Gambling Decisions: Key Factors

The introductory section establishes that gambling is a widespread behavior, not confined to any particular social class, although preferred gambling types might vary across classes. It mentions that Chinese and Jewish communities have been stereotyped as heavy gamblers. Cohen's research (1960, 1964, 1972) on gambling, uncertainty, and psychological probability is cited, highlighting interesting historical and mythical gambling practices. Data on per capita gambling expenditure in the late 1970s is presented, showing Australia as the highest (A$710), followed by the USA (A$440), New Zealand (A$210), the UK (A$95), and Canada (A$87). The text cautions that these figures might be misleading because they represent turnover, not necessarily net expenditure. The section emphasizes the relative lack of research on non-economic factors influencing risk-taking in gambling.

Two main theoretical approaches to understanding gambling behavior are identified: a learning theory perspective and a focus on personality characteristics. The latter has primarily concentrated on compulsive or pathological gamblers, examining traits such as locus of control, extraversion, neuroticism, and impulsivity. The limited attention given to non-economic influences and the under-representation of social gamblers in research are highlighted. The document notes that while some studies have speculated on the causes of gambling proclivity, many remain mere conjecture. The lack of comparative research between social gamblers and non-gamblers is also noted, contrasting with the emphasis on understanding the final stages of compulsive gambling.

The text points out a significant gap in gambling research: the lack of focus on social gamblers, defined as those who enjoy gambling without excessive involvement. While a prominent model of the progression to compulsive gambling includes a social gambling phase (Custer, 1982), there's surprisingly little research dedicated to this group. The document highlights the absence of comparative studies examining the performance of social gamblers versus non-gamblers. The paucity of evidence supporting commonly ascribed characteristics to social gamblers is also criticized, suggesting that many assumptions lack empirical backing. This lack of research is particularly surprising given the social gambler's role as a potential transitional stage before compulsive gambling.

The study involved 96 participants from Christchurch, New Zealand. The participants were divided into two groups: gamblers and non-gamblers. Twenty male non-gamblers served as a control group, selected randomly from the Christchurch telephone directory (every fiftieth home). Non-gamblers were defined as individuals who only purchased raffle tickets or lottery tickets costing one dollar or less, or gambled less than once a year. The gambler group consisted of 24 volunteers (from a total of 32 approached) who were randomly approached at a race track. Gamblers were defined as those who gambled at least once a week or wagered over 10% of their weekly gross income. The sample included 24 males and 12 females, with a slightly uneven distribution of genders across the two groups. The mean age of gamblers was 22.5 years (range: 18-25), and for non-gamblers, it was 21 years (range: 18-34).

The study utilized an experimental design involving a simulated gambling task. Participants were shown three dice for approximately three seconds, and an experimenter would then declare a win or loss. In four out of seven blocks of 15 trials, participants were given additional information about potential fluctuations in luck (good, bad, or ordinary). Other blocks had no such information. Gambling cues were manipulated to create different experimental conditions. In the 'cues' condition, the experimental setting resembled a real-life gambling environment, with visual and audio stimuli such as playing cards, a mahjong set, dice, New Zealand dollars, used race track betting tickets, and race track programs present. The 'no-cues' condition eliminated these stimuli. The pace of the experiment was designed to prevent participants from suspecting any manipulation of the dice.

The study acknowledges potential limitations in its sampling methodology. One group of gamblers was composed of volunteers from Gamblers Anonymous (GA). This is explicitly stated as not representative of the broader population of pathological gamblers, introducing a sampling bias. The study notes that GA members are highly motivated to address their gambling problems. The authors discuss this selection bias, acknowledging that generalisations must be made cautiously. The study further notes that discrepancies in findings across different studies on gambler characteristics might stem from varied methodologies and measuring techniques used, as well as diverse definitions and selection criteria for gambler samples, including the presence of subgroups within pathological gamblers themselves (Moran, 1978b).

A key focus was on comparing the risk preferences and decision-making strategies of gamblers and non-gamblers. The study found that gamblers tended to take greater risks, opting for choices with a lower probability of winning but higher potential payoffs. This contrasts with non-gamblers, who appeared to be primarily motivated by the potential payoff, seemingly less sensitive to the probability of success. This difference suggests a potential learning or experience effect, as indicated by some gamblers in Phillips' study (1972) who described their choices as compromises between probability and utility. The research also investigated the influence of factors like the amount of money available and the frequency of dice numbers on decision-making, revealing a greater impact on social gamblers compared to non-gamblers.

The study examined how previous gambling outcomes influenced subsequent betting behavior. It observed that a majority of gamblers increased their wagers over time, starting with low stakes and progressively increasing them. The influence of win/loss patterns on risk-taking was analyzed, noting that both gamblers and non-gamblers were affected by runs of wins and losses, but the nature of this influence varied individually. The research also considered the potential effects of arousal levels. It notes Edwards' finding (1954) that subjects alter their bets after losses but not after wins. Coombs, Donnell, and Kirk’s (1978) demonstration of the probability of losing dominating the effect of the amount to win is mentioned. The study found that a run of three losses after a double win had a particularly noteworthy effect on betting patterns across both groups.

The concept of the gambler's fallacy (also known as the Monte Carlo fallacy or the fallacy of the maturity of chances) is discussed in relation to the study's findings. The study notes that gamblers often fail to consider the independent probabilities of events, believing in a self-correcting process where deviations in one direction are followed by opposite deviations. This research observed that social gamblers consistently showed higher subjective confidence in winning than in making a correct choice, especially compared to non-gamblers who experienced a run of poor returns, further highlighting the gambler's tendency to overestimate the chances of success. The influence of subjective versus objective probability perceptions is considered. The study further notes that the factors most influential in subjects' choices (money available and frequency of dice numbers) were twice as influential for social gamblers than for non-gamblers, whilst the probability of winning or losing was mentioned infrequently.

The study investigated the relationship between locus of control and risk-taking behavior in gambling. Locus of control scales (James' I-E Scale, Rotter's I-E Scale, Nowicki-Strickland Locus of Control Scale, etc.) were used to measure generalized expectancies. Rotter (1975) suggested that these scales are particularly useful for predicting behavior in novel or ambiguous situations. The research found that individuals with an internal locus of control (internals) tended to take more risks than those with an external locus of control (externals) in the gambling tasks. This finding contrasts with some previous research suggesting that high risk-takers are externally oriented (Moran, 1978a; Conrad, 1978). Slovic (1964) and Strickland et al. (1966) offered contrasting hypotheses about the relationship between internal locus of control and risk-taking, suggesting that internals might become more ego-involved in such situations and therefore take greater risks.

The influence of arousal and sensation-seeking on risk-taking was explored. The Sensation Seeking Scale (SSS) (Zuckerman, Kolin, Price & Zoob, 1964) was used to measure a general tendency toward sensation-seeking. Walters & Kirk (1968) provided validation for the SSS in predicting risk-taking behavior, finding a negative correlation between sensation-seeking and the preference for lower probability, higher payoff options. Zuckerman's (1974) research showing a link between high sensation-seeking scores and a preference for gambling over other activities is also referenced. The study explored how prior experiences (wins/losses) and arousal levels affected the amount of money staked. Increased arousal was hypothesized to disrupt regulatory mechanisms, enhancing risk-taking (Cohen, 1964), potentially due to threat denial or ignoring punishment cues (Rule, Nutter & Fischer, 1971; Hare, 1968). The inverted U-shaped arousal function is mentioned as a potential explanation for the observed patterns. The possibility of gambling as an internally driven form of stimulation-seeking is discussed, akin to the 'jogger's high', acknowledging the role of biological and psychological factors.

The study briefly addresses gender differences in gambling behavior, citing previous research (Schneider, 1968; Deaux, White & Farris, 1975; Karabenick & Eddy, 1979; Karabenick, Sweeney & Penrose, 1983) focusing on preferences for skill or chance activities. The text notes that women participate in casino gaming at a higher rate than men but generally with lower stakes. Observations about disproportionate representation of women among slot machine players compared to craps players (Herman; Shapiro, 1982) are presented. The study also touches upon Greenson’s (1947) analysis of neurotic gambling, highlighting the components of hope for reward and testing luck/fate, linked to parental images. Post-experimental questionnaire data revealed that female participants perceived themselves as more unlucky compared to males following the gambling session. Cohen's (1960) findings on belief in luck (12% for females and 7% for males) are also referenced.

The study proposes using the repertory grid technique to overcome the limitations of laboratory studies and the difficulties of field studies in gambling research. The authors argue that this technique allows for the investigation of the cognitive aspects of gambling behavior, exploring the versatility and applicability of the method in this context. The choice of this method is partly justified by the prominent role of luck and superstition in the existing gambling literature; the repertory grid offers a way to analyze their influence on participants' gambling decisions. The repertory grid, as a method, is presented as a way to obtain a richer understanding of the cognitive structures that underlie gambling behavior, compared to traditional approaches which may not capture the nuances of individual perceptions and beliefs.

The section describes the theoretical underpinnings of the repertory grid technique, drawing on Kelly's personal construct theory. It highlights several key features of constructs within this framework. Constructs are characterized as having a range of convenience, meaning they can only be meaningfully applied to elements within a comparable class. The importance of using comparable elements is stressed to avoid a distorted picture of an individual's construing. The hierarchical arrangement of constructs is also explained, with some constructs subsuming others in a superordinate-subordinate relationship. This hierarchical structure reflects an individual's way of organizing their understanding of the world, constantly evolving through experience. The section also explains the use of bipolar rating scales (semantic differential) within the repertory grid methodology, referencing Osgood (1962) and Slater's work (1960, 1977) on component analysis for understanding the connections between constructs.

The section details how results obtained from repertory grids are typically presented and interpreted. It explains the use of two-dimensional diagrams in C-space (Ryle, 1975; Slater, 1976), showing the distribution of elements and constructs. The distance of elements from the multivariate mean (origin) represents the element's salience or importance in the construct system. The relationships between constructs are illustrated by their loadings, which define axes on the diagram, with opposite poles represented along the circumference of a circle or within a rectangle. The section then clarifies that elements and constructs in the grid are neither solely ‘provided’ nor solely ‘elicited,’ often being selected from a common pool of options. A comparison is drawn between the repertory grid and personality questionnaires, emphasizing the repertory grid's advantage in focusing on personally relevant items (Duck, 1973).