Judgement and Decision making and Problem Solving Flashcards
What is Bayes’ theorem?
Everyday life is full of cases in which the strength of our beliefs is increased or decreased by fresh or new information.
Need to know prior ratio (probability before data is collected) –> p(HA) / p(HB)
And the likelihood ratio (relative probability of collecting the observed data) –> p(D/HA) / p(D/HB)
Bayes developed a way of conceiving about judgements. Many things increase or decrease our inference of something, and Bayes came up with the formula.
Odds of something happening or not, observation, ability to detect when this thing is happening. Prior odds and posterior odds.
What is the Bayes illustration: Tversky and Kahneman (1972)?
- A taxi-cab was involved in a hit-and-run accident one night
- Of he taxi-cabs in the city, 85% belonged to the Green company and 15% to the Blue company
- An eyewitness identified the cab as a Blue cab
- However, when her ability to identify cabs under appropriate visibility conditions was tested, she was wrong 20% of the time and correct in 80% of cases
- The participant had to decide the probability that the cab involved in the accident was blue
Mustn’t ignore the ‘base rate’ probability, the prior ratio.
What is base rate neglect?
If people make snap judgements they focus on the posterior odds, we are biased to observing the data rather than getting the data in the first place.
People are not as good as understanding probabilities etc in base rate information in judgements, better at natural frequencies. This is because these are the types of figures we deal with in the natural world. More psychologically digestible.
About neglecting base rates…
Koehler (1996, p1) defined base-rate information as ‘the relative frequency with which an event occurs or an attribute is present in the population’
People often fail to take base rates fully into account. Why?
Tversky and Kahneman (1972): Taxi cab problem: participants said there was an 80% likelihood that the taxi was blue.
Tversky and Kahneman (1982) repeated study.
In the ‘causal’ condition, the problem was rephrased: Ps were told that although taxi firms were equal in size, 85% of the accidents involved green cabs
Estimated probabilities that a blue cab was responsible for the accident in causal and control conditions are shown below. Base rates are not so neglected in this example.
About heuristics and biases…
- Kahneman and Tversky suggest that we are prone to errors because we rely on simple heuristics
- Heuristics are simple, efficient rules, and can be hard-coded by evolutionary processes or learned
- Heuristics work well under most circumstances, but in certain cases lead to systematic errors or biases.
What is the representative heuristic?
Why do we fail to make proper use of base rate information?
Kahneman and Tversky suggested that people often use a representativeness heuristic (rule of thumb): "events that are representative or typical of a class are assigned a high probability of occurrence. If an event is highly similar to most of the others in a population or class of events, then it is considered representative" Kellogg, 1995
People judge the probability that an object A belongs to certain class of objects B.
For example in the Jack (lawyer/engineer) example or judge the likelihood of him being an engineer based on the similarity between the description of your stereotype of the job.
What is conjunction fallacy?
Further evidence of the representativeness heuristic is the conjunction fallacy (Kahneman and Tversky, 1983).
Mistaken belief that the probability of the conjunction of two events (A and B) is greater (ie more likely) than one of the two events separately (A or B)
A form of representativeness - a bias.
The conjunction fallacy tells us that it is less likely to be a combination of two things than to just be one.
Wrong to think that the conjunction of the two things together is greater than each of the single things. This is because of the representativeness.
What is the availability heuristic?
Estimating the frequencies of events on the basis of how easy or difficult it is to retrieve relevant information from long-term memory eg Tversky and Kahneman (1974)
Differences in availability (retrieval ability or fluency of examples) lead people to misjudge the relative frequency of the two categories of words.
(confirmation bias: orientated towards information that confirms our judgements and beliefs. Will orientate to ones that confirm our belief and ignore the ones that don’t. in todays day when we tell the news the things were interested in it pushes things our way and we ignore the things we don’t know or believe).
How available information is to you psychologically biases your judgement. Trying to rationalise a judgement.
What is the numerosity heuristic?
- Pelham, Sumarta, and Myaskovsky (1994) proposed a numerosity heuristic. This involves over-inferring quantity or amount from numerosity or number of units into which something is divided. For instance:
- People generally eat less when food is divided into small pieces
- Participants who were asked to estimate the value of sets of coins gave higher estimates when there were many coins
- High self-esteem seems to be enhanced if a single belief (eg I am creative, analytical, and verbal) about oneself is divided into several distinct statements (eg I am creative, I am analytical, I am verbal) (Showers, 1992).
We change our impression/inference of what’s going on based on the number of units that exist.
What is support theory?
Tversky and Koehler (1994) proposed their support theory:
- Any given event will appear more or less likely depending on how it is described. Need to distinguish between events themselves and the descriptions of those events. - A more explicit description of an event will typically be regarded as having greater subjective probability when it is mentioned explicitly compared to the same event described in less explicit terms. An explicitly description may draw attention to aspects of the event that are less obvious in the non-explicit description. - Memory limitations - people do not remember all the information (or possibilities) if it is not supplied - related to representativeness heuristic.
What is critical to making judgements is the assessment of how likely something is to occur, based on how its described. A more explicit description changes the subjective probability of the judgement.
More detail provided could mean that we think more likely, when actually not as less specific description encompasses more scenarios. Biases our judgement. This is seen in terms of insurance - ‘we cover everything’ or ‘we cover x, y, z, a, b, c…..’ and people will go for the second one.
What is evidence relating to support theory?
Johnson et al (1993) found supportive evidence in that Ps offered to pay more for insurance policies that covered a detailed range of illnesses than for one that covered all illnesses in general.
Support theory in conjunction with the discussed heuristics demonstrates when errors in judgements are made.
Insurance policy example.
Support theory tries to encompass why representativeness is so important and the detail.
What is a summary of heuristics?
• Heuristics are rules of thumb that we use to simplify decision making
• Overall, heuristics result in good decisions
• The loss in quality of decision is made up by the time saved
• However, heuristics can cause systematic biases or errors in decision even in experts
• There are some limitations however:
- Heuristic theory has failed to provide process models
- Some errors are made because participants misunderstand parts of the problem
- The research is detached from everyday life
- Individual differences are under-researched
Even though there is support theory, there are some limitations of looking at judgements just is that way. They are not a model - don’t provide a process model. Support theory goes some way to do that but doesn’t provide everything.
A lot of the time people misunderstand the question and this can lead to a number of errors being made. Using natural frequencies helps in this.
Individual differences: age, gender, personality, IQ, etc… means that we are unsure if these hold for all groups. So in this sense they are descriptive rather than predictive.
About fast and fugal heuristics…
- Gigerenzer and colleagues tried to enumerate some ‘simple heuristics that male us smart’
- Rapid processing of relatively little information
- One example was the so-called ‘recognition heuristic’
- A specific examples of the so-called ‘take the best heuristic’ - take the best ignore the rest
- Goldstein and Gigerenzer (2002) in their city recognition/population size example (Herne vs Cologne). Which has the larger population, Cologne or Herne?
- Which has the larger population, Herne or Cologne?
- Start by which city name you recognise: more likely this is a larger city
- Lets say you recognise both names
- What may be another clue
- Cathedral/football stadiums - cities with cathedrals/football stadiums usually have larger populations
- Because you know that cologne has a cathedral your answer is cologne
- Therefore you take the best heuristic. You search cues, stop after you find something that can discriminate between the options, then choose the outcome.
- However, Newell and Shanks (2003) found that the take-the-best is not always used. In particular, more information is considered (ie. Weighed) when the decision is important, for instance when you decide to marry someone. More complex decision making.
- We need to organise the cues hierarchically and this is not a fast or easy task.
- More research is needed to explain how and why certain heuristics are used over others.
With very little information idea of fast mapping. Show things in contrast to one another can get people very quickly to learn, through them discounting the rest of the information. Very successful way of learning - learning things in contrast to what they’re not.
With very little information can cascade to make a decision about something with very little knowledge.
You take the best heuristic having searched for cues. Once done that you make the decision - very fast.
However, wouldn’t use a fast and frugal heuristic in huge decisions like marriage etc. for more complex decision making/judgement don’t use something so fast and frugal.
What are causal models of judgement?
- If we’re so dumb, how come we’re so smart
- People are generally accurate in real life decisions, which is not reflected in artificial problems
- Easy to persuade people to take base rates into account when causal relationships are more explicit
What are the dual process models of judgement?
- Use of complex processes as well as heuristics
- System 1 - fast, automatic, associative, difficult to change, emotionally charges - most heuristics
- System 2 - analytical, slower, consciously monitored, flexible, effortful
What are normative theories of decision making?
- Focusses on how people should make decisions (best decisions/ideals) while de-emphasising how they actually make them. Typically developed by economists.
- Von Neumann and Morgenstern’s (1947) Utility Theory suggests that Ps treat decisions like gambles and that they seek to maximise utility.
- This was later modified to take into account subjective utility
How people should be making decisions. Came up with utility theory. We treat decisions like gambles and maximise the utility/benefit. Estimate the benefit vs the cost.benefit of the cost is going to be subjective to the individual.argue that we simply calculate the utility and cost.
What is expected utility theory?
- Expected utility = (probability of an given outcome) x (utility of the outcome)
- Eg lottery ticket had an 85% chance of winning £100. therefore the average over many weeks of lottery tickets would be £85. some weeks (15%) you would get nothing and a lot of the weeks 85% you would get £100. Therefore people should be willing to buy the ticket for anything up to £85
- They way the problem is described should not affect the decision
- However, people are not rational decision makers inline wit utility theory
- It does not seem that people make decisions in this way. Psychologists found big departures from the expected utility theory in how people make decisions
Rather than just cost vs benefit, also look at the probability of getting that.
What’s the benefit and how likely is it?
People do not make decisions on this basis - they are not rational. This model of decision making shows that this is how people should make decisions, but they don’t. the description of the decision problem should not influence the outcome. Should just be cost vs utility.
What are descriptive theories of decision making?
Focusses on and describes how people actually make decisions. Typically put forward by psychologists.
Prospect theory.
This approach considers how people decide amongst games (prospects).
Normative theories might be better in that they are how we should make decisions and make better choices, descriptive theories focus on how we actually make decisions.
The most developed of these is prospect theory.
What is prospect theory?
Kahneman and Tversky (1979, 1984) proposed a development of subjective expected utility hteory called prospect theory. This can be summarised sing the following graph.
Eg a loss of £10 has greater negative utility than the gain of £10 has positive utility.
Losses of any kind are weighted disproportionately to gains of the same amount. That is why the line is not linear. Steeper slope for loses.
Explains why there is risk seeking for loses and risk aversion for gains.
Show that they developed subjective utility theory and showed that we have the value of something and whether it’s a gain or a loss. Prospect theory predicts that it’s not linear. We have gains of certain values, where its not symmetrical, losses are worth more than gains. We are a lot more sensitive to losses than we are to gains - loss aversive. The slop is steeper for losses than it is for gains. We want to avoid losses.
This proposes that a loss of £10 has greater negative utility than a gain of £10 has a positive utility.
Risk seeking for loses, the loss may not come about. Happy to be risk aversive for gains. This changes how people make decisions.
In addition, people give too much weight to very small probabilities. Most refused this bet: $20 if a tosses coin came up heads and a loss of $10 if it came up tails. The bet provides an average expected fain of $5 per toss.
K and T also identified risk aversion in securing a gain and risk seeking in attempting to avoid a loss.
Dawes (1988) also identified the sunk-cost effect: extra resources are committed to support a previously unsuccessful decision.
What is risk seeking and risk averse decisions making?
In making choices, people are sensitive to outcomes and to degrees of risk.
However, people are also heavily influenced by how a decision is frames.
- When cast in terms of gains, people tend to avoid any risk - When cast in terms of losses, people seek out risk, presumably in hopes of avoiding it
This has also been shown to be the case with studies on surgical survival rates Edwards et al. (2001) 90% survival rate vs 10% death rate.
One of the things demonstrated nicely is framing effects. Framing effects are the way you cast information about the gains and the losses showing if people are risk averse or risk seeking.
People change their decisions dependent on the frame.