Correspondence bias and development PDF

Preview

Summary

lecture on correspondence bias and development...

Description

PSY2203 Social Psychology II Date Lecturer

05/10/2018 Louise Pendry

Lecture No. Lecture Title

6&7 Correspondence Bias & Developments in Attribution Research

Summary • The correspondence bias/FAB is when we attribute behaviour to the person over the situation. • Evidence for the correspondence bias has been collected over many years in many different paradigms: Jones and Harris’ attitude-attribution paradigm, Ross et al.’s quiz show paradigm, and Snyder and Frankel’s sex tape paradigm. • The actor-observer difference is the fact that when judging our own behaviour, we don’t have a correspondence bias. o We have self-knowledge and insight into our internal states. o Perspective taking  Storms (1973). • The causes of correspondence bias include: situations lack salience and go unnoticed by perceivers (Taylor & Fiske, 1975; Ross et al., 1977), underestimating the impact of the situation, differential memory decay, the belief that the person constructs the situation (Gosling et al., 2002), naive realism & the false consensus effect, anchoring and insufficient adjustment (Tverysky & Kahneman, 1974), cognitive busyness yields incomplete corrections of dispositional inference (Gilbert et al., 1988), prior information and context lead to inflated categorisation of behaviour (Trope, 1986). • Trope’s model for impression formation suggests that there are 3 sources of information that lead to behaviour categorisation and finally to attribution. The attribution can then be updated by situational inducements and prior information. o The FAB occurs when there is exaggerated behaviour categorisation from considering the stimulus, or insufficient adjustment for not accounting for the exaggerated categorisation. • Spontaneous trait inference (STI)  creating impressions without realising it. • STI paradigms: encoding specificity paradigm (Winter & Uleman, 1984), relearning paradigm (Carlston & Skowronski, 1995), reaction time paradigm (Uleman et al., 1996), and the falserecognition paradigm (Todorov & Uleman, 2002). • How the trait is applied (to the person or behaviour) doesn’t alter the correspondence bias. • Reasons for forming STIs include: distal goals, automatisation, dispositional inferences, and culture. • There are limits to the correspondence bias and STIs o Culture: differences between attribution in collectivist and individualistic cultures (Morris & Peng, 1994; Rhee et al., 1995; Krull et al., 1999; Morris et al., 1998). o Goals: accountability (Tetlock, 1985), outcome dependency (Vonk, 1999), suspicion of ulterior motives (Fein et al., 1990), the shift of focus of attention (Quattrone, 1982), information processing goals (Uleman & Moskowitz, 1994), suppression, and mindfulness (Hopthrow et al., 2017) can all reduce or eliminate the correspondence bias. • We rely on dispositions because: we need control (Miller et al., 1978), we misperceive situations, or we fail to use information. • Gilbert’s model for attribution has 2 stages - identification and attribution (where there can be situational correction; Gilbert et al., 1988). o When we are occupied we don’t have the capacity for situational correction. • Krull revised Gilbert’s model to produce the mixed model of social inference (three stage model) we either start with the disposition and correct for the situation, or vice versa. • Kids have good trait vocabulary from a young age and know how to apply it as well, but we don’t know if kids also correct for the situation. o There is evidence that they discount information even when alternative explanations are available. o With age children show more adherence to the three stage model of social inference (Haga et al., 2014).

Correspondence Bias and Spontaneous Trait Inferences (Moskowitz, pp. 267-309)  Correspondence bias/Fundamental Attribution Error  attributing behaviour to the person (i.e. dispositional factors) rather than the situation (i.e. situational factors) even when the situation plays an important role in influencing behaviour. o E.g. Abu Ghraib Case Evidence for correspondence bias  There are countless studies in the literature that demonstrate the power of the situation over behaviour (e.g. Milgram’s obedience studies, Latane & Darley’s bystander studies, Asch’s conformity studies).  Jones and Davis (1965) believed that increasing the power of the situation of behaviour should decrease the belief that behaviour is due to disposition (correspondence bias).  Jones and Harris tested Jones and Davis’ theory – but found that they were wrong. The Jones & Harris attitude attribution paradigm  The study conducted by Jones and Harris took place in the 60s in the middle of the Cuban missile crisis – at this time, a student would expect their peers to be anti-Castro.  In their study, participants were given essays to read that were either pro-Castro or anti-Castro. Some of them were told that the essays were written freely (i.e. the writer chose which position to take, whether pro-Castro or anti-Castro), the rest were told that that writer was given a position to take in writing the essay and therefore did not pick whether to be pro or anti-Castro.  They were then asked whether they thought the essayist was pro or anti-Castro, by giving a rating from 10 to 70, where the higher the number the stronger the pro-Castro attitudes they believed the writer had. Expected results Choice No Choice   



Essay direction Pro-Castro Anti-Castro 60.00 15.00 20.00 20.00

Actual results Choice No Choice

Essay direction Pro-Castro Anti-Castro 59.62 17.38 44.10 22.87

As was expected there was a difference between pro- and anti- positions, when the writer was antiCastro they were seen as less pro-Castro and vice versa. There was also an effect on choice, so when the person was believed to have chosen to write proCastro they were seen as such, and vice versa. However, when the participants were told that the writers had no choice in their position, the situation should have resulted in them adjusting their opinions, i.e. if they were forced to write a pro-Castro essay those were not their real attitudes and they should have thought the writers to be anti-Castro as was the norm at the time. o But the participants interpreted the writers’ attitudes as pro- or anti- regardless of whether they had chosen the position or not. o The participants failed to account for the situation – they did reduce their interpretations a little bit (44.10 v. 59.62) but not as much as would have been expected. Our behaviours appear to be always considered a result of our disposition even if they are not – the implications of this are powerful and, in some cases, worrying.

Is correspondence bias an artefact of methodology?  Jones (1979) proposed a series of limitations to the methodology of the original study which could have affected the results: o In the context of an experiment, research participants would not have expected to receive information that was completely invalid to the question they were being asked, therefore, if they received an essay, they would assume that it had some truth to the attitudes of the person in question or they would not have been given it.



o The essay could have been written very well, this could lead the participants to conclude that only someone who could write so eloquently and passionately about a subject would actually believe it – therefore that could have persuaded them that the person was pro- or anti-Castro more than the actual contents of the essay. o The participants could have assumed that the no-choice condition was actually the one where they had the most choice – affecting the results collected. Other limitations: o Participants maybe didn’t understand the study – they don’t know that debaters are assigned topics. o Did not believe that someone would write an essay on something that you don’t believe in.

The Ross, Amabile, and Steinmetz quiz show paradigm  Ross, Amabile and Steinmetz in 1977 wanted to further test the correspondence bias in the form of a quiz show.  Participants were put into groups of 3 and each participant was given a role: the quizmaster, the contestant, and the observer.  The quizmaster was clearly told that they had to make questions in their own area of expertise, whatever that may be – this point was also made clear to the other participants in the group.  The observer would then be asked what they thought about the intelligence of the quizmaster and the contestant – ideally, the observer should adjust for the situation and realise that the questions were impossible for the contestant to answer, but this was not what they found.  The contestant was routinely seen as less intelligent than the quizmaster. The Snyder and Frankel sex tape paradigm  In this experiment male participants were given a soundless tape of a woman being interviewed by a man.  Half of them were told that the woman was being interviewed about her sex life, the other half were told that she was being interviewed about a bland topic.  The assumption is that accounting for the situation would suggest that the women in the sex tape was not dispositionally anxious as it was clearly the situation that made her anxious – however, the participants saw the sex tape woman as more dispositionally anxious than the bland tape woman.  This provides evidence to say with certainty that the correspondence bias is a real phenomenon. The actor-observer difference  Jones and Nisbett (1971) coined the term actor-observer difference – when judging our own behaviour, correspondence bias disappears.  One possible reason for this is that we have self-knowledge and insight into our internal states, whilst an observer only has the behaviour they observe and the situation they are in. We have no need to infer our motives from behaviour as we would have to do as observers, but we can analyse those motives directly.  There are differences in consistency and distinctiveness information for both the actors and the observers – the observer does not know how the actor behaves usually. Actors can detect covariation and identify changes in their behaviour over time to account for their behaviour in the moment, but an observer only has what they see in front of them in that moment. o But this does not mean that actors have perfect insight into their behaviour, they can be biased to favour one explanation over an undesirable one.  If someone has been in multiple frustrating situations such that they frustration is building up and they then meet someone who tips them over the edge, the observer and the actor will perceive the outburst in very different ways.  There are different aspects of the environment that are salient to the observer and to the actor. The actor would focus on their environment and the observer, whilst the observer would have as a focal point the actor’s behaviour – situation and behaviour are not seen equally by both parties.

Perspective taking  According to Heider (1958) in order to accurately attribute behaviour, we should be able to understand the psychological state of the other.  The idea of perspective taking has been consistently regarded as an essential ability of social functioning: o Davis (1983) found that perspective taking positively correlated with social competence and self-esteem. o Piaget (1932) considered it to be a big cognitive milestone. o Kohlberg (1976) appreciated its importance in moral reasoning.  Perspective taking has been found to have powerful effects on behaviour, inciting altruism and aggression when needed – being able to empathise with someone makes it more likely that we will help them, since we experience the emotions the target must be feeling.  However, the ease of the existence of the correspondence bias suggests that perspective taking is not the default method of social functioning – however, by participating in this process, observers will be able to understand what guides actors’ behaviour other than the situation. Perspective taking and the fundamental attribution error  The assumption of incorporating perspective taking into the fundamental attribution error is that it should eliminate it – by taking the perspective of the other person, there should no longer be an actor-observer effect.  Storms (1973) o Set up a situation where there were two observers and two actors. o The following diagram shows how they were set up, the arrows indicate what the individual is looking at:

o Next the participants were either shown the tapes from this original set up, or they were shown the tapes of the experimental set up, as shown by the diagram, again the arrows indicate what they are looking it (the actors looked at videos of themselves):

o The participants were then asked to rate their own behaviour if they were the actors, or the behaviour of the actor if they were the observer, on several personality dimensions, and for each dimension they had to say how much they thought that influenced personal characteristics and the situation. o The hypothesis was that actors would attribute behaviour to the situation and observers to disposition – to an extent this was what they found, but actors did not attribute as much to





situation as was expected, such that they equally attributed behaviour to both situation and disposition.  These were the same results as the Ross et al. (1977) study – actors attribute behaviour to situation and observers to disposition. o But, when their perspective was shifted they found that actors attributed behaviour more to disposition, and observers to situation – therefore perspective taking influenced the actorobserver effects. Regan & Totten (1975): they simple asked participants to take the perspective of the other person without actually placing them in that perspective, and they found the shift from dispositional to situational attributions as well. Whilst there is evidence that perspective taking is an effective method of accurately attributing behaviour, it is clear that this does not happen naturally or spontaneously, so they may be some others factors that contribute to the over-attribution of actors’ behaviour to disposition.

The causes of correspondence bias 1. Situations lack salience and therefore go unnoticed by perceivers  Since situational factors cannot physically be seen people will attend to what they can see and assume that that is the dominant causal factor for the behaviour.  Taylor & Fiske (1975): o When a participant was observing a conversation was seated to look at confederate 1, they said that confederate 1 was the dominant individual in the conversation, however, if they were seating to look at confederate 2, they would say the same of this confederate. o Shifting the observers’ perspective produced completely different results.  Once salient behaviour has been observed, other situational factors that caused that behaviour go unnoticed, so the only explanation the observer can produce for the behaviour they are observing is disposition.  Observers have the tendency to assume everything comes from one perceptual unit – behaviour came from the person, so the person is the cause.  Other study example: Ross et al. (1977) – participants overlooked the role-conferred advantage and focused on disposition instead.  If you don’t see a situation, you can’t subtract it out of the behavioural attribution – when people look at contestants on quiz shows they assume that they can be better than them, but they don’t consider the stress of being on television or having millions of people watching you. 2. Underestimating the impact of the situation  Perceivers may recognise the situation but still show correspondence bias, this could be because: o A) they are not aware of the situational constraints on the actor and how this affects their behaviour in the situation. o B) they are not aware of the power of those constraints, so perceivers will not view a situational force as strong enough to influence the behaviour of the actors even though it does (e.g. in Milgram’s study participants were not expected to comply and shock the confederates, but they did)  underestimated constraints. 3. Differential memory decay for pallid versus salient events  Sleeper effect  over time the situation fades into memory and you only remember dispositional causes for past behaviour. 4. The situation is the person  People will assume that the person constructs their situation, and to an extent this is true.  Gosling et al. (2002)

o Participants were brought to 5 office buildings to look at 94 offices and were asked to form an impression of the person who inhabited that office simply by seeing what they had in the office and without actually meeting the person. o Each office occupant had filled out a personality questionnaire and nominated 2 people to also fill out a questionnaire on the occupant’s personality. o The questionnaires of the participants and the office members were then compared. o They found that the personality scales were highly correlated, and that the participants were able to accurately predict the office occupant’s personality.  The effect becomes a bias when people assume that actors have influenced their situation when in fact they have not. 5. Naïve realism and the false consensus effect  False consensus is when we assume that everyone else would behave in the same way that we behave, this leads us to make less dispositional inferences about our own behaviour, since we assume there is high consensus.  If someone behaves in a way that is not expected, then their behaviour is seen as completely their own and will lead to dispositional explanations. 6. Anchoring and insufficient adjustment  The idea of anchoring is that judgements will be biased by initial estimates.  Tversky & Kahneman (1974) o Participants were asked to estimate how many African countries are in the United Nations. o They saw the experimenter spin a wheel with numbers on it and were then asked whether the actual number was higher or lower than that initial number from the wheel. o Participants who started with the number 10 gave substantially lower estimates than those who were given the number 65 as an anchor – even though the number was completely random. Tversky & Kahneman (1974)  o Participants were asked to multiply a string of numbers in 5 seconds – an impossible task. They only changed the order in which the number string was given. o Once group got the numbers 1-8 in ascending order, the other in descending order. o Those who got the numbers in the ascending group gave the median estimate of 512, those in the descending group gave the median estimate of 2,250.  Jones (1979) extended this to people suggesting that when we make a dispositional inference about a person, we will then use that as an anchor for our future inferences about them. o However, people don’t start with a neutral inference, and the idea is that future information will influence and change that initial inference but people suffer from insufficient adjustment so that a first impression can never be invalidated. 7. Cognitive busyness yields incomplete corrections of dispositional inference  Making adjustments from initial inferences requires cognitive effort, therefore, if a person has cognitive load they would succumb to the correspondence bias and be unable to correct for the situation.  Gilbert, Pelham & Krull (1988) o Replicated the Snyder and Frankel sex tape paradigm but now half of the participants were given a cognitive load – they were told to rehearse the discussion topics in the video. o They found that when participants had a cognitive load they were unable

to account the situation and considered the woman in the video as equally dispositionally anxious in both conditions (sex and bland). 8. Prior information and context lead to inflated categorisations of behaviour  Trope’s (1986) model shows how cognitive load is not needed for the correspondence bias to occur anyway. o The model considers 3 types of information that feed in to the final attribution: the observed behaviour, the situation, and prior information. o His model begins ...