunities has been found to be associated with gambling participation and problem gambling rates^44,45. There is a strong social gradient with gambling, where problem gambling is often higher in groups with lower socioeconomic status or for those who may be experiencing social deprivation^46,47. Similarly, online gambling allows gamblers to choose their mode of access based on convenience and when they want to bet⁴⁸. Combining these factors are findings that electronic gambling machines (EGMs) may be clustered in locations with lower area-level socioeconomic status⁴⁹.

In some instances, leisure activities are limited in rural areas, so the motivation to gamble is due to a lack of alternative activities⁵⁰. In many rural areas, gambling venues may become social hubs due to lack of competition. Complicating matters is that many of these communities may have large populations of individuals who have either drive-in drive-out or fly-in fly-out schedules, providing less structured schedules and more time to engage in gambling, but also serving as a motivator to engage in gambling with the intention of changing this lifestyle.

Study aims

Given the limited understanding of gambling behavior in comparing rural and urban areas, and the seemingly complex nature of gambling in rural areas, this study aims to further examine differences between gamblers residing in these locations. The data for this study are from the annual Interactive Gaming Assessment, a legislated annual assessment examining the impacts of legalized online gambling in Pennsylvania⁵¹, which included an oversampling of individuals living in rural areas during the second year of the report.

This exploratory study specifically aimed to determine if:

• there are any differences in the prevalence of online or offline gambling between urban and rural areas
• individuals living in urban and rural areas tend to gamble on different formats, as well as which region tends to engage in greater numbers of formats
• individuals that gamble online in urban and rural regions are more likely to present as being at risk for problem gambling
• there are distinct motivations for engaging in online gambling among individuals living in urban and rural areas
• there are different beliefs regarding the harms of gambling and the legality of gambling between individuals residing in urban and rural areas
• there are different demographic characteristics among those who gamble online and offline in urban and rural areas.

Methods

Sampling methodology

A dual-frame random digit dial survey sampled a combination of landline and cell phone numbers to produce a probability-based sample of adults in Pennsylvania. Between October 2021 and June 2022, a total of 26 403 landline numbers and 30 472 cell phone numbers were drawn, with 17 028 and 22 350 numbers remaining respectively after screening. 

To increase the likelihood of reaching potential respondents, phone calls were staggered over days of the week and times of day. With the exception of numbers that were confirmed disconnected numbers, fax machines, or businesses on the first call attempt, every number was dialed at least three times, covering each calling period: weekday daytime (Monday through Friday between 10 am and 5 pm), weekday evening (Monday through Friday between 5 pm and 9 pm), and weekend (Saturday and Sunday between 10 am and 9 pm). Voicemail messages, providing the name of the interviewer, the reason for the call, and a number for the participant to call back, were left when possible. If potential participants called back and indicated that they did not wish to be contacted further, calls to their number were discontinued. The number of call attempts to landlines was capped at 15 calls and for cell phones it was eight calls. The most recent birthday method was used to randomize the individual selected for landline calls. Cell phones were treated as belonging to an individual. All participants provided consent to participate.

A total of 154 962 calls were made to 39 179 numbers (22 349 to cell phones; 16 803 to landlines). An average of 4.59 (standard deviation (SD) 3.49) calls were made to landline numbers and 3.48 (SD 1.48) calls were made to cell phone numbers. The resolution rate (percentage of numbers in the total sample for which eligibility was determined) was 15.29% for landlines and 16.58% for cell phones. A total of 1953 individuals completed the survey; in addition, 43 individuals partially completed the survey. The overall response rate for the survey was 10.90% (12.70% landlines and 9.30% cell phones; American Association for Public Opinion Research (AAPOR) response rate 3, a measure that includes an estimate of cases of unknown eligibility⁵²), while the cooperation rate was 54.40% (59.90% landlines and 49.40% cell phones; AAPOR cooperation rate 2, the number of complete interviews as a fraction of the sum of complete, partial and non-interviews and cases of unknown eligibility)⁵². In comparison, other dual-frame random digit dial studies conducted in Pennsylvania since 2020 have reported response rates of less than 1%⁵³ and 1–4%⁵⁴. Similarly, a recent national study reported a response rate of 6%⁵⁵.

Participants

Of the 1996 individuals who either partially or fully completed the survey, 20 were excluded from analyses for not including the county that they currently reside in, and 42 surveys were excluded because of an incomplete section on either online gambling or offline gambling; leaving a final sample of 1934 participants (1021 cell phones; 913 landlines) comprising 53.1% women respondents (46.4% men; 0.3% another gender; 0.3% preferred not to answer). The average age of participants was 58.37 (SD 18.52; range 18–97) years.

Due to the unavailability of individual ZIP codes, we relied on county classifications as proxies for distinguishing between urban and rural areas. To do this, we utilized metropolitan statistical areas (MSAs) and non-MSA classifications, in conjunction with the rural and urban county definitions provided by the Center for Rural Pennsylvania⁵⁶. MSAs are those regions characterized by the presence of at least one urbanized area with a population exceeding 50 000 as well as the neighboring regions with connections to the central area. Some further delineate into micropolitan statistical areas, those regions where the urbanized area has a population ranging from 10 000 to 49 999 – though, for the purposes of this study, we will consider these regions to be included with the non-MSAs. Our hybrid classification largely aligns with the rural and urban county definitions provided by the Center for Rural Pennsylvania, which includes micropolitan areas in their list of rural counties, albeit with the inclusion of certain MSA counties as well⁵⁶. Utilizing this classification, we determined that the majority of the participants resided in rural counties in Pennsylvania (74.1%).

Among rural participants the average age was 53.34 (SD 17.71) years, 53.1% identified as women, 89.5% identified as being White only, 3.8% identified as being of Hispanic, Latino/a, or Spanish origin, 28.0% had a bachelor’s degree or higher, and 50.2% were presently employed. Comparatively, among the urban participants the mean age was 46.04 (SD 17.42) years, 54.4% identified as men, 61.3% identified as White only, 18.0% identified as being of Hispanic, Latino/a, or Spanish origin, 37.8% had a bachelor’s degree or higher, and 55.5% were presently employed.

Measures

Demographics: In addition to their current age, gender identity, and county of residence, participants were asked to indicate race, ethnicity, highest level of education completed, current employment status, current marital status, and personal income over the previous 12 months.

Online gambling participation: Participants were read a list of online gambling formats and were asked to indicate which formats they had participated in over the previous 12 months. Formats included online table games, online poker, online slots, online sports betting, online fantasy sports, iLottery, other legal online gambling, offshore sports books, offshore poker sites, offshore casinos, and other offshore online gambling. Offshore online gambling was recoded for analyses to reflect whether individuals engaged in any offshore/illegal betting, combining the four options (ie an individual who engaged in one form of offshore or illegal gambling would be coded as ‘1’ to indicate ‘presence’, as would an individual who indicated that they engaged in four types of illegal or offshore betting). A composite variable, ranging from 0 to 8, was calculated to determine the total number of online gambling formats engaged in by each individual (including any offshore betting).

Offline gambling participation: Participants were read a list of offline gambling formats and asked to indicate which they had participated in over the previous 12 months. Formats included Pennsylvania casinos, casinos outside Pennsylvania, Pennsylvania-based lotteries, Pennsylvania instant lotteries, fantasy sports, betting on sporting events, horse racing or off-track betting, private lotteries, bingo, private poker or card games, cash bashes, or other offline formats. A composite variable, ranging from 0 to 12, was calculated to determine the number of offline gambling formats engaged in by each individual.

Brief Problem Gambling Screen (BPGS): The BPGS⁵⁷ was used to assess whether individuals who engaged in online gambling in the previous 12 months had experienced any problems with their gambling. Given the primary aim of the original survey this measure was not provided to those who had not gambled online. This five-item screen for problem gambling assesses preoccupation with gambling, needing to gamble with larger amounts of money to achieve the same level of excitement, gambling longer than intended or with more money than intended, and borrowing and/or selling possessions to get money to gamble, as well as attempts to reduce or cease involvement in gambling. The measure has both high sensitivity and specificity, and a classification accuracy of 95.9%.

Online gambling motivations: Participants who had gambled online in the previous year were read a list of possible reasons that people may elect to gamble online and asked to indicate (yes/no) whether that was a reason that they had gambled online. Motivations included to win money, for enjoyment, because of the convenience, for the challenge, for competition, because of the availability, for stimulation, as an escape, because of the ability to bet at your own pace, because of anonymity, for the flexibility in stake size, for the high speed of play online, and other reasons^58,59.

Gambling attitudes measure: Personal beliefs about gambling were assessed to measure attitudes toward gambling as they influence engagement and the development of gambling problems (RJ Williams 2003, unpublished data). Participants were asked two questions on their beliefs about gambling. First, they reported perceived harms and benefits of online gambling with five response options ranging from 0 (‘the harms far outweigh the benefits’) to 4 (‘the benefits far outweigh the harms’). Second, participants reported perceptions on the legality of gambling (four response options: ‘all types of gambling should be legal’, ‘all types of gambling should be illegal’, ‘some types of gambling should be legal and some should be illegal’, and ‘don’t know/unsure’).

Data weighting

Data raking procedures were used to adjust the sample weights to match the known population characteristics of Pennsylvanian adult residents. Weights were calculated on age (18–64 years, 64–99 years, or not reported), race (White only, Black/African American only, Asian only, American Indian or Alaska Native only, other only, two or more, or not reported), ethnicity (Hispanic, Latino/a, or Spanish origin, non-Hispanic, or not reported), and gender (man, woman, some other gender, or not reported). The weighting schema achieved targets based on Pennsylvanian populations estimates⁶⁰ using complete data records. In the final sample 126 participants were missing data on one or more variables used in calculating their weights; missing data were replaced using the linear trend.

Data analysis

Statistical analyses were conducted, including checks for normality assumptions on continuous variables. Descriptive statistics were summarized as counts and percentages. Chi-squared tests (2 × 2) were used to evaluate group differences using bivariate and nominal data, while Mann–Whitney U-tests were employed for analyzing continuous data, due to the violations of normality.

Ethics approval

This research was declared exempt by the Pennsylvania State University Institutional Review Board.

Results

Online gambling

There was a significantly higher prevalence of online gambling in urban (14.1%, 95% confidence interval (CI) 11.41–16.70%) compared to rural regions (8.8%, 95%CI 7.25–10.38%), χ² (degrees of freedom (df) 1, n=1935)=12.713, p<0.001, odds ratio (OR) 1.69. Looking specifically at the rate of illegal or offshore online gambling, there was no difference between those in urban (2.9%, 95%CI 1.62–4.17%) than those in rural (2.0%, 95%CI 1.22–2.76%) regions, χ² (df 1, n=1933) =1.519, p=0.218, OR 1.46. Regarding the number of online gambling formats engaged in by people who do gamble online, we found that those in rural regions engaged in significantly more online gambling formats (median 2.00) than those living in urban regions (median 1.00), as determined by Mann–Whitney U-test, U=3169, p<0.01. Table 1 shows a detailed breakdown of the reported engagement in each online gambling format by region. Online table games, online poker, online slots, and other online gambling were all significantly more popular among online gamblers in rural regions than in urban regions. There was significant offline gambling participation among online gamblers, with 90.9% of online gamblers in urban regions and 89.5% of online gamblers in rural regions also gambling offline in the past year; there was no significant difference in these rates, χ² (df 1, n=206)=0.073, p=0.788, OR 1.13.

The median age of those who gamble online in urban regions (33.00 years) was not significantly different than that in rural regions (median 34.64 years), U=4437.00, p=0.555. There was no significant difference between those in urban versus rural regions who gamble online in terms of gender, race, ethnicity, marital status, or income (Table 2). Individuals who gamble online in both regions were predominantly men, White, not currently living with a partner, and the majority reported a household income exceeding US$50,000 (A$66,500) per year. There were significant differences in the educational attainment of online gamblers based on region, with online gamblers living in rural regions having lower levels of education than online gamblers living in urban regions, p<0.001.

Examining differences in motivations to gamble online between those in urban versus rural regions, there was no significant difference in the number of reported motives between those in urban regions (median 9) and those in rural regions (median 7), U=2918.00, p=0.057. An examination of the endorsement of specific motives to gamble online indicated that those living in urban regions were significantly more likely to report that their motivation to gamble online was to win money, for stimulation, because of the ability to bet at your own pace, and for the flexibility in stake size compared to those in rural regions of the state (Table 3).

When examining differences in scores on the BGPS, we found no significant differences in the proportion of individuals who gamble online and score 1 or higher on the BPGS. Specifically, 38.5% of those living in rural regions and 33.3% of those living in urban regions scored 1 or higher, χ² (df 1, n=205)=0.564, p=0.453.

Offline gambling

There was a significantly higher prevalence of offline gambling in urban (72.9%) versus rural (66.6%) regions, χ² (df 1, n=1935)=8.078, p<0.01. Examining the number of offline gambling formats engaged in by people who do gamble offline, we found that those in rural regions engaged in significantly more offline gambling formats (mean 2.69, SD 1.71; median 2.00) than those living in urban regions (mean 2.40, SD 1.49; median 2.00), U=107 586.00, p<0.01. Table 4 shows a detailed breakdown of the participants who reported engagement in each offline gambling format by region. Among offline gamblers, urban residents reported significantly more engagement in gambling in Pennsylvania-based casinos and offline fantasy sports betting than rural residents, while PA Lottery, PA Instant Lottery, private lotteries, cash bashes, and other forms of offline gambling were reported significantly more often among offline gamblers in rural regions than in urban regions.

The age of those who gamble offline in rural regions was significantly greater than in urban regions (rural median 55.00 years v urban median 44.00 years), U=99 634.00, p<0.001. There was no significant difference between those in urban versus rural regions who gamble offline in terms of education or income (Table 5). In rural regions, offline gamblers were significantly more likely to be women (p<0.01) and to be White (p<0.001) than those in urban regions. Offline gamblers in urban regions were more likely to be of Hispanic, Latino, or Spanish origin (p<0.001); they were also more likely to be currently living on their own (p<0.01) than offline gamblers living in rural regions.

Gambling attitudes

Table 6 details the reported attitudes toward online gambling for urban and rural participants. There were no reported differences in perceptions about the harms of online gambling. On the other hand, rural participants were less likely to endorse that all forms of gambling should be legal (p<0.01).

Discussion

To date, little research has been devoted to the examination of differences in gambling between individuals who live in rural and urban communities. This study examined differences in gambling behaviors and beliefs among those living in urban versus rural regions in Pennsylvania. In particular, we examined differences between urban and rural residents in (1) prevalence of online and offline gambling, (2) total engagement and types of formats of online and offline gambling, (3) rates of problem gambling associated with online gambling, (4) number and motivations for engagement in online gambling, (5) attitudes toward gambling, and (6) demographic difference in online and offline gambling.

In regard to prevalence of online and offline gambling, prevalence rates of engagement, both online and offline, were higher in urban areas than they were in rural ones. Urban residents were 1.7 times as likely to engage in online gambling and 1.4 times as likely to engage in offline gambling. In Pennsylvania, there are greater opportunities to gamble offline, with all Pennsylvania casinos located within urban areas. In terms of online gambling, there may be more restricted access to broadband connections, with 81% of rural residences lacking reliable internet services⁶¹. Interestingly, there were no differences in engagement in illegal or offshore online gambling where those living in rural communities may have been expected to have higher rates of engagement in illegal or offshore gambling due to being further away from legal gambling opportunities in metropolitan areas.

In regard to total engagement and types of formats of online and offline gambling, while urban residents were 1.6 times more likely to report online gambling and 1.1 times more likely to report offline gambling than those living in rural regions, rural residents engaged in a greater variety of formats of both online and offline gambling. This may be of particular concern due to findings that high involvement in gambling has been found to be positively related to problem gambling; that is, those that participate in many types of gambling are more likely to experience a gambling problem^62-65. This finding is analogous to findings in Australia that although there were lower participation rates in gambling, rural residents tended to spend larger amounts compared to those in urban areas – suggesting this greater level of involvement among those that do gamble in rural areas³². Rural residents were more than 95% as likely to engage in other forms of online gambling, 60% more likely to engage in online slots, 57% more likely to engage in online poker, and 49% more likely to engage in online table games than those who gamble online and live in urban regions. This engagement in online slots may be of particular concern, with slot play being consistently found to be predictive of problem gambling status^62,66,67. The online table engagement may also be concerning due to findings of increased problem gambling rates among those playing online roulette and online blackjack⁵⁹.

There were several differences with respect to engagement in different offline gambling formats. Those living in urban areas who gamble offline were more likely to gamble at Pennsylvania-based casinos and to engage in offline fantasy sports gambling than offline gamblers who reside in rural areas. This was unsurprising due to the distribution of casinos being exclusively in urban areas in Pennsylvania, giving those in urban areas easier access to this form of gambling. Offline gamblers in rural areas were more likely to engage in all forms of lottery (PA Lottery, PA Instant Lottery, and private lotteries), as well as to engage in cash bashes and other forms of gambling than offline gamblers in urban areas. This finding suggests that greater access to in-person gambling formats may increase the likelihood of engaging in gambling behaviors, and this should be monitored as other types of in-person gambling formats (eg video game terminals) become available in all areas.

In relation to rates of problem gambling associated with online gambling, we found that while more individuals who live in rural areas and gamble online tended to score a 1 or higher on the BPGS than those living in urban areas who gamble online, there were no significant differences in scores between those living in either area. It should be noted that this was only a problem gambling screen, and not a full assessment of problem gambling that categorizes individuals in multiple ways (eg recreational gambler, at-risk gambler, problem gambler), so it is not known whether there are differences in problem gambling prevalence between the two areas. Looking at the existing literature regarding problem gambling in rural versus urban areas, at least among Veterans Affairs service users, living in rural areas presented less risk of problem gambling³⁹, though gambling has changed significantly following the timing of this study, in particular the expansion and legalization of online gambling. In other studies, within the US, access to opportunities was often confounded with location, with the number of opportunities to gamble within locations being predictive of gambling problems⁴¹. Despite there being no significant differences in scores on the BPGS, the increased number of formats engaged in by those living in rural areas, the preference for certain online gambling types (ie online slots and online tables) by rural online gamblers, and the increased (though non-significant) number of individuals who score 1 or higher on the BPGS, there are concerns about potentially higher rates of problem gambling or riskier gambling in these rural areas.

Our fourth aim was to examine whether there are different motivations for engaging in online gambling among individuals living in urban and rural areas. We found that those living in urban areas were more likely to endorse motivations to gamble online as being to win money, for stimulation, because of the ability to bet at your own pace, and for the flexibility in stake size. The motivations for winning money or stimulation may be considered enhancement motives. In the literature it has been proposed that motivations can be for enhancement, coping, monetary, recreation, or social motives^68-71. Some literature has found that enhancement motives have been found to consistently predict greater gambling behavior, while both coping and enhancement motives predict gambling problems⁷¹. Looking at whether there are differences in motivations between men and women, it was found that enhancement motives are a better predictor of problem gambling in men, while coping motives are a better predictor among women⁷¹. With online gambling being more common among men, and motivations only being assessed among the online gamblers, seeing these motivations commonly endorsed is not surprising. Other studies have found that all motivations other than monetary motives are associated with engaging in greater numbers of gambling types, and that coping and enhancement motives are particularly associated with problem gambling severity⁶⁹. What this suggests is that there may be a greater vulnerability to problem gambling among those living in urban areas due to increased endorsement of these motivations, though there were no significant differences in the endorsement of other motives that may be classified as coping or enhancement types.

In examining differences in beliefs regarding the harms of gambling and the legality of gambling between individuals residing in urban and rural areas, there was only one notable difference. In terms of beliefs about the harms and benefits of gambling, there were no significant differences – individuals in both urban and rural areas tended to believe that the harms of gambling outweigh the benefits. In terms of beliefs about the legality of all forms of gambling, it was found that in rural areas individuals were less likely to believe that all forms of gambling should be legal compared to those living in urban areas. A 2003 survey of rural Pennsylvanians found that 39% did not approve of legalized casino gambling, suggesting that approval has increased following legalization and gambling expansion⁷².

Finally, we did find several differences between those gambling online and those gambling offline who reside in urban and rural areas. Among online gamblers, it was found that the demographic profiles were largely the same; however, rural online gamblers tended to be less educated than online gamblers living in urban areas. For offline gambling, there were a number of demographic differences between rural and urban areas. There was less of a gender difference in gamblers in both regions; however, in rural areas a significantly greater proportion of women took part in offline gambling than those in urban areas. Offline gamblers in rural areas also tended to be older than those in urban areas, and the majority of offline gamblers in rural areas were White, while there were ethnically more Hispanic, Latino, and Spanish origin individuals who gamble offline in urban areas. Many of these demographic differences associated with gambling behaviors can be explained by existing differences between those living in rural versus urban areas; the average age of those living in rural areas was significantly greater than in urban ones, a greater proportion of women lived in rural areas than in urban areas, those living in rural areas were more likely to endorse their race as being only White than those in urban areas, there were a greater proportion of participants of Hispanic, Latino/a, or Spanish origin in urban areas, and educational achievement was higher in urban areas than in rural ones. 

Limitations

A limitation of this study already alluded to is the lack of a full assessment of problem gambling with the study due to time constraints with the phone survey. This limits our ability to determine whether there is any difference in the prevalence of problem gambling between the two regions. While we can see that riskier behaviors are being engaged in, in terms of a greater number of formats being engaged in by both online and offline gamblers in rural regions, the problem gambling screen revealed no significant differences between the two groups, and perhaps a full assessment that better delineated between recreational, at-risk, and problem gamblers (such as the Problem and Pathological Gambling Measure)^73,74 would have found differences between these two groups. In addition, the problem gambling screen was only administered to those who had gambled online in the previous year, restricting the ability to determine whether there were any differences among those who gamble strictly offline; this was due to the original legislative motivation of the assessment, which was to assess the impacts of legalized online gambling in Pennsylvania, leading to a focus on gambling problems as they relate to online gambling. A second and related limitation is that the survey did not include full assessments of the frequency and expenditure for specific gambling formats, or even gambling generally, which again was due to restrictions in the length of the phone survey. While the number of formats individuals engage in is a demonstrated indicator of gambling involvement, as well as potential risk of problem gambling^62,63, having this supplementary information would be valuable. Here too, motivations were limited to only the online gamblers and only motivations for gambling online. There may be differences in motivations, though prior literature has found those in rural communities to generally have the same motivations to gamble as those in non-rural communities⁵⁰.

A third limitation was our use of telephone surveys, which, although we used both landlines and mobile numbers, had a low response rate of 10.90%, reflecting steadily declining response rates to telephone surveys⁷⁵. To better approximate the Pennsylvania population, we weighted the data, based on gender, age, race, and ethnicity; however, it cannot be ignored that those who did participate in the survey may differ from those who did not in some significant ways. In addition, this use of phone surveys may have led to more pronounced response biases such as social desirability, in which participants may have responded in ways that they felt would seem more desirable to the interview or just (in their belief) generally more desirable than would be found if web-based survey methods were used⁷⁶. Also, participants may have felt their responses would not be anonymous, which may have impacted decisions to answer specific questions, or even the entire survey. Additionally, this limited our final sample size, restricting the ability to better analyze certain subgroups including looking at potential differences between our mixed-mode gamblers (those who gamble online and offline) and our exclusively online gamblers from both urban and rural regions. Prior literature has found that most who gamble online tend to be mixed-mode gamblers, with only a small proportion being exclusively online gamblers^77,78, though there are some differences between these gambler types, particularly with regard to problems and intensity, with mixed-mode gamblers tending to see higher rates of problems and gamble more intensely.

Finally, the survey did not include questions enquiring about other potential differences that may exist between the regions. It would have been valuable to have included assessments regarding such factors as mental and physical health, substance use, religiosity, and even individual ZIP codes to better determine the socioeconomic status of areas that individuals reside in, to better determine whether any differences in gambling behaviors may be accounted for due to differences in these other determinants of behaviors.

Conclusion

This study has a number of important implications for policy and practice. It is clear that gambling online and offline is present in both urban and rural regions. In addition, this study revealed that, at least within Pennsylvania, those in rural regions who do gamble tend to engage in gambling on a greater number of gambling formats both online and offline, which has been demonstrated to be associated with problem gambling^62,63. However, for those in rural areas, it is often the experience that there are impediments to receiving help, due to issues such as the experience of social familiarity with the service providers or fears of being stigmatized when seeking help⁵⁰. Furthermore, these communities are less likely to have many specialized services that can be accessed by those who may be experiencing a problem with gambling. Self-help methods are often less available in these areas as well; for example, within Pennsylvania only 49 listed in-person Gamblers Anonymous meetings are available, with only a single meeting available in a rural area⁷⁹. In recent years, online meetings have become more available and popular; however, this option has been met with mixed results, with some developing new recovery strategies, while others have experienced a breakdown in the group dynamic and behaved in a less collectivistic way⁸⁰. In some rural areas, access to this service may be limited due to lack of access to the high-speed broadband internet connections that may be required for telehealth-type treatment services^18,81. In-person services should be made more available in these rural areas, including specialized training for GPs in identifying possible signs of problem gambling and including screening in their general assessments.

Looking to policy and prevention, the placement of messaging about gambling (both advertising and prevention messages) as well as the placement of gambling venues (including the placement of individual machines as well as gambling-dedicated venues such as casinos) should be considered. With the placement of gambling venues, several studies have found that a greater density of gambling availability (in particular EGMs), perceived or actual, is associated with higher problem gambling severity rates^45,74. In rural areas with low populations this perception of availability may be exacerbated by limited options for social gathering, where the limited local shopping and dining options may also be locations that house gambling opportunities⁵⁰. The often lower socioeconomic status of these areas may also make these individuals more vulnerable to gambling problems, and increased concentration of gambling machines could expose an already vulnerable population to an augmented risk of harm⁸². Finally, the provision of school-based prevention should be made equally available in both urban and rural areas, providing equal opportunity to all youth to receive information about gambling, including how to identify which behaviors may constitute gambling and the potential risks associated with more involved gambling.

Funding

Funding for this research was provided by the Pennsylvania Gaming Control Board through the Pennsylvania General Assembly Act 42 of 2017, relating to utilizing revenues generated from interactive gaming licensees.

Conflicts of interest

The authors have no conflicts of interest to declare.