Judgements and Decisions
The rational normative model
The Rational model of the homo economicus: expected value
- Expected value of an event
- Value x probability
- Expected value of a decision alternative
- The sum of all possible returns multiplied by their repsective
probebilities
- e.g. value of econometrics degree: probability of graduating in
econometrics (0.65) * starting salary (€2200) = 1430
- Deviation from the mode → Illustration: Bingo
- According to calculations, you should be willing to pay a
maximum of €4,75 for one bingo card…
- The bingo card costs €5.50, but people play anyway
- Actual behaviour deviates often from rational decision rules,
because a number of values are subjective:
- Value is subjective
- Probability is often determined subjectively
- Probability and value are not always combined
in a rational manner
Deviations from the normative model
Not Homo Economici: Prospect theory
- Describes how psychological (subjective) values and probabilities differ from
objective probabilities and values
- Deviations are not entirely random, the:
- Can sometimes be systematically described
- Need not lead to bad decisions
- Theory consists of two elements, similar to the normative model
- Value
- Objective value €5,-
- Subjective value, perhaps €11,- , perhaps
nothing
- Also applies to money;
- From nothing to 100 feels great
- From nothing to 200 feels twice
as great?
- Law of diminishing returns, afnemend stijgende
curve. Subjective value increases in a steeper value, but the slope
becomes smaller over time.
- Example: the Asian Disease. People prefer the
certain option when it comes to gains. People become more risk-
taking when it comes to losses.
- The Reflection effect: a gain becomes
subjectively less valuable, the higher it is. This is the same for losses,
but losses feel more intensely than equal gains.
, - Loss aversion
- The value function is less steep
for positive values (gains) than for negative values (losses),
losses feel more intensely than equal gains.
- Probablity
- Small probabilities (< 10%) given too much
weight
- Moderate probabilities (> 10%) given too little
weight
- The theory does not indicate what happens
above 90%
- But 100% probability is given decision weight 1
Other Biases in Decisions and Judgements
- Bias in finding information
- In line with theory (tunnel vision)
- Bias in sample
- e.g. size, selective sample (e.g., only friends)
- Bias isn seeing covariation
- Illusory correlation → Lecture 7 on stereotyping
- Regression effect, suggests that we over estimate the meaning of extreme
effects. Extreme effects level out over time.
- Regression to the mean
- Extreme effect will, on average, be less extreme
at another point in time
- Effect applies to stable context, so, the same chef for a long
time.
- In instable context, extreme observation can be indicative of
change (e.g., new chef)
- Dilution effect
- Information 1 → He’s an alcoholic
- Additional ‘non-diagnostic’ information → Perhaps an
alcholic (not so sure)
- Diagnostic information diluted with non-diagnostic information
- Judgement becomes more moderate and less
certain
- Especially in observers who are motivated to
form an accurate impression
Heuristics
- Rules of thumb when estimating probability and quantity
- Representativeness
- The more characteristics A shares with B, the
more likely it is that A and B are associated, B can be the
consequence of A, or B can be an exemplar of category A
- Can lead to accurate judgements
- Can lead to biases, e.g., if initial probabilities
are ignored (base-rate fallacy). Initial probability is much greater that a
man works in construction industry than in a library
- Conjuction Fallacy