Financial Literacy and Gambling Behaviour: A Systematic Review · O'Connor F, Ivory N

Results

The search identified 115 papers. Seventy-six papers remained for abstract and title screening following removal of duplicates. Forty-seven papers were excluded following title and abstract screening and a further 12 removed following full text review, resulting in 9 articles retained as meeting the inclusion criteria (see Fig. 1). It should be noted that two studies published by Amonhaemanon may have been based on the same dataset as the sample sizes are identical, although we could not find any commentary to that effect in either publication. As such, these studies will be discussed as if they have been conducted on independent samples. Year of publication ranged from 2018 to 2023, with the majority of articles (89%) published since 2021. Of the nine articles reviewed, three were conducted in the United States of America, two in Thailand, and one each in Austria, China, Italy, and Japan. Sample sizes ranged from 627 to 26,566. Six studies reported the mean age and standard deviation of the sample and a further two reported percentages within age ranges. One study provided no information regarding age of the sample. Most of the studies recruited adults; with the sample mean ages ranging from the mid to late 40s. One study was conducted on adolescents. Information relating to sex/gender was not reported in all papers but was able to be sourced for all studies from online supplementary materials where necessary. The proportion of males in the samples ranged from 27 to 83%. Further sample characteristics of the included studies are summarised in Table 1.

Key Findings: Associations between Financial Literacy and Gambling

While the research findings were somewhat mixed, the majority of studies found increased financial literacy was negatively and significantly associated with gambling behaviour (n = 6; Amonhaemanon, 2022; 2023; Becchetti et al., 2018; Cho, 2022; Wang et al., 2023; Watanapongvanich et al., 2021). Two of these studies found this negative association between financial literacy and gambling when using education level as an instrumental variable to address potential endogeneity between financial literacy and gambling behaviour (i.e., Cho, 2022; Watanapongvanich et al., 2021). An additional two studies found a negative relationship between financial literacy and gambling which did not reach statistical significance (Razen et al., 2021; Watanapongvanich et al., 2022), while only one of the nine studies found a positive and statistically significant relationship between financial literacy and gambling behaviour (Grable et al., 2021). No studies found a positive but not significant association between financial literacy and gambling.

Measurement of Financial Literacy

The majority of studies (n = 8) measured actual or objective financial literacy with items that test knowledge of financial literacy concepts (i.e., Amonhaemanon, 2022; 2023; Becchetti et al., 2018; Cho, 2022; Razen et al., 2021; Watanapongvanich et al., 2021; 2022; Wang et al., 2023). All but one of these studies (Wang et al., 2023) drew upon Lusardi and Mitchell’s research, using either the Big 3, the Big 5, or some combination therein, sometimes adding items to these widely recognised scales. Specifically, three studies employed the Big 3 (Becchetti et al., 2018; Watanapongvanich et al., 2021; 2022), two studies administered the Big 3 but also included additional items (Amonhaemanon, 2022; 2023), one study used only two of the three items from the Big 3 (Razen et al., 2021), and one study utilised the Big 5 with the addition of one extra item (Cho, 2022). Wang et al. (2023) employed a comprehensive multidimensional measure of financial literacy, which included five factors: Financial Knowledge (20 items), Financial Capacity (15 items), Financial Management Values (12 items), Financial Ethics (9 items), and Wealth Values (9 items). Two studies measured self-report perceptions of financial literacy or financial literacy confidence (Amonhaemanon, 2022; Grable et al., 2021). These studies used a single item that asked participants to report their perceived financial literacy (e.g., “How knowledgeable are you about personal finance topics?” Grable et al., 2021).

Measurement of Gambling Behaviour

The majority of studies (n = 7) measured gambling behaviour in terms of frequency. Several of these studies only asked a single question to measure gambling frequency (i.e., Watanapongvanich et al., 2022; Razen et al. 2021; Cho, 2022) and these questions varied across studies. For example, Cho (2022) employed an objective time frequency (i.e., daily, weekly, monthly etc.), while Amonhaemanon (2022; 2023) considered the frequency of participation in the national lottery draw (i.e. every draw, most draws etc.). Amonhaemanon (2023) also measured the amount of money spent gambling within the past year and multiplied this by the frequency of gambling to create a composite variable which they called gambling intensity. Razen et al. (2021) measured frequency subjectively with a Likert scale (i.e., never, seldom, occasionally, etc.).

Only two studies attempted to measure whether the gambling was problematic (Wang et al., 2023; Grable et al., 2021). Wang et al. (2023) administered a 19-item scale to measure four dimensions of gambling behaviour: (1) impact on life, work, and relationships; (2) excessive win expectations (whether players expected to win excessively); (3) compulsive disorder (measuring difficulty controlling gambling); and (4) depressive signs (measuring negative emotions regarding gambling). Grable et al.’s (2021) approach was less direct but could be argued to indirectly tap into problem gambling. They used a single item asking participants how likely they would gamble a day’s wages at the casino (from 1 = very unlikely to 10 = very likely). Gambling such a large amount of money in a very short period of time could be considered a risky behaviour that is likely to be positively associated with problem gambling.

Methodological Quality

The methodological quality of the included studies was generally very good. Key strengths included the use of very large sample sizes and strategies to control for potential confounding variables in most studies, such as including an endogenous regressor within a probit model. However, reporting of sample characteristics was often unclear or incomplete, with many studies omitting key demographic information. While this doesn’t necessarily reduce the quality of the research itself, it does make it a little harder to properly interpret the findings or assess their generalisability to other populations.

Only one study provided information regarding the reliability or validity of their measures (i.e., Wang et al., 2023). While internal consistency was rarely reported, this largely reflects the use of very short scales, often single items, to measure financial literacy and/or gambling behaviour. However, greater attention to psychometric reporting, particularly validity, is still warranted. Future research in this area would benefit from the use of larger, multi-item scales that better support the evaluation and reporting of reliability and validity.