Lecture 10: Processing Uncertainty Flashcards
What did Gigerenzer et al (1998) find when he administered HIV tests to counsellors and asked them the likelihood of a man in a not a risk group having HIV given a positive test result?
Found the professionals could not accurately indicate the likelihood of this man having HIV. They failed to account for the base rate. The correct answer was that the man was 50% likely to have HIV given that he was in the low risk group and tested positive.
What is Bayes theorem? What tends to be the case though?
Bayes theorem rests on the mathematical analysis of the nature of probability and allows us to estimate posterior probability of a hypothesis given prior knowledge (base rate) and conditional probabilities (ie high v low risk group). This is a normative or prescriptive model of human logic however it appears that humans use more descriptive models.
Why is it argued that people don’t think logically?
That it humans are incapable of processing heaps of information. Brains would explode
What did Kahneman and Tversky (1973) find when looking at people and base rate neglect?
Had participants read descriptions of people and indicate how likely those people were to be an engineer or lawyer. In one group ps were told that within the sample there were 70 engineers and 30 lawyers and in the other group the opposite amounts were true. Found that people relied on the stereotypical information and the description and ignored the base rate when giving a career description. When no information was given, ps took into consideration the base rate. And when no career relevant info was given then ps guessed at a chance level.
What type of heuristic is likely to be at play from the findings of Kahneman and Tverskys (1973) study?
The representativeness heuristic - the probability of an event is estimated by evaluating how similar it is to the essential properties of the population. Ie stereotypes are taken into account out of proportionately.
Given the breast cancer example what would specificity and sensitivity represent?
Specificity - the likelihood of not having the disorder and testing negative (correct negative)
Sensitivity - the likelihood of having the disorder and testing positive (correct positive).
Air draw the little table to work out the probability of having something given a positive result (specificity, sensitivity, base rate).
Drawn
What do Gigrenzer and Hoffrage (1995) argue is a better way to represent probability estimations?
They argue it is better to represent the data as natural frequencies to better mirror the environment and are less likely to neglect base rate and be effected by heurisitcs.
Describe the results from gigrenzer and hoffrage (1998) that support that natural frequencies are a better method of displaying likelihood data.
Gave 48 physicians the breast cancer information either in frequencies or probability format. Found that they were more accurate at estimating the likelihood of the woman having breast cancer given a positive result when they were given the information in frequency format.
How did the defence phrase OJ Simpsons defence case ? How did Gigerenzer (2002) rebut this logic?
The defence argued that out of all the women that were battered annually by their partners (4 million) only 1432 were mirdered by their partners. Only 1 in 2500 cases of abuse ended in murder by partners. Gigerenzer (2002) said that the amount of people not murdered by their partners also needs to be taken into consideration. When it is, you find that the likelihood of being murdered by your partner given the cause is more like 89% or 8/9.
When taking into consideration if someone who tests positive to a something actually has the disease, what do you have to incorporate?
The base rate of the disease occurring in the population.
The likelihood of having the disease and getting a positive result and the likelihood of not having the disease and getting a negative result.
