Griffiths Psychology AS OCR

Preview:

Citation preview

Griffiths (1994)

The role of

cognitive bias

and skill

in fruit

machine

gambling

Professor Mark Griffiths

Aim

• To compare the behaviour and

cognitions of regular and non-

regular gamblers to see if there

were differences between the two

groups

Four main Hypotheses

1. RGs will be no different from NRGs in terms of

skill levels (ability to affect outcome positively)

2. RGs will produce more irrational (contrary to

reason) verbalisations than NRGs

3. RGs will report themselves (subjective) as

being more skill orientated than NRGs

4. (Ps in the ‘thinking aloud condition would take

longer to do the task)

Method

• Quasi field experiment

Advantages

Disadvantages

Method

• Quasi field experiment

Advantages Higher ecological validity

(sometimes low demand

characteristics- not in this case)

Disadvantages Little control of variables – hard to

establish cause and effect links

Can be difficult to replicate

Sometimes ethical issues

Participants/Sample• 60 people (mean age 23.4 years)

• 30 RGs (29 M 1F) – gambled at least once per week

• Gender imbalance is due to fruit machine gamblers being mainly M in this country.

• 30 NRGs (15M 15F) – gambled once a month or less (but had used fruit machines at least once in their lives)

• Self – selected volunteers – via posters around local college campuses

• A number of the RGs were recruited via a RG known to Griffiths - snowball sampling

Independent Variable

• IV = regular gamblers (Gs)

or non –regular gamblers (NRGs)

(Naturally occurring)

I won

because

I was

quick

07/04/2014

Dependent variables

(1)The „objective‟ DVs – hypothesis 1 –testing differences in skills levels

• measured by observation of 7 aspects of behaviour – quantitative data:Total number of plays in session

Total minutes of play in session

Total plays per minute in session

End stake – total winnings

Total number of wins in session

Win rate (time) – time between wins

Win rate (plays) – number of plays between wins

Dependent variables

• The „subjective‟ DVs- Qualitative data

(2) Cognitive activity – i.e. what Ps were

thinking – hypothesis 2: differences in

irrational verbalisations

• measured by „thinking aloud‟ technique

(3) Perception of SKILL – hypothesis 3

• measured by post - experiment semi

structured interviews

Controls

All participants played same machine

„Fruitskill‟ (except those who didn’t!) –

controlling situational variables

Randomly assigned to thinking aloud /

non-thinking aloud – controlling participant

variables

PROCEDURE

Procedure – the task• All Ss were tested individually

• Each participant was given £3 to gamble

on a fruit machine (equal to 30 free plays)

in a local arcade

• The game selected was FRUITSKILL

(though some players moved on to other

games- see next slide) *

• They were asked to try to stay on their

machine for at least 60 gambles (i.e. aim

to break even & win back £3)

Procedure – the task

* 3 RGs objected, either because they were

not familiar with FRUITSKILL or because

they preferred a different machine

* further RGs began on the FRUITSKILL

machine, then changed machines, each

changing at least 3 times

Procedure – the task

• At 60 gambles they were allowed to keep

the £3 or carry on gambling

• RGs & NRGs were randomly allocated to

the “thinking aloud” (TA) “non-thinking

aloud” (NTA) conditions

‘Thinking Aloud‟ Method

• Considered the best method for accessing

cognitive processes – it requires Ps to verbalise

every thought that passes through their mind.

• The Ps were given lapel microphones and

given certain instructions:

•Say everything that goes through your

mind

•Do not censor your thoughts – even if it

seems irrelevant to you

•Keep talking continuously

•Speak in complete sentences if

possible – but don‟t worry if you can‟t

•Do not try to justify your thoughts

Self-reports in post-experimental

semi-structured interviews

The three main questions that were asked to measure perceptions of skills were:

1) Is there any skill involved in playing the fruit machine?

2) How skilful do you think you are compared to the average person?

3) What skill (if any) is involved in playing the fruit machines?

RESULTS

„This

machine

hates me!‟

Hypothesis one looked at differences

between RGs and NRGs.

• The study found only two significant

differences:

– Regular gamblers had a higher playing rate of

8 gambles per minute. NRGs had a playing

rate of 6 gambles per minute.

– RGs who thought aloud had a lower win rate

(in plays) and therefore made fewer gambles

between each win than the other groups.

Hypothesis 1 contd.

• The study also found 2 interesting, but not

significant, findings:

• 1. Regular gamblers were seen to spend

more time on the fruit machine by having

more gambles using the same initial stake.

• 2. There were no significant differences in

the amount of total winnings between

those who thought aloud and those who

didn’t.

Hypothesis 1: testing the

difference in skill levels

• On the whole there were no significant differences between regular gamblers & non regular gamblers

• Hypothesis 1 is supported.

• (there is also some support for hypothesis 4 that ‘thinking aloud’ means you take longer to gamble)

Hypothesis 2. - RGs will produce more

irrational verbalisations than NRGs

• The verbalisations gathered from the ‘thinking aloud’ method were analysed by content analysis

• The author produced a coding system. Helooked through the transcripts and intuitively identifyied 30 utterance categories

• The author then categorised the statementsmade by each participant using this coding system

• They were then counted – turning qualitative data into quantitative data

Reliability of the coding system

• Reliability was checked by using two other raters

• Inter-rater reliability was low because:

• 1) One rater knew very little about fruit machine gambling & therefore couldn’t understand the terminology

• 2) The second rater had not been present during the recording of the utterances & therefore had no context & could not make as much sense of the utterances as the author

07/04/2014

Examples of coding system

categories

Content Analysis

categories

NRG RG

Machine personification 1.14 7.54

Explaining losses 0.41 3.12

Talking to the machine 0.90 2.64

Swear at machine 0.08 0.o6

Reference to skill 1.47 5.34

Verbalising confusion 4.81 1.72

1. Examples of rational verbalisations

• Reference to winning

e.g. ‘I won forty pence I think’

• Confusion/non-understanding

e.g. ‘What’s going on here?’

2. Examples of irrational verbalisations

• Personification of the fruit machine

e.g. ‘The machine likes me’

• Explaining away losses

e.g. ‘I lost because I wasn’t concentrating’

07/04/2014

More examples of

IRRATIONAL

VERBALISATIONS

This ‘fruity’ is not in a good

mood

It wants its money back

Putting only a quid in ‘bluffs

the machine’

The machine … hates me

This machine won’t pay out

happily

Results of verbalisations

• Regular gamblers made a significantly higher percentage of verbalisations in the following irrational category:

• Personification of the fruit machine - “The machine likes me”

• Regular gamblers made significantly higher percentage of verbalisations in the following rational category:

• Reference to the number system - “I got a 2 there”

Results of verbalisations

• Non regular gamblers made significantly

higher percentage verbalisations in the

following Rational categories:

• Questions/statements relating to confusion

– What’s going on here?/I don’t

understand this.

• Miscellaneous utterances – I think I’ll get a

bag of crisps after playing this.

Hypothesis 2: Overall Results

• Similarities: Overall both groupsused more rational than irrational verbalisations

• Differences - Regular gamblersproduced significantly more irrational verbalisations (14%) than non regular gamblers (2.5%)

• Supporting hypothesis 2

Heuristics

RGs used a variety of heuristics, showing cognitive bias in their verbalisations :

For example:

• Flexible attributions e.g. ‘ two nudges, gotta be…oh you son of a bitch, you changed them’ (talking to the machine)

• Illusory correlations: the belief that skills like using the nudge button well has a high impact on winning when in reality it does not.

Results from semi-structured interviews.

Hypothesis 3 – Perception of Skill

Interview Question NRGs RGs

1) Is there any skill

involved in playing

the fruit machine?

2) How skilful do you

think you are

compared to the

average person?

3) What skill (if any)

is involved in playing

the fruit machines?

Results from semi-structured interviews.

Hypothesis 3 – Perception of Skill

Interview Question NRGs RGs

1) Is there any skill

involved in playing

the fruit machine?

mostly chance equal chance and

skill

2) How skilful do you

think you are

compared to the

average person?

Below average Above average

OR

Totally skilled

3) What skill (if any)

is involved in playing

the fruit machines?

Feature skills

Knowing when

machine will pay

out

07/04/2014

Additional findings

• Of the 14 RGs who managed to ‘break

even’ (60 gambles) – 10 (71%) stayed on

machine until they lost all the money

• Of the 7 NRGs who broke even – 2 (29%)

stayed on until lost all the money

Griffiths explanation of his results:

Griffiths says that gamblers know they

will lose money but by using familiar

machines and having some level of skill

they are able to maximise their playing

time.

Gamblers play with money not for it -

staying on the machine is the objective

Conclusions - Questions

Answer the following questions (give

reasons for your answers):

• Do regular gamblers behave

differently?

• Do regular gamblers think differently?

• Do regular gamblers think they are

more skill orientated?

Broad Conclusions - answers

• RGs Behave differently = No as there

were no overall differences on the seven

objective measures

• RGs Think differently = Yes, as regular

gamblers produced more irrational

verbalisations than non regular gamblers

• Differences in perceived skill orientation =

Yes, as regular gamblers were more skill

orientated in their self-report ratings

Conclusion P= RGs have different cognitive

thought processes than NRGs

E= RGs think there is more skill involved than there actually is (as the skills identified had a minor influence on outcome). Also RGs made more irrational verbalisations.

C= However, only fruit machine gambling is studied so the findings may not apply to other forms of gambling.

Applications/Implications

• The results of this study may be used to

rehabilitate ‘problem gamblers’.

• It shows that gamblers may have certain

cognitive biases (ways of thinking).

• Thus, it suggests that they may be helped

by cognitive- behaviour therapies to

change their thinking e.g. listening to ‘think

aloud’ recordings to highlight their irrational

verbalisations

Ethics

• Carried out according to BPS guidelines

• The participants gave fully informed

consent and had the right to withdraw

• Protection of Ps – they were given money

to gamble – could be seen as encouraging

gambling.

Ecological ValidityThe setting:

• A field experiment – so ecological validity is high

- took place in a naturalistic setting (arcade) on a typical

fruit machine

Money:

• However - Using someone else’s money may reduce the

excitement & risk taking involved in gambling.

• Nevertheless the researchers believed that allowing

participants to keep their winnings may compensate for

this

* Also: thinking aloud condition - low EV – not how

gamblers usually behave

Quantitative and Qualitative

data• Hypothesis 1 – behavioural measures of

skill – quant data

• Hypothesis 2 – irrational verbalisations –

qual data turned into quant by coding

system

• Hypothesis 3 – perceptions of skill – qual

data gained through semi-structured

interviews

Evaluation

• State strengths & weaknesses of the study using the following evaluation issues:

• Sample

• Methodology

• Ecological validity

• Validity

• Reliability

• Type of data collected

Evaluation - Strengths

• A variety of data collection methods were

used – observation, interview, ‘talking

aloud’ – allows triangulation and increases

validity.

• Large amounts of data: quantitative –

allowing objective statistical comparisons -

and qualitative data - giving richness and

insight.

Evaluation - Weaknesses

• Inter-rater reliability was low – this suggest the

descriptions of categories was only understood

by Griffiths – this might be a source of

subjectivity and bias.

• Sample: Volunteer sample of students – not

representative of wider society. Snowball sample

also means sample not representative – p’s

likely to be similar.Gender imbalance in RGs -

but this reflects the problem in the UK

• Field study – extraneous variables hard to

control – reducing validity and reliability

Evaluation - Weaknesses

• ‘Thinking aloud’ produces descriptions of

thinking NOT explanations of why they are

thinking it. Also – open to

falsification/demand characteristics. Some

people find it difficult and in fact the RGs in

this study were silent for some of the time–

reducing validity.

• Overt observation – open to

falsification/demand characteristics –

reducing validity

Alternatives1. Self report questionnaire – ask the participants what they were

thinking while they were playing –using open questions.

Advantages:

Ps could play normally without being disturbed by ‘thinking aloud’ .

They could explain their thoughts in phrases non-regular gamblers would understand

Disadvantages:

would rely on memory

Would allow them time to censor their thoughts – social desirability

Effect on results:

• If Ps didn’t remember what they’d thought, validity would be decreased. Validity would also go down if they censored their thoughts. But on the other hand playing normally (without talking aloud )would increase the validity of the research.

• Being able to explain their thoughts might increase inter-raterreliability when the content was analysed

Alternatives2. Use more forms of gambling e.g. horse racing, dice, cards.

Advantages:

Increase validity of the data - getting wider information about the

way gamblers think. Make comparisons between thought

processes shown in different types of gambling.

Disadvantages:

• Time consuming.

Effect on the results

• Might find gamblers use similar cognitive processes across different

types of gambling. Or, might find that some kinds (eg poker)

genuinely require more skill, and that people use different heuristics

where those are concerned.

Recommended