A sequential mixed methods design was used to achieve the objectives of this study. First, a population study was conducted to generate a portrait of gambling practices of adult online gamblers and assess the impacts of the COVID-19 pandemic. Second, qualitative semi-structured interviews were conducted with a sub-sample of online gamblers to gather information on their lived experiences during the pandemic, including any changes in their practices and the reasons for such changes. Qualitative data were used to inform the main quantitative component of this paper.

Survey component

Study design and participants

Quantitative data were derived from the first wave of the longitudinal study [40], which aimed to document the impact of the COVID-19 pandemic on gambling habits. The target population consisted of non-institutionalized adult (18 years or older) online gamblers who gambled during the 12 and or 24 months preceding the survey, who spoke French or English and lived in private households throughout the province of Québec. The study employed a hybrid sampling technique to reach participants either by telephone or through a web panel.

Telephone sample

A probabilistic telephone-based sample was generated to produce estimates of online gambling prevalence and practices in the adult population of Québec. The sampling list included cellular phone users to compensate for the decreasing response rates observed among people reached via landline phones, and landline ownership biases observed for young adult and low-income respondents [41]. Representation among these groups was particularly important for the estimation of gambling participation, as online gambling is disproportionately popular among young adults [42].

Data were collected using computer-assisted telephone interviews over three periods (May to June 2021, July to August 2021, and September to October 2021). Private household (66%) and cellular phone (34%) numbers were selected by random digit dialing. The sample was stratified geographically according to the seven distinct regions along the rural-urban continuum based on Statistics Canada’s classifications. Consenting respondents were screened for online gambling participation, and only those who reported online gambling after February 2019 were retained for the remainder of the interview. Of those contacted via the estimated 162,126 valid phone numbers, 15,817 (9.8%) agreed to participate and reported valid responses to the online gambling participation screener. The final sample consisted of 1,381 participants (8.7%) after excluding a large proportion of ineligible non online gambling participants, and 8 participants who had incomplete answers on the gambling participation questions. Participants were compensated with an entry to win one of four $500 CAD prepaid VISA cards.

Web panel sample

Given the low expected prevalence of online gambling in the general population, we sought an additional online sample recruited through a web panel. This online sample was intended to reach online gamblers more effectively and achieve adequate conceptual coverage [43]. Notably, however, these panels are typically made up of self-selecting participants, and as a result tend to overestimate people who gamble online and those who gamble more intensely [44].

Data were collected using a web panel (Léger Marketing, Montréal) over three periods (May to June 2021, July to August 2021, and September to October 2021). The sample was selected per quota to match the age, sex, region, and linguistic distribution based on census data. In total, 8,537 people out of 37,530 (22.7%) agreed to complete the online gambling eligibility screen. The final sample consisted of 3,295 participants excluding participants who were not eligible or did not complete the survey. As compensation, participants earned credits redeemable for gift cards on the survey firm’s website. The final survey sample consisted of 4,676 participants, including 4,376 current online gamblers and 300 online gambling quitters. Given that this study focused on current online gamblers, i.e., those who gambled during the first year of the pandemic, all post-pandemic online gambling quitters (n  = 300) were removed from the analyses exploring the impact of the pandemic. The sample comprised a majority of men (65.1%), people aged between 25 and 64 years (76.8%), with a secondary (36%) or college (38.4%) education, who were employed (66.9%), married or in a common law partnership (59.9%), and reporting an average annual household income of $30,000 to $59,999 (25.6%), with an overrepresentation of people reporting an income of $75,000 and more (49.4%).

Survey measures

Online gambling profiles

Respondents were asked to report their gambling participation during the past 12 months (i.e., during the first year of the COVID-19 pandemic and the implementation of public health measures) for nine online gambling activities (e.g., lotteries, bingo, slot machines, poker, casino games excluding poker, sports betting, esports betting, and day tradingFootnote 1), and during the 12 months preceding the pandemic.

Based on these questions, gamblers were assigned to one of the three profiles: (1) post-pandemic quitters are those who gambled online the year preceding the pandemic but did not report any online gambling activity during the first year of the pandemic, (2) current gamblers including both (a) new online gamblers, those who did not report any online gambling activity the year preceding the pandemic but gambled online during the first year of the pandemic, and (b) continuing online gamblers, those who reported online gambling the year preceding the pandemic and during the first year of the pandemic. To identify current online gamblers who migrated from offline to online gambling, participants were asked to report any offline gambling participation both during the first year of the pandemic and the year preceding it.

Problem gambling status

Participants completed the Problem Gambling Severity Index (PGSI), a validated questionnaire with reasonable temporal stability that is widely used to assess the frequency and severity of past-year gambling problems [45]. The PGSI consists of nine items that are scored from 0 to 3 (“Never”, “Sometimes”, “Most of the time”, “Almost always”) and summed. Respondents were categorized as: non-problem gamblers (score of 0), low-risk gamblers (1 or 2), moderate-risk gamblers (3 to 7), or problem gamblers (8+), based on the responses they provided for the past 12 months, which corresponds to the first year of the pandemic.

Impacts of the COVID-19 pandemic on gambling patterns

Respondents were asked to report whether (1) the frequency with which they spend money, (2) the amount of money they spend, and (3) the time they spend gambling has increased, decreased or remained unaffected by the COVID-19 pandemic for each of the gambling activities they reported betting on in the past 12 months.

Reasons for change in gambling behaviours during the pandemic

Respondents were asked if they consider that, since the start of the pandemic in March 2020, they gamble online less, as much as, or more than before the pandemic, or if they started or stopped gambling online during the pandemic. Based on their response, they were directed to a list of reasons where they can choose all the answers that apply. A similar list of nine reasons was provided to respondents who declared having started or increased their gambling online, with items referring to availability/safety, relational reasons (family, spouse), feelings of isolation, finances, social interaction, and relaxation/stress reduction. A similar list of nine reasons was provided to respondents who declared having stopped or reduced their gambling online, with items referring to not wanting to gamble around family members/children, feeling as though they gamble too much, financial strain, lack of interest, and mental health. Further questions queried respondents’ sex, self-identified gender, age, level of education, income, and work and marital status.

Weighting and analytical procedure

Survey data were weighted to adjust for non-response and the cluster sampling design, as well as to align the results with the information on the Québec adult population available from the census including age, sex, education, living conditions (living alone vs. living with others), and CMA. Estimates were produced separately for the web and telephone samples given significant differences in age distribution.

Since telephone- and web-based surveys potentially suffer from different methodological limitations in estimating online gambling prevalence in the population, we have employed two estimation methods based on weighted combinations of the telephone and web data: (1) an integration weight based on the proportion of individuals that were screened for eligibility in each of the two samples (65% telephone; 35% web), and (2) an integration weight that gave the telephone sample greater influence (80% telephone; 20% web). Jointly reporting the results of these intermediate scenarios, we seek to reduce sampling biases that may be present in each when reported individually.

For estimates among online gamblers, sampling weights were calculated jointly for the telephone and online samples to adjust for non-response and the clustered sampling design, as well as to align the results with the weighted distribution of gamblers according to the telephone sample, including age, sex, education, living conditions, CMA, and profile. Propensity scores were used in logistic regression to control for socio-demographic differences in the samples.

Analyses were conducted to estimate gambling prevalence and gambling problem severity among Québec adults and to verify whether these habits changed since the declaration of the COVID-19 pandemic. Estimates were produced with 5% confidence intervals. Descriptive analyses and Fisher’s chi square tests were performed to compare groups of online gamblers using Stata SE 16.1 (StataCorp; College Station, TX).

Qualitative component

Sample selection and participants

After having completed the quantitative survey, participants were asked for their permission to be contacted to participate in the qualitative phase of the study; 2,277 agreed to be contacted (50.2%). A total of 415 participants who had gambled in the past 12 months were sent an invitation, and 98 interviews were scheduled. Of these, 96 semi-structured interviews were conducted. The sample was formed using the non-probabilistic maximum variation sampling method [46, 47], based on the following diversification criteria that are recognized as having a potential impact on the phenomenon under study: trajectories with online gambling (continuing, migratory and new online gamblers), direction of the changes in online gambling habits due to the pandemic (increase, decrease and no change), gambling activities (controlling exclusive lottery), gender and age group. The criteria were established based on self-reported responses to the survey. Among these profiles, over half (n = 58) reported having increased their gambling habits during the pandemic, 17 reported having decreased them, 16 mentioned that their gambling habits remained stable, and a small proportion were new gamblers (n = 5).

Interview grid and procedure

The interview grid explored (1) the online gambling habits (e.g., practices, benefits, harms, motivations) and their variations over the past 12 months; (2) the impact of the pandemic and health measures on online gambling practices (turning points, benefits, harms, reasons for changes, etc.); (3) the use of responsible gambling measures; and (4) and the use of help services. However, this paper reports on the impact of the pandemic on online gambling practices only.

Gambler’s lived experiences during the pandemic were explored via semi-structured interviews which took place between July and November 2021 via Zoom and lasted approximately 1.5 h. After having presented each participant with the consent form and received their verbal consent to participate in the study, the interview was conducted by one of six interviewers (i.e., one of the principal researchers and five research assistants who were in graduate programs in the fields of psychology, sociology and health science). The final qualitative sample was composed of 37 participants who identified as women and 59 as men. The mean age of the group was 48 years, within a range from 18 to 82 (SD  = 1.4); 20% of participants were aged between 18 and 34 and 40% between 35 and 54. Approximately 68% of participants were in a relationship (married, common law partnership) and 26% were single. A small proportion of participants (17.7%) lived alone. The vast majority (84%) had a post-secondary diploma. Just over half of the sample (56.3%) was employed. The median annual household income was between $75,000 and $99,000, and 13% had an income of less than $30,000.

Participants in the qualitative study reported an average of 2.5 online gambling activities (min = 1, max = 7, SD = 1.4). Three quarters bought lottery tickets online, and 43.5% purchased scratch tickets. Just over half played online casino slots and almost 22% played other online casino games. Nearly 18% played online poker and around 15% practiced day trading. Finally. around 15% of participants reported taking part in online sports betting, 12% in online bingo, and 2% in esports betting.

Data analysis

The interview content was recorded in audio format, then transcribed with the aid of automatic transcription software and validated by a research assistant. NVivo 12 software was used to carry out thematic content analysis on the data by following the method proposed by Paillé and Muchielli [48]. The themes were identified using a mixed approach, i.e., according to pre-established themes drawn from the interview guide (deductive), and according to themes that emerged directly from the participants’ words (inductive). The codebook was developed in close collaboration between the two principal researchers and the research professional responsible for the analyses. Bi-monthly meetings were held throughout the process to discuss content coding and theme development. Review and updating of the codebook continued on a regular basis throughout the analysis process to ensure the credibility of the analyses (internal validity) [46, 49].

Source link