Summary – Judgment under Uncertainty – Tversky and Kahneman:
Heuristics and Biases (1974). Amos Tversky and Daniel Kahneman.
Biases in judgements reveal some heuristics of thinking under uncertainty.
Many decisions are based on beliefs concerning the likelihood of uncertain events
expressed as odds or probabilities limited validity.
People rely on heuristic principles which reduce the task of assessing probabilities and
predicting values due to subjectivity and limited validity, this can lead to common biases.
There are main types of heuristics that are employed in making judgements under uncertainty
lead to biases these biases are enumerated.
Representativeness Heuristic probabilities are evaluated by the degree to which A is
representative of / resembles B assessing or ordering by the degree of similarity to a
stereotype (probability) leads to errors, because probability ≠ similarity the terms
named hereafter are factors that have no effect on representativeness but should have effect on
probability:
1) Prior probability of outcomes when no specific evidence or explanation is given in
an experiment, prior probabilities are properly utilised (based on given information);
when (uninformative) evidence is given, they are ignored.
2) Sample Size when subjects failed to appreciate the role of sample size, even when
it was emphasised in the formulation of the problem neglect of sample size.
3) Chance gambler’s fallacy expectation that essential characteristics of a random
process will be represented (in each of its parts).
4) Predictability only predicting in terms of favourableness of a description no
regard for considerations of predictability.
5) Illusion of Validity unjustifiable confidence of due to a fit between the output and
input correlation decreases accuracy / probability, even as it increases confidence.
6) Regression towards the mean / average value to recognise the import of regression
the thought that punishment is more effective than a reward.
Availability a useful clue for assessing frequency of a class biases:
1) Retrievability of instances
2) Effectiveness of search set neglecting the fact that different tasks draw more search
sets into existence.
3) Imaginability incorrect evaluation of a frequency of classes.
4) Illusory correlation co-occurring event resistant to contradictory data.
Adjustment and Anchoring situational estimates based on the formulation of an issue are
biased towards initially chosen values (chosen by computation) underestimating
probabilities of failure in complex systems:
1) Insufficient Adjustment when an initial value, chosen through wrong computation,
is given and not enough adjustment is made to it.
2) Conjunctive and Disjunctive events conjunctive events are often overestimated,
while disjunctive events are underestimated.
Biases that occur due to the reliance on judgmental heuristics.
A better understanding of heuristics and their resulting biases could improve judgements and
decisions in situations of uncertainty.
Heuristics and Biases (1974). Amos Tversky and Daniel Kahneman.
Biases in judgements reveal some heuristics of thinking under uncertainty.
Many decisions are based on beliefs concerning the likelihood of uncertain events
expressed as odds or probabilities limited validity.
People rely on heuristic principles which reduce the task of assessing probabilities and
predicting values due to subjectivity and limited validity, this can lead to common biases.
There are main types of heuristics that are employed in making judgements under uncertainty
lead to biases these biases are enumerated.
Representativeness Heuristic probabilities are evaluated by the degree to which A is
representative of / resembles B assessing or ordering by the degree of similarity to a
stereotype (probability) leads to errors, because probability ≠ similarity the terms
named hereafter are factors that have no effect on representativeness but should have effect on
probability:
1) Prior probability of outcomes when no specific evidence or explanation is given in
an experiment, prior probabilities are properly utilised (based on given information);
when (uninformative) evidence is given, they are ignored.
2) Sample Size when subjects failed to appreciate the role of sample size, even when
it was emphasised in the formulation of the problem neglect of sample size.
3) Chance gambler’s fallacy expectation that essential characteristics of a random
process will be represented (in each of its parts).
4) Predictability only predicting in terms of favourableness of a description no
regard for considerations of predictability.
5) Illusion of Validity unjustifiable confidence of due to a fit between the output and
input correlation decreases accuracy / probability, even as it increases confidence.
6) Regression towards the mean / average value to recognise the import of regression
the thought that punishment is more effective than a reward.
Availability a useful clue for assessing frequency of a class biases:
1) Retrievability of instances
2) Effectiveness of search set neglecting the fact that different tasks draw more search
sets into existence.
3) Imaginability incorrect evaluation of a frequency of classes.
4) Illusory correlation co-occurring event resistant to contradictory data.
Adjustment and Anchoring situational estimates based on the formulation of an issue are
biased towards initially chosen values (chosen by computation) underestimating
probabilities of failure in complex systems:
1) Insufficient Adjustment when an initial value, chosen through wrong computation,
is given and not enough adjustment is made to it.
2) Conjunctive and Disjunctive events conjunctive events are often overestimated,
while disjunctive events are underestimated.
Biases that occur due to the reliance on judgmental heuristics.
A better understanding of heuristics and their resulting biases could improve judgements and
decisions in situations of uncertainty.
