The Role of Cognitive Bias and Skill in Fruit Machine Gambling
Griffiths (1994) (Summary)
Abstract
Recent studies suggest that individuals who engage in gambling behaviour demonstrate a strong cognitive bias. This study looked at variables relating to the cognitive psychology of playing a fruit machine and compared the results of 60 participants (30 regular and 30 non-regular gamblers) playing fruit machines in a British amusement arcade. Subject completed a self-report and were observed whilst playing the machines, The results showed that regular gamblers make more irrational verbalisations that non-regular gamblers and also that regular gamblers can gamble more times with the same amount of money (but not significantly so).
Background
Humans tend to hold consistent biases when processing information and early studies into gambling behaviour focused on the illusion of control as a cognitive bias held by regular gamblers. Wagenaar (1988) suggests that it is distortions in cognitive processes which lead to a particular way of reasoning. Gambling behaviour has nothing to do with personality, education or social environment.
There are six cognitive distortions which are useful in understanding gambling:
- Illusion of Control: for eg. Choosing a specific lottery ticket or playing a favourite machine. This leads to gamblers believing they have a greater chance of winning.
- Flexible Attributions: this is where gamblers attribute their successes to their own skills but attribute losses to external factors (internal / dispositional or external / situational attributions)
- Representativeness: this relates to the ‘gambler’s fallacy’ – that the chance of winning increases with the number of ongoing losses.
- Availability Bias: this relates to the way in which winners and winning are highly publicised (eg. Announcing lottery jackpot winners or the noise of machines paying out) – this suggests winning is more commonplace than it is in reality.
- Illusory Correlations: this relates to all the superstitious behaviours gamblers exhibit (for eg. Rolling dice in a particular way).
- Fixation on Absolute Frequency: this relates to gamblers focusing on how much they win compared to other people, with no consideration for the fact that they gamble more and also lose more than other people.
Regular fruit machine players believe their actions to involve skill – this is stimulated by features such as ‘nudge’ and ‘hold’. Some addicted players also say they know they will lose all their money in the long run, but try and stay on the machine for as long as possible, indicating that they also see this as something that can be attributed to skill.
Pathological fruit machine gamblers have a greater perceived level of skill that less regular gamblers.
Methodology
This study investigated a number of factors, including:
- whether the playing of the machine involved perceived or actual skill
- the differences between the thinking aloud of regular and non-regular gamblers (the ‘thinking aloud method’)
- subjective measurement of skill
Design
Quasi-experiment / field experiment – the independent variable was the regularity of gambling and had two conditions: regular / non-regular gamblers. Conducted with full support of arcade manager on an individual basis.
Participants
- 30 regular and 30 non-regular fruit machine gamblers recruited via posters around the college and university campuses.
- All had to have played fruit machines at least once in their life.
- Regular gamblers (29 males, 1 female, mean age 21.6 yrs) had to play at least once a week.
- Non-regular gamblers (15 males, 15 females, mean age 25.3 yrs) had to play once a month or less.
- The gender inequality reflects the male dominance in fruit machine gambling.
Data Collection Methods
Self-report via a semi-structured interview after the experiment
Observations of participants ‘thinking aloud’
Hypotheses
H1: There would be no differences between regular and non-regular fruit machine gamblers on objective measure of skill (observations) based on seven behavioural dependent variables:
- Total plays: total number of plays during session
- Total time: total time in minutes of play during one session
- Play rate: total number of plays per minute
- End stake: total winnings in number of 10p pieces
- Wins: total number of wins
- Win rate (time): total number of minutes between each win
- Win rate (plays): total number of plays between each win
H2: Regular gamblers would produce more irrational verbalisations than non-regular gamblers.
H3: Regular gamblers would be more skill oriented than non-regular gamblers on subjective measurement (self-report interview).
H4: Thinking aloud subject would take longer to complete the task than non-thinking aloud subjects.
Procedure
Each gambler was given £3 to gamble on a fruit machine – 30 ‘free’ plays at 10p each. All were asked to try and stay on the machine for 60 gambles (ie to win back the £3 and spend it again). They were told they could keep any money at won at the end / choose to continue gambling – this was to counter the possibility that not playing with one’s own money takes away the excitement of playing.
Both the regular and non-regular gamblers were randomly assigned to one of two groups: the ‘thinking aloud’ or ‘non-thinking aloud’ groups. Previous research has shown that the ‘thinking aloud method’ does not influence behaviour, but it does slow it down.
All participants were instructed to play the same FRUITSKILL machine unless they had any objections.
All ‘thinking aloud’ participants were given standardised instructions about saying everything that came into their mind and not worrying about talking in complete sentences. The verbalisations were recorded on to a portable tape recorder via a lapel microphone. The researcher was standing nearby recording the seven DVs identified above. All verbalisations were transcribed within 24 hours to ensure that the context was not forgotten.
Results
Only two significant differences were found (p=0.01): regular gamblers had a significantly higher playing rate (8 gambles per minute) than non-regular gamblers (6 gambles per minute). Regular gamblers had a significantly lower win rate in the thinking aloud group than any other group.
ANOVA showed that regular gamblers stayed on machines longer, but this was not statistically significant. There was also no significant difference in the playing times of the two groups.
Content analysis was performed on the transcriptions of the thinking aloud condition. The coding for this was “intuitively constructed” by the researcher based on 30 categories of utterance and a further miscellaneous category. Each type was then tallied and given a percentage weighting. Regular and non-regular gamblers were then compared using t tests (significance level p<0.01).
The reliability of this coding system is unknown as it was based on verbalisations that were specific to the machine and the context at the time and therefore could only be understood by the researcher.
Regular gamblers made significantly more verbalisations where they were personifying the machine (p<0.001) and in referring to the ‘number system’ (p<0.01). Non-regular gamblers made more utterances in relation to confusion and non-understanding (questions / statements) and miscellaneous utterances. There were two categories of utterance (mind going blank / frustration) that only applied to non-regular gamblers.
Regular gamblers made more verbalisations fitting the categories of ‘swearing at the machine’, ‘referring to winning’ and ‘explaining away losses’ – these were approaching significance but were not significant.
The majority of verbalisations were rational for both groups, but regular gamblers made significantly more irrational verbalisations (14%) than non-regular gamblers (2.5%) (p<0.001). Gamblers used a variety of heuristics (problem solving techniques) many of which relate to personifying the machine:
Hindsight bias: “I had a feeling it wasn’t going to pay very much after it had just given me the feature…”
Flexible attributions: “I’m losing heavily here… (the machine’s) not giving me the numbers I want. I’ve just taken a quid off it so it wants its money back now.
Erroneous perceptions: “I’m only gonna put one quid in to start with because psychologically I think it’s very important… it bluffs the machine – it’s my own psychology.”
“Can I win more than 10p this time?...No!!...Obviously the machine’s being a bit of a bastard at the moment.”
In the post-experiment interview, questions asked about skill, for eg. “Is there any skill involved in playing a fruit machine?”. Most non-regular gamblers said ‘mostly chance’, whereas most regular gamblers said ‘equal chance and skill’. Regular gamblers rated themselves of at least average skill, with most saying they were above average or ‘totally skilled’. Most non-regular gamblers said they were below average and 7 stated that as it was a game of chance they were as skilled as anyone else.
The were similarities in the response to the question ‘What skill (if any) is involved in playing fruit machines?’ although regular gamblers said knowledge of the ‘gamble’ button was skilful, as was knowledge of the ‘feature’ button and knowledge of when the machine will / will not pay out. This last skill is used as an excuse for not winning by regular gamblers. 10 / 14 regular gamblers carried on playing until they had lost everything (those 14 being the ones who had reached 60 gambles), compared to only 2 / 7 of the non-regular gamblers.
Discussion
On the whole the null hypothesis is supported – there is no difference between regular and non-regular gamblers, although there were some differences. Regular gamblers stayed on the machine for longer, suggesting there is some skill involved (turning smaller wins into bigger ones for eg.). Regular gamblers play faster and more often, which leads to the fixation on absolute frequency bias. The thinking aloud group also took longer to play, but not significantly so.
A few gamblers objected to playing on the chose machine, which supports Langer’s (1975) illusion of control due to familiarity with the machine. The playing until all money is lost indicates that gamblers play with rather than for money and the need to play on a familiar machine extends the time played. This suggests that gambling functions as a leisure activity or form of escapism.
Ladouceur et al (1990) found that 80% of regular gambler’s verbalisations were irrational, which is far greater than the 14% found, although in this study all verbalisations were recorded, so a direct comparison is not possible.
The validity of the ‘thinking aloud’ method must be questioned, as it only identifies what the participants are thinking, rather than why. Understanding why these cognitive biases occur may reveal developmental or other reasons, rather than just heuristics. More research is required into this.
Fisher previously reported that there are three skills motivating gambling: choosing a machine, knowing the reels and using the gamble button. According to regular gamblers in this study, the gamble button could be played at a particular time to better effect, but gaming industry information says that the gamble button operates entirely by chance, giving the illusion of skill (a pseudo-skill).
Some skills Griffiths refers to as ‘idiot skills’ such as knowing to hold winning symbols, or knowing not to play a machine that has just paid out all its available money. However, he also suggests that knowing the reels, the number system and light oscillation are real skills. The FRUITSKILL letters light up – often they light as far as FRUITSKIL then the machine clears it – this is what Griffiths refers to as the near miss.
The real difference between regular and non-regular gamblers appears to be cognitive. This resulted in them personifying the machine more.
This understanding of irrational gambling bias may help rehabilitate problem gamblers, helping them to modify their thought patterns (Cognitive Behavioural Therapy) or to playback recordings so they can hear for themselves. The four who did hear their recordings could not believe how they were thinking. One was diagnosed as a pathological gambler using DSM-III-R criteria – after the study his gambling behaviour reduced partly through hearing the recording.
Therapists could use detailed structure of an individual’s gambling experiences to moderate or eliminate the motivation to gamble. If cognitive biases stimulate persistent gambling then audio playback may be a way forward for rehabilitation.