Judgements Flashcards
Kidney cancer in USA
Low prevalence countries were:
rural, sparsely population, republican, Midwest, south, west
but high prevalence countries were the same
Why do low prevalence and high prevalence countries have the same rural lifestyles?
Sparsely population - small sample, more variability
if done a year later, same pattern found but the particular countries with high and low prevalence rates would differ
The Gates foundation
Invested 1.7 billion in a programme to create small schools, because they found a large proportion of small schools getting better results than average, didn’t look at size of worst performing schools, would’ve found the they do worse than average
Small sample = more variability found
The law of small numbers
People take small samples to provide accurate estimates just as large samples do (the law of large numbers) but small samples yield extreme results more often than small samples do
What are the misperceptions of randomness?
People are misled by randomness
bomb strikes on London, falling in some areas but not others, thought certain areas were deliberately missed - but this isn’t true, no evidence it wasn’t random
do basketball players get a hot hand which causes them to score baskets from several shots in a row? the sequence of them scoring loads in a row was just random
How to we make judgements?
We use heuristic rules, easily applied, gives us quick answers, correct most the time but can lead us astray
making proper judgements takes a while. e.g. insurance company - know young people have more accidents so insurance should be more, but need to research this
What are the three heuristics?
Anchoring and adjustment
Availability
Representativeness
What is anchoring and adjustment?
When making a judgement about how likely something is, you have a starting point (the anchor) and then have to adjust this to be more accurate
Example of anchoring and adjustment
A wheel of fortune stops at either 10 or 65, people asked to read result and decide if proportion of nations in the UN that are African is larger or smaller than that number, have to make a numerical estimate
people move away from initial suggestions
if landed on 10, move to 25%
if landed on 65, move to 45%
actual figure 30%, adjustment not as much as it should be
What are 65 and 10 called in the wheel of fortune?
Anchors - people do not adjust enough from poor or arbitrary anchors
What can anchors be used in?
Negotiations - prices for used cars
Is adjustment used by system 1 or 2?
It can occur as a deliberate process but it is also subject to unconscious influences (shaking heads leads to moving further away from anchor, nodding head to less)
also evidence of a priming (system 1) component to anchoring via evoked images
A real world example of anchoring and adjustment
Estate agents valuing real properties with high and low anchors showed an anchoring effect of 40%
How do you work out the anchoring effect?
It is the difference between the final estimates divided by the difference between the anchors
What are availability and representativeness based on?
Fundamental cognitive processes
availability - retrieval from memory
representativeness - judgements made from similarity
What is availability?
It is used whenever people estimate frequency or probability by the ease of which instances or associations could be brought to mind
Examples in which availability misleads
Do more English words have R as first letter or third letter? think more have r as first letter because we think of words due to the letter they begin with, but more have it as third letter
famous names experiment - when asked to remember if they were famous or not, remember more as being famous because they come to mind more easily
Availability - domestics
Estimating members of couples of their relative contribution to household chores
results usually add to over 100%, your own contributions are more readily accessed from your memory than partners
people overestimate their contribution to causing arguments
Are more examples harder to recall?
List 6 or 12 instances of situations in which you have been assertive and rate how assertive you are
asking for 12 led to lower ratings, because the last instances were difficult to bring to mind
background music can make things harder to remember
Cause of death study
Effects of media coverage
deaths by accident overestimated compared with deaths by stroke
death by lightening underestimated compared with death by botulism - as this is in news more
What is representativeness?
Making a judgement based on whether something depending on how well it fits with a stereotype or prototype - based on similarity of what you think something is like
Tom W experiments
Rank base rates for graduate students in 9 areas of study
rank each area for how well a description of tom W fits typical graduate student in that area
rank areas for how likely tom w is now a graduate student in that area
People know there weren’t many computer scientist graduates in the past, but the description fits this so people thought it was a computer scientist
Going against representativeness
Billy Beane - manager of basketball team, overruled his scouts who were suggesting hiring players who seemed to fit the physical stereotype or good basketball players - look at play statistics instead
he formed a very successful team for a very low cost
What are the two problems of using representativeness?
Ignoring base rates (computer scientist graduates were relatively rare when the exp were carried out)
Using poor or useless information - the information in description of Tom is described as old, based on psychological tests of uncertainty
Representativeness and the conduction fallacy
Given a description of Linda and then asked about what she is:
less likely to be both but still pick it - ignoring base rate
most choose second option, but the prob of being both is less likely than being one
more likely to be a bank teller with a feminist movement because she fits the stereotype, people choose this because option 2 represents the description of her more, even if it statistically less likely
the description of Linda Is crucial in producing the judgement
Is less more?
Value of two dinner sets, one contains all the items in the first, a few more good items and a few broken
in a direct comparison, the set with more items is valued higher
but when the values are given by different people, the set with fewer itemises valued higher
Probability of a cab involved in accident being blue or green
A cab was involved in hit and run, a blue and green company run in the city. 85% in city are green and 15% are blue, the witness identified the cab as blue. the court tested reliability of the witness and conclude that the witness correctly identified one of the two colours 80% of time and failed 20% of time.
people believe it was 80% likely to be blue, but it was actually 41%
Cab story - if the problem is changed to a causal story
85% of accidents are caused by drivers of green cabs, performance then improves considerably
simple base rates are hard to engage with and use, causal base rates are easily incorporated into a story
Experience with flying instructions
Praise for a well executed manoeuvre was often followed by poorer performance next time
Criticism for a poorly executed manoeuvre was followed by better performance
praise is ineffective and criticism is effective
Double marking
sometimes marking is overestimated/undermarked
for 2 examiners, the best prediction is they will give it the same mark
if mark of first examiner is known, best prediction is that second examiners mark will be closer to overall mean
if mark of second examiner known, best prediction is that first examiners mark will be closer to the overall mean - other things equal, spread of marks will be the same
What is regression to the mean?
Moving towards the mean, we have poor intuitions about phenomenon so don’t recognise it in the real world
Restaurant example of regression to the mean
A really excellent meal at a restaurant on one visit is likely to be followed by a slightly disappointing one on the next visit
