Human Relationships studies

Aim (Beaman et al.)

To determine whether people can be trained to overcome bystanderism

Procedure (Beaman et al.)

The sample was made up of 80 introductory psychology students divided between 4 conditions. In the helping film condition, participants watched a film explaining why people don’t help (Latane and Darley’s explanation of diffusion of responsibility, informational social influence, and evaluation apprehension) and simulated the studies done to demonstrate this. In the helping lecture condition, participants attended a lecture with the same content as the film, but without the simulations of the studies (they were just described). In the obesity and emotion lecture condition, participants attended a lecture describing the emotional factors related to obesity (no reference to helping behavior). There was also a control condition. After watching the film or lecture, participants were given a questionnaire to make sure they paid attention and understood the content. Two weeks later, the participants were told they were going to participate in a study on communication. The participant and a female confederate were told to enter a room set up with recording equipment for the communication experiment. There was a male confederate lying on the floor appearing in need of help, and the female confederate was instructed not to help or comment on the man lying on the floor. Helping was indicated by going to the victim, calling out to the victim, or telling the female confederate to get help.

Results (Beaman et al.)

Around 42% of participants who attended the helping film/lecture helped (no significant difference between these groups). Around 25% of participants in the obesity lecture/control group helped (no significant difference between these groups). The sample was matched for gender and no statistically significant difference in helping between genders was found. This indicates that teaching about helping behavior may increase prosocial behavior.

Evaluation (Beaman et al.)

Strengths: 

• Experiment (cause and effect can be established)

• Easily replicable (reliability can be established)

• Participants were debriefed and asked if they were able to determine the real aim of the communication study (any who said they did were excluded)

• Relatively high ecological validity despite being highly controlled

Limitations:

• Ethical concerns about deception (but it was necessary to avoid expectancy effect)

• Sampling bias (only US psychology students)

• Study only looked at emergency situations (not other forms of helping)

Aim (Decety et al.)

To determine whether there are differences in neural activity between incarcerated psychopaths and incarcerated controls when presented with empathy-eliciting stimuli depicting other people in pain

Procedure (Decety et al.)

80 incarcerated men were classified as having high, medium, or low levels of psychopathy using the Hare Psychopathy Checklist. Participants were placed in an fMRI and completed two counterbalanced conditions: being shown images of people being physically harmed and being shown people with facial expressions of pain.

Results (Decety et al.)

Psychopathic participants (high and medium scorers) showed significantly less activation in the ventromedial prefrontal cortex and lateral orbitofrontal cortex, which form part of the brain’s empathy circuit. This suggests that capacity for empathy is hard-wired and cannot be taught through programs promoting prosocial behavior.

Evaluation (Decety et al.)

Strengths: 

• Standardized procedure (can be replicated, so reliability can be established)

Limitations:

• Cause and effect cannot be established from MRI (correlational)

• Reductionist when only biological approach is considered

• Low ecological validity (doesn’t emulate empathy response in real life)

• Generalizability issues (only incarcerated people were studied)

• Construct validity (empathy isn’t the same as prosocial behavior)

Aim (Bradbury & Fincham)

To determine the impact of communication style on romantic relationships

Procedure (Bradbury & Fincham)

All participants were married couples that lived together and did not attend marriage counseling. The average length of marriage was 8.5 years. All participants completed a survey to determine their level of marital satisfaction. Before the observation, each participant individually filled out a questionnaire about marital problems. They individually asked each participant about the cause of and who was responsible for a common problem on their and their partner’s questionnaire. They were also asked the cause of and who was responsible for a problem their partner did not report. Each couple was then observed and recorded for 15 minutes as they discussed solutions to the problem they both reported. Three researchers independently coded the recording to determine relationship-enhancing and distress-maintaining communication patterns.

Results (Bradbury & Fincham)

Couples that reported lower levels of marital satisfaction in the initial questionnaire more frequently used distress-maintaining patterns of communication (attributing problems to the partner and viewing their intentions as selfish), and the interactions between these couples were viewed as hostile. Couples that reported higher levels of marital satisfaction in the initial questionnaire more frequently used relationship-enhancing patterns of behavior (not blaming partners for problems and assuming that negative actions were intentional).

Evaluation (Bradbury & Fincham)

Strengths: 

• Researcher triangulation (increases reliability)

Limitations:

• Correlational (bidirectional ambiguity and cause/effect cannot be established)

• Other variables like mental illness were not measured and could have influenced communication style

• Not all discussed problems were as serious between couples - distressed couples tended to discuss problems that were more difficult to resolve than non-distressed couples

• Doing the marital satisfaction questionnaire before the observation could have influenced behavior, and there was no counterbalancing to control for order effects

• All participants were from Western cultures, which could influence communication style (sampling bias)

• Cross-sectional design, so changes over time cannot be measured

Aim (Collins & Miller)

To investigate the link between self-disclosure and liking in relationships

Procedure (Collins & Miller)

A meta-analysis was carried out on research articles on the topic of self-disclosure. Researchers used journal articles published between 1955 and 1992, as well as studies in academic textbooks, and focused on isolating key terms relevant to self-disclosure and liking. The studies were varied, including lab experiments and questionnaires, and effect size was calculated from the studies.

Results (Collins & Miller)

Liking appears to be associated with self-disclosure, indicated by positive correlations and effect size between the two variables. The trend was more pronounced for questionnaires, but still present for lab experiments. Research also found a link between intimacy and liking in that people disclose more information to people they already feel closer to. Researchers concluded that self-disclosure plays an important role in the maintenance of relationships.

Evaluation (Collins & Miller)

Strengths: 

• Data triangulation increases the validity of the study

• Lower risk of researcher bias with meta-analyses

• Multiple statistical measures used leads to an objective and consistent procedure, which increases reliability of findings

Limitations:

• Researchers don’t have control over the secondary data, which limits reliability

• Lack of ecological validity as statistical measures do not account for how and why self-disclosure occurs, only that it does

Aim (Darley & Latane)

To determine whether the number of people present in an emergency situation influences likelihood of helping

Procedure (Darley & Latane)

Student participants were told they would be interviewed about personal problems faced by students as they move to university, and that they would be interviewed over an intercom to ensure anonymity. Some were told that there were five other people in a discussion group with them, some were told that there were two other people, and some were told that there was only one other person in the study. The comments the participants heard from other group members were pre-recorded. In the middle of the interview, one of the pre-recorded voices cried out for help and made choking noises as if they were having a seizure.

Results (Darley & Latane)

If participants thought they were the only person, 85% attempted to get help. If participants thought there was one other person, the rate of helping was 65%, and if participants thought there were four other people, the rate of helping was 31%. These results indicate that believing someone else will help lowers the probability of someone taking responsibility (diffusion of responsibility).

Evaluation (Darley & Latane)

Strengths:

• Lab experiment so highly controlled (lowers likelihood of extraneous variables)

• Cause and effect can be established

• Standardized and replicable procedure (reliability) 

Limitations:

• Ethical concerns of deception and undue stress/harm (probably couldn’t give informed consent)

• Low ecological validity (participants could only hear the victim, not see them)

• Sampling bias (only students, might not want to upset professors by interrupting study)

Aim (Piliavin et al.)

To determine how situational factors influence prosocial behavior

Procedure (Piliavin et al.)

The participants were an opportunity sample of travelers on the NYC subway observed between 11 am and 3 pm. While on a 7.5 minute stretch between stations, one of two scenarios would occur: a man with a cane who appeared ill or a man who appeared drunk (smelled of alcohol and carried a bottle in a bag) would fall down. The “victims” fell down 70 seconds after the train left the station and stayed on the floor until they were helped. A confederate was instructed to help the “victim” if nobody else offered assistance after 70 seconds. The data was recorded by two researchers. The independent variables were the appearance of the man who fell and the number of people on the train. The dependent variables were frequency of help, speed of help, sex of helper, movement away from the victim, and verbal comments.

Results (Piliavin et al.)

78% of the time someone helped the victim (95% for the man with the cane and 50% for the drunk man), and more than one person helped 60% of the time someone helped. The median response time for helping the man with the cane was 5 seconds, and the median response time for helping the drunk man was 109 seconds (it seems as though people were weighing the costs and benefits of helping the drunk man). 90% of the helpers were male (statistically significant). More comments were made the longer the victim was on the floor, and more comments were made when the victim appeared to be drunk. Diffusion of responsibility was not observed (the larger the group, the quicker people helped).

Evaluation (Piliavin et al.)

Strengths: 

• High ecological validity (field experiment)

• Covert observation, so there likely weren’t demand characteristics

Limitations:

• Low internal validity because of the field experiment (difficult to conclude cause and effect)

• Time taken to help was assumed to indicate weighing the costs and benefits of helping (problematic construct validity)

• All victims were male, which could have impacted reactions

• All American participants (limited generalizability)

• Ethical concerns (deception, no informed consent, no debriefing, potential undue stress about the situation or worrying about not having helped)

• Arousal-cost-reward model assumes people make rational decisions and does not explain altruism

Aim (Toi & Batson)

To determine whether manipulating participants’ level of empathy would influence likelihood of helping

Procedure (Toi & Batson)

Participants listened to an interview about a fellow student named Carol who broke both her legs in a car accident and spent the last month in the hospital. Although she recovered, she may need to drop the introductory psychology course, which would put her a year behind everyone else. In the low empathy condition, participants were asked to listen closely to the information in the interview. In the high empathy condition, participants were asked to imagine how the person being interviewed felt about her situation. After listening to the interview, all of the participants were given a questionnaire about their emotional response. Then they were given the opportunity to help Carol by going over their notes with her. Participants were given the possibility to “escape” this responsibility by being told whether or not they would see Carol in the future. In the easy escape condition, participants were told that Carol’s legs were still in casts, so she was studying at home, but could easily arrange transportation so they could help. In the difficult escape condition, participants were told that Carol was in the same psychology discussion group as them, and that she would be back in class next week.

Results (Toi & Batson)

Participants in the low empathy condition reported feeling less empathy than participants in the higher empathy condition. They also helped less when the escape was easy than when it was difficult. This indicates that the low empathy participants egoistically wanted to reduce their own distress. Participants in the high empathy condition had high rates of helping even when escape was easy. This indicates that the high empathy participants altruistically wanted to reduce Carol’s distress.

Evaluation (Toi & Batson)

Strengths: 

• The results have been replicated, indicating high reliability

Limitations:

• Only short-term altruism was investigated

• Did not consider personality factors (participant variability could influence results)

• Construct validity (difficult to measure level of empathy)

• Low ecological validity (you generally see the person you are helping before you help them)

• Demand characteristics (social desirability, particularly for “altruism”)

• Can’t fully control what participants are focusing on during the interview (limited internal validity)

• Only female participants/female person in need (limited generalizability, empathy could have different impacts on different genders)

Aim (Drury et al.)

To determine whether priming a group identity would make it more likely for people help others during an emergency situation

Procedure (Drury et al.)

Participants experienced a virtual reality simulation of a fire in the London metro. They could make the decision to either help people or push them out of the way to escape safely. Before the simulation, the participants were primed by reading a news report about a fire in Kings Cross Station that killed 31 people, and were asked to imagine the sensory details associated with that scenario. The participants in the group-identification condition were given a scenario stating they were coming back from an England football match on a train with fellow supporters. The participants in the individual-identification condition were given a scenario stating they were travelling back to the university after going shopping. To make the group identity salient, the other people in the group condition simulation wore the same colored shirts, while the other people in the individual condition simulation wore different colored shirts.

Results (Drury et al.)

Participants with a higher in-group identification were more likely to give help, and those without in-group identification were more likely to push people out of the way.

Evaluation (Drury et al.)

Strengths:

• High internal validity (controls for extraneous variables)

• High mundane realism

• Replicable, which increases reliability

• Highly ethical because undue stress/harm and deception were avoided

Limitations:

• Low ecological validity because the participants didn’t feel like their life was in danger

• Independent samples means participant variability could impact results

Aim (Wedekind)

To determine whether one's major histocompatibility complex (MHC) affects mate choice

Procedure (Wedekind)

The sample was made of students at the University of Bern, Switzerland, and each participant was “typed” for their MHC, with a wide variety being included in the sample. The men and women were taking different courses, so that they likely didn’t know each other, and the women taking oral contraceptives were noted. The men were given a T-shirt to wear for two nights and place in an open plastic bag during the day. They were given perfume free detergent to wash clothes and soap for showering, and asked not to smoke tobacco, drink alcohol, eat spicy foods, or have sex. The women were asked to use a nose spray for two weeks leading up to the experiment to support the regeneration of the nasal mucous membrane and prevent colds, and were also given a copy of the novel Perfume to heighten their sense of smell. The women were asked to rank the smell of 7 of the T-shirts each placed in a cardboard box with a smelling hole, preferably within the second week after menstruation when their sense of smell is most sensitive. 3 of the boxes had shirts from men with similar MHC genes to the women, 3 had shirts from men with different MHC genes, and 1 was an unworn control. The shirts were ranked from 1-10 for intensity and pleasantness/sexiness, and the women were alone in the room where they smelled the shirts.

Results (Wedekind)

Women not taking oral contraceptives rated men with different MHC genes as having a more pleasant smell, and women taking oral contraceptives rated men with similar MHC genes as having a more pleasant smell, indicating that the MHC gene does impact mate choice.

Evaluation (Wedekind)

Strengths:

• The study has been replicated 

• Ability to detect MHC is important as studies have shown that couples with more similar MHCs are more likely to have miscarriages or children with a lower birth weight

• Double-blind (minimizes researcher bias and demand characteristics)

• Ethical standards met (informed consent and debriefing)

Limitations:

• Reductionist argument

• Not representative (similar age and culture of participants)

• Construct validity (pleasantness/sexiness does not necessarily indicate attraction)

• Low ecological validity (you don’t determine whether you like someone based on smell alone)

Aim (Moreland & Beach)

To test the validity of the mere exposure effect under naturalistic conditions

Procedure (Moreland & Beach)

All of the participants were undergraduate psychology students taking classes in a large lecture hall. Four female confederates attended a different number of class sessions over the course of the experiment. One attended no sessions, one attended 5 sessions, one attended 10 sessions, and one attended 15 sessions. Each time one of the confederates attended a class session, they would arrive a few minutes before the class began, walk slowly towards the front of the lecture hall, and sit where all the other students could see her. She listened and took notes during the lecture, and a few minutes after class ended, she would slowly get up and walk to the back of the lecture hall to leave. None of the confederates were allowed to interact with the students. At the end of the semester the students were shown pictures of the four women and asked to rate them on a scale of 1-7 for several traits, including attractiveness, popularity, intelligence, warmth, honesty, and success. The students were also asked whether they knew the woman or if she looked familiar.

Results (Moreland & Beach)

Both men and women responded to the women in similar ways. When the women attended no classes, she was rated a 3.62 for attractiveness, when she attended 5 classes she was rated a 3.88, when she attended 10 classes she was rated a 4.25, and when she attended 15 classes she was rated a 4.38. The participants did not say they found the women familiar. This demonstrates that the mere exposure effect can occur under naturalistic conditions.

Evaluation (Moreland & Beach)

Strengths: 

• High ecological validity (high mundane realism of situation)

Limitations:

• The women may have had different levels of baseline attractiveness (confounding variable)

• All participants were American university students (generalizability issues)

• Difficult to prevent participants from interacting with the confederates (low internal validity)

Aim (Carrere & Gottman)

To determine whether psychologists could accurately predict whether a couple in marriage counseling would end up divorced based on only a few minutes of conversation

Procedure (Carrere & Gottman)

Researchers developed the Specific Affect (SPAFF) coding system to measure positive and negative affects during the interaction. The positive affects included interest, validation, affection, humor, joy, and the negative affects included disgust, contempt, belligerence, domineering, anger, fear, defensiveness, whining, sadness, stonewalling. SPAFF also included neutral affects. Newlywed couples were recruited using purposive sampling to stratify the sample in terms of economic background and ethnicity. Each husband and wife individually completed a survey and discussed the results with a researcher to identify one or two problems in their relationship. The identified issues were the basis of a 15 minute discussion between the couples that was recorded and then coded using SPAFF. Researchers then checked in with the couples once a year for six years to determine whether they were still married.

Results (Carrere & Gottman)

The differential in positive emotions and negative emotions was about 50 for married husbands and 45 for married wives. The differential was about -15 for divorced husbands and -20 for divorced wives. Gottman concluded that communication styles play a central role in success or failure of relationships, and further analysis led to the development of his Four Horsemen of the Apocalypse theory. He argues that these communication styles place couples in a vicious circle where their negative behaviors reinforce the negative behaviors of their partners.

Evaluation (Carrere & Gottman)

Strengths: 

• SPAFF is a standardized test with high reliability

• Researcher triangulation establishes inter-coding reliability, increasing internal validity

Limitations:

• Selection bias in that participants volunteered for the study (they are likely either very happy with their relationship or very unhappy)

• Reductionist

• Extraneous variables that occurred over the six years could play a role in divorce (problematic internal validity for longitudinal and prospective studies)

• Cause and effect cannot be established (correlational)

Aim (Felmlee)

To test the fatal attraction hypothesis as an explanation for the end of relationships

Procedure (Felmlee)

300 students at the University of California completed a questionnaire in which they listed the qualities that first attracted them to a recent partner as well as the qualities they liked least about the person.

Results (Felmlee)

Felmlee found that about 30% of breakups were fatal attraction breakups. She found three common fatal attraction patterns: fun to foolish, strong to domineering, and spontaneous to unpredictable. This data provides some support to fatal attraction theory.

Evaluation (Felmlee)

Strengths: 

• Large sample size increases reliability

• Anecdotal evidence and real-world observation support Felmlee’s findings (increased validity)

Limitations:

• Limited generalizability (only American university students)

• Cultural bias towards individualistic cultures (relationships in collectivistic cultures tend to be centered more around family priorities than individual preferences)

• Self-reported data cannot be verified (low internal validity)

Aim (Dawes et al.)

To investigate whether one’s decision to cooperate is based on greed or a fear of loss

Procedure (Dawes et al.)

Participants were allocated into groups of 7 people and each participant was given $5. They were told that if at least 3 people gave the $5 to the researcher, everyone would receive $10 (if you gave the $5 away and got $10 you would have $10, but if you kept the $5 and other people donated you would have $15). If less than 3 people gave away the $5, then nobody would get more money and the money given away would not be returned. There was also a no fear condition in which researchers told participants that if not enough people gave the $5 to the researcher they would get their money back, and a no greed condition, where participants were told that nobody would end up with more than $10 total (even the people who didn’t donate when other people did).

Results (Dawes et al.)

In the control condition and the no fear condition, 50% of participants donated the money. In the no greed condition, almost 90% of participants donated the money. This indicates that it is not fear of losing money, but rather greed that led people to not cooperate.

Evaluation (Dawes et al.)

Strengths: 

• Random allocation and researcher manipulated IV means cause and effect can be established

Limitations:

• Low mundane realism, so low ecological validity

• Participant variability is likely an issue, as some people could just be more greedy or altruistic than others

Aim (Kerr)

To determine whether watching someone else cooperate influences likelihood of cooperation

Procedure (Kerr)

Undergraduate students were asked to pump air using a small rubber bulb for 30 seconds, with the air being collected and measured with a spirometer. In one condition the participant was alone, in another they were working with a capable partner who worked hard, in another with a capable partner who was not working hard, and individually but at the same time as someone else doing the test and not putting in much effort.

Results (Kerr)

Participants working with a partner who worked hard pumped less air than those who worked by themselves, demonstrating the free-riding effect. Participants working with a partner who did not work hard pumped less air as well, demonstrating the sucker effect because they did not want to do more than their fair share of work. The results in the individual condition at the same time as another person had results similar to the control condition. This demonstrates that being willing to cooperate involves not wanting to do more work than everyone else and not wanting to do more work than you need to.

Evaluation (Kerr)

Strengths: 

• Standardized procedure increase reliability

• Random allocation and researcher manipulated IV allows cause and effect to be established

Limitations:

• Independent samples design means participant variability could influence results (some people might be stronger and therefore pump more air than others)

• Low mundane realism, so low ecological validity (it wasn’t an important task so people might have been less likely to cooperate)

Aim (Fein & Spencer)

To test the theory of threatened egotism as an explanation for prejudice

Procedure (Fein & Spencer)

All participants were male psychology students, and they were randomly allocated to either the control or negative feedback condition. All participants took an online test called “The Reasoning and Verbal Acuity Battery” which included impossible questions. The control group was told that the test was a fake intelligence test, but the negative feedback group believed it was an actual measure of intelligence. All of the negative feedback group did poorly on the test. The participants were then given one of two scenarios: both about a struggling male artist (Greg) in New York and his hopes to write his own play, but in one it is mentioned that he lives with his girlfriend Anne, and in the other the word “partner” is used without a name (this was meant to imply that Greg is gay). After reading the scenarios, participants rated Greg’s personality on 3 generic traits, 7 stereotypical of gay men like femininity, and also whether they would like Greg as a friend and how similar they thought they were.

Results (Fein & Spencer)

Participants who thought they had done poorly on the intelligence test rated the “gay implied” Greg’s personality more stereotypically gay than those who did not think they did poorly. Those who thought they did poorly were also less likely to say they liked Greg than the control condition regardless of whether it was implied that he was gay (although there was more dislike in the gay implied condition).

Evaluation (Fein & Spencer)

Strengths: 

• A pilot study was conducted that found that explicitly saying Greg was gay caused participants to recognize that the study involved stereotypes about gay men → they changed the procedure to minimize demand characteristics

• Highly standardized and can be replicated for reliability

• High internal validity

Limitations:

• Low ecological validity (artificial and does not indicate how people would act in a real situation)

• Assumes that the men had strong self-image because they were at a competitive university → not necessarily true

• Participant variability regarding attitude towards gay men could be an issue

Aim (Harris & Fiske)

To observe the role of the prefrontal cortex and the amygdala in reacting to "extreme out-groups" - homeless and addicts

Procedure (Harris & Fiske)

22 participants were randomly allocated between seeing images of people and objects. Before being placed in the fMRI, participants were asked to assign a series of neutral photos an emotion of either pride, envy, pity, or disgust on a computer screen so that they understood the task. In the fMRI, participants were shown six sets of ten images, including images of people with disabilities, rich businessmen, elderly people, American olympians, and homeless people. After seeing the image they used a joystick to select the emotion they felt upon seeing the image.

Results (Harris & Fiske)

Regardless of the selected emotion, participants in the objects group showed no activity in the medial prefrontal cortex, although this part of the brain was almost always active in the people group. When people saw images of addicts or homeless people, the amygdala was activated as well as a series of disgust reactions. The insula, typically a response to garbage or human waste was activated, although the medial prefrontal cortex, which is activated when we think about people, was not, suggesting that the brains of the participants did not perceive the homeless or addicts as people. This indicates that prejudice may have an evolutionary component.

Evaluation (Harris & Fiske)

Strengths:

• The study is a true experiment (cause and effect is implied)

• Repeated measures within the people group (the different categories of photos shown, decreases participant variability)

• fMRIs do not allow for demand characteristics

Limitations:

• Unclear the extent to which marginalized groups cause activation of the insula

• Small sample size (because fMRIs are expensive, more research is required to determine reliability)

• Biased sample (all Americans attending Princeton University)

• The study did not observe homeless people seeing images of other homeless people for example, so it is not conclusive that the brain is necessarily responding to the person being a member of the out-group

Aim (Phelps et al.)

To determine the relationship between neural correlates of unconscious evaluation and level of prejudice

Procedure (Phelps et al.)

White American participants were shown the faces of Black and white men (both familiar and unfamiliar) in an fMRI and then took the Implicit Association Test as a standardized test for ethnic prejudice.

Results (Phelps et al.)

Those who had higher levels of prejudice showed more activation of the amygdala when shown the faces of unfamiliar Black men. The amygdala did not respond to the same extent when participants were shown the faces of familiar and positively regarded Black men. This indicates that observing members of different social groups activates the amygdala differently, and that the amygdala plays a role in unconscious social evaluation.

Evaluation (Phelps et al.)

Strengths: 

• Demand characteristics are not an issue in an fMRI

Limitations:

• fMRIs do not indicate causality

• Only tests white Americans, who may have different levels of prejudice than other groups (sample is not representative)

• Reductionist

• Seeing a picture may not lead to the same response as seeing a real person (low ecological validity)

Aim (Sherif et al.)

To study the development of group organization and test Realistic Group Conflict Theory

Procedure/results (Sherif et al.)

The participants were 11-12 year old boys with similar backgrounds, none of whom knew each other. The participants thought they were attending a summer camp and were divided into two groups. During activities the groups started dividing tasks and hierarchies emerged within the groups. The groups developed their own norms and began ridiculing group members who did not do well at the activities to maintain social control. The groups eventually developed a clear group identity as the Eagles and the Rattlers. Conflict was then introduced through competitions and games. The groups started refusing to interact with each other, called each other names, and turned against their previous friends. Solidarity within groups and hostility towards the out-group increased. Fights and bullying became common, which demonstrates that conflict can result from group identity and righting for resources. The participants were asked to list features of both groups, and they listed much more favorable characteristics for their own group than the other group. Researchers wanted to stop the conflict, and after initial attempts did not work, they tried to introduce a common goal that would force the groups to work together. They introduced superordinate goals, like needing to work together to pull the camp truck after it broke down. Doing so led to less negative ratings of the other group and a reduction in hostility.

Evaluation (Sherif et al.)

Strengths: 

• Field experiment, so very high ecological validity

Limitations:

• Low internal validity (many extraneous variables, like what the boys said to each other or did when not being observed by researchers)

• Reduction in hostility after superordinate goals could have been due to demand characteristics (self-report)

• Ethical concerns about deception, and participants were not protected from harm (some children developed signs of anxiety)

• Theory may be slightly oversimplified since conflict between children isn’t the same as real-life conflicts 

• Sampling bias, so results are difficult to generalize

Aim (McDoom)

To understand the mechanisms of group polarization

Procedure (McDoom)

Case study of the Rwandan genocide including content analysis of Radio Milles Collines broadcasts, surveys, and interviews, focusing on comparing data between the North of Rwanda (hight threat of war) and the South (low threat of war)

Results (McDoom)

Over time, language on the radio broadcasts became much more negative about out-group (the Tutsis). As the threat of war became greater, divisions between the two groups became more salient. Negative language further increased both on the radio and in communities. When the genocide began, all members of the Tutsi out-group were viewed the same, and Hutus were pressured to express loyalty by condoning the genocide. In the South, however, there was minimal conflict because there was not a threat of war.

Evaluation (McDoom)

Strengths: 

• High ecological validity 

• Data triangulation

Limitations:

• Data is analyzed after the event, so you are relying on self-reported data and reconstructive memory

• Sampling bias as many people likely are not comfortable talking about the war, particularly if they were involved

• People might still be loyal to group beliefs instead of their own

Aim (Paluck)

To determine whether media could be used to reduce conflict between the Hutus and Tutsis in Rwanda

Procedure (Paluck)

Participants were randomly allocated to either a control group that listened to a soap opera about reproductive health or the experimental group that listened to a radio soap opera about reconciliation between Hutu and Tutsi characters. Nearly all the participants had been in Rwanda at the start of the genocide. After a year of listening to the soap operas, researchers carried out interviews, focus groups, and observations.

Results (Paluck)

Participants in the experimental group demonstrated more trust of the out-group and were more willing to cooperate with them, even if their beliefs about what happened during the genocide hadn’t changed. Paluck argues that feelings of empathy towards the soap opera characters might have led to the development of empathy for real Rwandans. Using observational learning instead of direct contact may allow people experience intergroup contact in a less threatening way.

Evaluation (Paluck)

Strengths: 

• A stratified sample was used to ensure that the participants were representative of the Rwandan population

• Data triangulation increases validity and reliability

Limitations:

• Demand characteristics, particularly social desirability in interviews