Greg Francis 5/27/081JudgmentIIE 269: Cognitive PsychologyGreg FrancisLecture 33Why clinicians need to be careful.Cognition We have talked a lot about theproperties of cognition perception memory attention language Now we start to discuss what is typicallythought of as thoughtJudgment Drawing conclusions from knowledge Why was I in a bad mood yesterday? Does she really like me? Will I pass the exam? What do we do? strategies for evaluating evidence influential factors inductive judgment Describe people’s behaviorInfluences Many factors affect judgment especially interested in those that seem arbitrary Availability Detecting covariation Base rates ReasoningAvailability Judgments are usually based on patterns ofobservations summarize data to find a pattern frequency estimate: how often has this eventoccurred? For example, your friend recommends amovie you have to consider the quality of your friend’srecommendations in the past often good, or often not good?Availability This approach to judgments means that yourdecisions will be based on what you canremember If you remember some things more than others, theywill have a bigger influence on your judgments A heuristic is a strategy that risks error to gainefficiency we use lots of heuristics in judgments Typically to our advantage, its not worth the effort totry to remember everything that is related to ourjudgmentGreg Francis 5/27/082Availability effects from bias Whatever biases our memory can affect ouravailability heuristic Consider causes of death Which is more common motor vehicle accidents or stomach cancer? homicides or diabetes? Rather important for deciding how to spendgovernment dollars!Availability effects from bias Media reportsbias whatseems familiarto us homicide andcar wrecks areheadline news(in some towns) stomach canceris hardlymentioned at allNotice Even when you know that such biases mayexist, you will probably not compensate for it Consider an effect called anchoring when an answer to a question is observed,subjects use the answer as a reference point even used when the answer is obviouslyirrelevant to the questionAnchoring Tversky & Kahneman (1974) (three groups) (1) subjects estimate the percentage of Africannations in the UN (actual number is 35%) (2, 3) subjects watch an experimenter spin awheel and watch it land on a “random number”» wheel actually rigged to land only on 10 or 65» subjects are split into two groups saw 10 -> give estimate of 25% African nations saw 65 -> give estimate of 45% African nations Thus fundraisers often suggest a donationCovariation Sometimes two things vary together exercise and stamina owning CDs and going to concerts hot weather and thirst We often have to judge how strongly twothings covary study time and test score education and salary spanking and deviant behaviorCovariation These types of judgments are also afundamental basis of science aspirin and heart attacks broccoli and cancer aluminum and Alzheimers religion and obesity Note, you have to make multipleobservations even to talk about covariationGreg Francis 5/27/083Illusory covariations Let’s look at an interesting example of a casewhere people report detecting a covariationwhere there actually is not Psychologists often use Rorschach tests toidentify certain personality traits ink blots subjects report what they see psychologist judges reports Do responses really covary with certainpersonality traits? can people (psychologists) detect this covariation?Illusory covariation Chapman & Chapman (1971) created many (fictional) written transcripts of asubject’s responses to a set of ink-blots created many (fictional) trait descriptions ofpeople who supposedly made the responses,e.g.:» “believes other people are plotting againsthim” (paranoid)» “has sexual feelings toward other men”(homosexual) Randomly paired a transcript with a persondescriptionIllusory covariation Subjects were shown many transcript-description pairs and asked to determinewhich response-types covaried withhomosexuality because of the random pairings, there was nocovariation of responses to inkblots with the trait ofhomosexuality in the description subjects reported it anyway» claimed that homosexual respondents werelikely to perceive buttocks in the inkblotsClinicians beware Interesting, the illusory covariation is onereported by clinicians Using real data, clinicians also claim there is arelationship between reports of buttocks andhomosexuality however, statistical tests show that there is no suchcovariation in the real data many clinicians continue to insist the relationshipexists, even though it does not Be careful interpreting Rorschach tests!Assessing covariation It seems then that people are often quite badat judging covariation “professional wisdom” is often untrustworthy if it isbased on personal experience instead ofsystematic study But people can fairly accurately judgecovariation if they simply work with data many problems are related to the availabilityheuristic we remember best what agrees with our viewsBase rates Consider the following problem: I have a friend who is a professor. He likes to writepoetry, is rather shy, and is small in stature. Which ofthe following is his field:» a) Chinese studies.» b) psychology. How do traits covary with profession?Greg Francis 5/27/084Base rates Many people say the friend is in Chinesestudies small stature, poetry, seem stereotypical of that field But, there are many more psychologyprofessors than Chinese studies professors so the odds are greater, overall, that the person is inpsychology this is the base rateBase rates Suppose 90% of Chinese scholars fit the stereotype And that only 5% of psychologists fit the stereotype Suppose there are 10,000 Chinese scholars 200,000 psychologists Then we get 90% of 10,000 => 9,000 5% of 200,000 => 10,000 So there are actually more psychologists that fit thestereotype than Chinese scholarsBase rates People often ignore base rate information whenthere is other information available Diagnostic information (stereotypical traits) areused over base rate information so is irrelevant information Often leads to serious