Please enable JavaScript.
Coggle requires JavaScript to display documents.
Behavioral eco-lect3: Judgement under risk and uncertainty part 4…
Behavioral eco-lect3: Judgement under risk and uncertainty part 4
Confirmation bias
Bayesian updating:
Consider an alternative way of writing Bayes’ rule:
P (H | E ) = PEH x P(H) / P(E|H) x P(H) + P(E|−H) x P(−H)
Where H stands for ‘hypothesis’ and E for ‘evidence’
Then:
P(H) = the probability you assign that the hypothesis is true (called the ‘prior’)
P(E|H) = the probability of getting the evidence obtained, conditional on the hypothesis you assign
P(H | E) = the probability that you should assign to the hypothesis, given the evidence
Example flipping a coin: we want to test if it is ‘unfair’. slide 24-27/34
Confirmation Bias cont.
However in real life this is not always the case
People see different sets of evidence.
People interpret the same evidence in a different way.
The latter case is called
‘confirmation bias’: a tendency to interpret evidence as supporting prior beliefs to a greater extent than warranted.
In the example above, rational people come to the same conclusion after seeing some evidence. This is independent of initial beliefs.
Some more on adjustment
The Linda problem
see example 31-32/34
Disjunction fallacy
: underestimating the probability of a disjunction (a string of events, one of which have to happen)
This happens in case of a
‘disjunction’
:
in probability-language: when an ‘or’ occurs in the calculation
People tend to use the probability of one event as an ‘anchor’, and then adjust upward insufficiently.
Example 34/34
Conjunction fallacy
: overestimating the probability of a conjunction (a string of events, all of which have to happen)
This happens in case of a
‘conjunction’
:
in probability-language: when an ‘and’ occurs in the calculation
People tend to use the probability of one event as an ‘anchor’, and then adjust downward insufficiently.