Mark Griffiths - Thinking and Gambling

Revision cards for the study 'Thinking and Gambling'by Mark Griffiths


Rational choice theory would suggest that people would people would never gamble. Griffiths wanted to investigate whether cognitive bias (any kind of faulty thinking that could be used to explain winning and losing) could explain the behaviour of regular gamblers.

The specific types of cognitive bias he looked at were:

a) illusion of control - where gamblers believe that games of chance are in fact based on skill, and that they have that skill.

b) flexible attributions - where gamblers see losing as due to bad luck, but winning due to skill.

These cognitive biases become part of a set of thinking (ideas and rationalisations) that gamblers use to justify their behaviour. These are called heuristics.

1 of 9


The aim was to investigate cognition and gambling, and in particular to find out whether regular gamblers (RG) think and behave differently to non-regular gamblers (NRG).

Hypotheses 1 - The behavioural hypothesis

  • That there are significant differences in the behaviours of RG and NRG.

Hypotheses 2 - The cognitive hypothesis

  • That there are significant differences in the thought processes of RG and NRG.

Hypotheses 3

  • Regular fruit machine gamblers would describe themselves as highly skilled, and believe that gambling requires a lot of skill.

Hypotheses 4

  • Gamblers in the thinking aloud condition would take longer to complete the task than those tested in the non-thinking aloud condition.
2 of 9


  • Self-selected sample
  • The participants were volunteers who responded to a poster placed in the local university.
  • There were 30 regular gamblers and 30 non-regular gamblers.
  • The regular gamblers included 29 men and 1 woman.
  • The non-regular gamblers were equally divided (15 men, 15 women)
  • The regular gamblers were defined as those who gambled on fruit machines at least once a week.
  • The non-regular gamblers were those who gambled once a month or less.
3 of 9


The procedure took place in a real amusement arcade. The IV was whether the gamblers were regular or non-regular. As this variable was not manipulated the procedure can be called quasi-experimental.

The methods included

  • measuring the winnings and time
  • content analysis of the 'thinking aloud' recordings
  • semi-structured interviews

It was ethical because Griffiths gained fully informed consent from the volunteers. All participants using the same 'fruitskill' machine, and were randomly allocated to the thinking aloud groups.

4 of 9

Semi-structured Interview

A semi-structure interview is an interview without set questions. There is generally a theme that the interviewer has to question around.

eg. An interview on a famous singer would have a theme. For instanse, ask questions about new single. So, the interviewer may choose what questions to ask, plus, if the interviewee says something more interesting, the interviewer can ask about it.


  • Flexible, sensitive
  • Reliable and easy to analyse


  • Open to experimenter bias
  • Difficult to replicate
5 of 9

Gambling Behaviour and results

Each participant was given £3, which was equivalent to 30 plays, and they were asked to stay on the machine for 60 plays. This meant, in effect, that they had to break even. Any winnings after 60 plays they could keep, or they could carry on playing. There were seven DVs - variables that were measured. These produced quantitive data.

Only two results were significant. These were:

a) that RG had a higher playing rate, and

b) RG who thought aloud had a lower win rate in plays.

As only two results were significant Griffiths rejected hypothesis 1. However, there is some evidence that RG were more skillful as 14 managed to 'break even' and 10 stayed on machine until they lost all the money. Only 7 NRG broke even and 2 stayed on machine until they lost all the money. Griffiths also found that winning was not necessarily the main aim - staying on the machine for as long as possible was more important, for playing was enjoyable in itself.

6 of 9

Gambling cognition and results

The participants in each condition (NG and NRG) were randomly assigned to two groups: thinking aloud (TA) and non-thinking aloud (NTA). In the TA condition the participants were encouraged to say whatever came into their heads while playing. These were qualitative and quantitative data. The results were recorded at the time and written up within 24 hours.

The results showncognitive differences between the two groups:

  • Irrational verbalisations - The RG were more likely to have irrational verbalisations eg. personifying the machine, talking and swearing at it. Examples of personification are: 'the machine hates me', 'this fruity is in a good mood'. Examples of explaining away losses are: 'I lost there because I wasn't concentrating'.
  • Rational verbalisations - The NRG were more likely to record confusion and lack of understanding of the game, whereas the RG were more likely to refer to winning situations.

RG were significantly more likely to report skill being involved, even though fruit machines are based on chance. These results came from semi-structured interviews.

The cognitive findings therefore support the second hypothesis.

7 of 9


  • There are only minor behavioural differences between RG and NRG but there are important cognitive differences in both their rational and irrational speaking.
  • There was also an irrational belief in skill by RG. Griffiths suggest that when confronted later with their verbalisations, many RG were surprised and embarassed. This would suggest that CBT could be effective for people with a gambling problem.
  • The results support explanations of gambling that are based on cognitive bias and heuristics.
8 of 9


  • Quasi-experiment - Difficult to establish cause and effect
  • Sample - Volunteers might not be representative of gamblers or students. There may also be demand characteristics. Huge gender differences, which may be considered representative, but it makes it difficult to generalise).
  • Ecological Validity - Research carried out in a real arcade, with real money, however, thinking aloud may cause lower ecological validity as people generally dont do this. Thinking aloud may also lower reliability.
  • Quantitative and Qualitative Data - Using a mixture of these is called triangulation. This allows a mixture of in-depth data, and replicable data.
  • Usefulness - Griffiths found that fruit-machine gamblers could be treated by allowing them to listen to their 'thinking-aloud' recordings. The insight they gain about their cognitive bias is the first step towards correcting it and changing behaviour. Therefore Griffith's study supports the use of CBT - cognitive behavioural therapy.
9 of 9


No comments have yet been made

Similar Psychology resources:

See all Psychology resources »