A Study on Imaged Contact as an Effective Method for Prejudice Reduction

Imagined contact has been shown to be an effective method of prejudice reduction when no other option is available. Crips, Turner, and Lambert (2007) previously studied the effects of imagined contact with outgroup members on prejudice towards that outgroup. This study replicates that of Crisp and colleagues with regards to outgroup anxiety towards Muslims. We had our participants imagine first contact with a Muslim stranger or neutral undescribed stranger. Participants then reported their comfort level working with various demographic groups, including Muslims. The data did not show any significant reduction in the experimental condition, therefore not supporting the findings of Crisp and colleagues (2007). However, some moderating factors—such as political orientation or overall anxiety levels — do have significant effects on comfort levels with Muslims.

Prejudice Reduction Through Imagined Contact

Introduction

The United States is divisive right now on numerous issues; prejudice and discrimination are at the heart of many of these political and social debates. While many people are interested in extended discussion of these issues, few suggest genuine progress. However, one tool to decrease prejudice between groups-especially with an outgroup — is contact between groups. One extension of contact—imagined contact where participants imagine interacting with a member of another group—is more convenient. Imagined contact has the potential to bridge gaps of knowledge and prejudice reduction, but only if it is a truly effective and genuine phenomenon.

Literature Review

Contact between members of different groups should lessen intergroup hostilities through promoting positive intergroup attitudes, but only under certain conditions (Allport, 1954). According to Allport, these conditions include equal status and cooperation for common goals, and most importantly, a positive interaction (1954). There are two mechanisms of intergroup contact that could reduce prejudice under the right circumstances.

The first mechanism is the reduction of intergroup anxiety. Intergroup anxiety reduces the desirability to initiate intergroup contact in the first place, increasing prejudice (Stephan & Stephan 1985). Intergroup anxiety results from negative expectations, but not necessarily previous experiences, of rejection or discrimination (Stephan & Stephan, 1985). According to Stephan and Stephan, successful group contact as outlined by the conditions Allport set forth (1954) should cause the groups to realize that they have no reason to fear each other, reducing intergroup anxiety and prejudice.

The second mechanism is reduction through changing affections and cognitive consequences: perceptions, stereotypes, and judgements about a group (Crisp, Turner, & Lambert, 2007). Affecting these factors reduces prejudice by counteracting the “outgroup homogeneity effect” (Quattrone and Jones, 1980). This effect is observed when people perceive outgroup members to be more homogenous than ingroup members (Quattrone and Jones, 1980).

When there is a successful contact, people are more likely to see members of other groups as individuals, rather than assigning group characteristics (Crisp et al., 2007). However, there would be less of an effect if previous group characteristics were positive, or if the contact was negative (Crisp et al., 2007). Reducing prejudice through increasing variability of outgroup trait descriptions has been found to reduce tensions between Hindus and Muslims in Bangladesh (Islam & Hewstone, 1993), which is relevant to this study’s focus on Muslims as an outgroup.

However, real contact may not always be feasible as a tool to reduce prejudice between groups. Crips and colleagues theorized that imagining contact, rather than direct contact, may also reduce prejudice between groups (2007). Crisp, Stathi, Turner, and Husnu also noted that imagined contact had four key strengths over Allport’s direct contact: it can be used in situations when actual or extended contact isn’t possible, it is a cheap and accessible treatment, it is safe, and it may make people more open to positive interactions in the future (2009). Imagining contact reduces prejudice through the two mechanisms described above by activating the mind’s associated knowledge structures (Crisp et al., 2007). If contact is successful, concepts associated with that successful interaction, such as comfort — an indication of reduced intergroup anxiety—and a greater variance of traits—changing perceptions, stereotypes, and judgements about a group—should be activated (Crisp et al., 2007). When thinking about the outgroup, the activated concepts should cause people to refer back to any positive interactions, and would influence perceptions of outgroups.

Bargh and colleagues believe thinking about social categories may generate negative intergroup attitudes and behavior, especially when thinking about an outgroup, and may prime negative associated knowledge structures (1996). This finding may point to an important caveat in this study. However, Crisp and colleagues believe that thinking about the interaction itself rather than an outgroup category and placing thoughts in the context of the interaction — how they behave towards the interaction partner and vice versa — would counter the priming of negative attitudes (2007).

Crisp and colleagues ran a study in which undergraduate participants were instructed either to imagine an outdoors scene (the control group) or contact with an elderly person (the experimental group) (2007). The participants in the experimental condition were asked to list different ways in which they could classify the imagined elderly person after an imagined conversation (Crisp et al., 2007). After this imagined contact, all participants were told that the researchers may be conducting a study in the near future with a local old people’s home, and asked to rate their preference working with a young person and an elderly person on a 1-9 Likert-type scale (1 = not at all, 9 = very much) in order to assess the suitability of pairings in the researchers’ follow-up study (Crisp et al., 2007). Crisp and colleagues found that while there was significant intergroup bias against the outgroup (old people)in the control group, the participants in the experimental group did not show significant difference between ingroup (young people) and outgroup (old people) evaluations on working preferences (2007). Crips and colleagues also found similar effects when participants were instructed to imagine homosexuals as an outgroup; intergroup anxiety was reduced and variability of perceptions increased (2007).

Further research investigating imagined contact showed that it is most effective when mental representation of the contact scenario was elaborated, and/or when the partner for the outgroup was typical of the outgroup rather than atypical (Asbrock, Gutenbrunner, & Wagner, 2013). Asbrock and colleagues (2013) were interested in how this might apply to people with different personality traits, such as those who endorse right-wing authoritarianism (RWA) and social dominance orientation (SDO). Researchers concluded that those high in RWA indicated lower levels of negative emotions toward Turkish people, and greater willingness to engage in future contact with Romani people, both after imagined contact (Asbrock, Gutenbrunner, & Wagner, 2013). While those high in SDO showed less negative emotions after imagined contact toward Turkish people, they did not feel more willingness to engage in social contact with Romani people (Asbrock, Gutenbrunner, & Wagner, 2013). This finding sheds light on a significant critique of imagined contact, which is the possibility that for many participants, although it decreases self-reported negative emotions towards outgroups, it may not change behaviors.

This study is designed to examine the reliability of these findings (Crisp et al., 2007) when applied to college students imagining interaction with people of Muslim faith. After reviewing current literature, the outgroup targeted for prejudice reduction through imagined contact was Muslims; this group has in recent years been the subject of discrimination. This study hypothesizes that imagined contact with a Muslim, in comparison with imagining a neutral contact, will decrease prejudice measured by comfort levels working with a Muslim, as operationalized by Crisp and colleagues (2007).

Methods

This study is a conceptual replication of Crisp, Turner, and Lambert (2007)’s first experiment. The study is a conceptual replication rather than a direct one due to time constraints, demographic differences between Claremont and the subject pool of the original experiment, and different interest in the group being measured for bias against.

Participants

The participant pool was American college students, specifically students attending one of the Claremont Colleges. We recruited through social media outreach to gather participants to take a Qualtrics survey, targeting social media circles of Claremont Colleges students. Of the 165 participants who took the survey, 48 participants finished. Three participants were dropped from data analysis, as further explained in the results section, resulting in 45 observations going through data analysis.

Materials

The survey platform was Qualtrics, making the survey most efficient and easily disseminated to our participant pool.

The study was a between-subjects design. In the experimental condition, participants were asked to imagine contact with a Muslim. In the control condition, participants were asked to imagined contact with an undescribed stranger. This scenario is a deviation from the original study as Crisp, and colleagues (2007) asked participants to imagine an “outdoor scene” as the control. The study used an undescribed stranger so that we could control for the effect of just imagining contact with any person, instead of with a specific outgroup. According to the theory of availability heuristics, the stranger imagined should be the most typical member—and therefore a member of the ingroup—as opposed to imagining a Muslim (Kahneman and Tversky, 1986), making a salient contrast between our experimental group and control group. Thus, the independent variable of interest is what group the participants were in.

The dependent variable was the participant’s reported prejudice against Muslims. Prejudice was operationalized through measured self-reported comfort level with working with Muslims, one of the scales used in Crips and colleague’s original study (2007), except with the outgroup changed from elderly people to a Muslim. The scale was also shortened from a 9-point to a standard 7-point Likert scale to make choosing a comfort level easier for participants.

After the participants signed the consent form, participants were asked filler questions, mainly about demographics such as age. We created two sets of instructions, designed to either invoke participants’ imagination of a detailed interaction with an outgroup member, or their imagination of an interaction with an undescribed stranger. Participants assigned to the imagined contact condition were asked the following question: “We would like you to take a minute to imagine yourself meeting a Muslim for the first time. Imagine their appearance, the conversation that follows and, from what you learn, all the different ways you could classify them into different groups of people.” Participants assigned to the control condition were asked: “We would like you to take a minute to imagine yourself meeting a stranger for the first time. Imagine their appearance, the conversation that follows and, from what you learn, all the different ways you could classify them into different groups of people.” Participants were also instructed to “List the different ways in which you could classify the stranger following the conversation you just imagined”. In both conditions, participants were given space to write as they imagined the scene. A minimum-character validation was put in place to ensure that participants were actually participating in the task. This task was designed to reinforce the effect of the imagining an interaction.

Following this manipulation, participants completed a measure to assess their level of intergroup bias. They were told the researchers “may be carrying out a study shortly in conjunction with a local mosque where we will get either non-Muslims to converse with other non-Muslims, Muslims to converse with other Muslims, or non-Muslims to converse with Muslims. We are gauging whether people would be willing to take part in such a “conversation” study (it would be carried out in the Psychology department, would last 20 minutes, and you would receive $5). If you were to sign-up for this study, can you rate your comfort level for each of the following pairings. Keep in mind this is NOT a commitment to participate. We just would like to gauge people’s interest. All answers will be confidential. Please answer openly and truthfully.”

Participants were asked to mark their preference for working with another non-Muslim (‘Non-Non’ pairing) and then their preference for working with a Muslim (‘Muslim-Non’ pairing) on a 7-point scale (1 = not at all, 7 = very much). An ‘Muslim-Muslim’ pairing rating scale was included to reinforce the impression that Muslim people were also being asked the same questions in preparation for this supposed future study. These pertinent questions were asked along with filler questions concerning comfortability working with Christians, white people, Asian people, and those of the opposite gender. After completing the dependent measure, participants were asked what they thought the aim of the study was, to determine whether they were suspicious that the pre-test was intended to measure for demand characteristics. In the final portion of the survey, participants provided demographic information (see Appendix A-G).

Results

Data Set

Out of the 165 participants recorded on Qualtrics, 48 participants completed all questions. One participant who took fewer than 5 seconds imagining contact was dropped from the initial data set (threshold was 5 seconds imagining contact and 10 seconds answering questions about comfort working with members of different groups). This threshold was chosen because if the study demanded that participants actually took a whole minute as instructed in the survey to imagine a Muslim or a stranger, most participants would be dropped from data analysis. From initial summary statistics of time taken, five seconds was chosen as a threshold that allowed enough time to imagine a contact scenario.

When processing the data set, one participant who correctly guessed the hypothesis was dropped, and also dropped one participant who identified as Muslim, as the interest is in whether imagining contact with Muslims would decrease prejudice for non-Muslims, not for Muslims, as they are already part of the targeted outgroup.

Descriptive Statistics

For descriptive statistics, around 55% of the participants were randomly assigned to imagine interacting with a Muslim. Across both experimental groups, comfort level when working with a Muslim on a 7-point Likert-type scale—with higher numerical values indicating more comfort—yield M = 4.91, SD = 2.24. When participants imagined interacting with a Muslim, comfort level had M = 4.56, SD = 2.33. When participants imagined interacting with a stranger, comfort levels had M = 5.35, SD = 2.11. Although it appears imagining a Muslim actually decreases comfort levels, as the statistical inference shows, that are not statistically significant (see Appendix H).

Statistical Inference

For statistical inference, the data was analyzed with multivariable regressions. Crisp and colleagues (2007) used a similar regression method to analyze data from one of their experiments. Prejudice, operationalized as comfort level of non-Muslims when working with Muslims was measured on a 1 to 7 Likert type scale, with a higher score indicating more comfort (comfort); this variable is also our explained variable. Our main explanatory variable of interest is the dummy stranger_muslim (1 = imagined muslim, 0 = imagined stranger). Because the value of stranger_muslim is randomized between participants, no omitted variables can bias our estimated effect of imagining a stranger versus imagining a Muslim. However, two additional variables were held constant to obtain the most accurate effect of imagining a stranger or a Muslim: ease of comfort working with non-Muslims (nonmuslimwork) and political views (politics). Ease of comfort working with non-Muslims was chosen because it could affect comfort; if someone is naturally uncomfortable working with anyone, regardless of religion, then he/she would be expected to have a difficult time working with a Muslim due to preferences unrelated to religion. Politics was held constant because politically conservative people are more likely to display decreased comfort in the presence of outgroup members (insert citation); the effect of muslim_stranger is best predicted when politics is held constant. After checking for correct functional form, and possible errors in variance with a Breusch-Pagan test, the final regression function is as follows (β0 is the intercept and ui. is the error term): comforti.= β0 + β1muslim_strangeri. + β2nonmuslimworki. + β3politicsi. + ui.

This regression yielded β1 = -0.446, p = 0.43. This result indicates that when participants imagined interacting with a Muslim, comfort levels decreased, on a 7-point scale, by 0.446. However, because this result is far from statistical significance, we cannot draw any conclusions (see Appendix I).

The results of a similar regression, with politics omitted, yielded β1 = -0.188, p = 0.764, were also non-conclusive (see Appendix I).

Discussion

The results did not support the hypothesis that imagined contact with outgroups would decrease prejudice.

The study’s limitations could have contributed to the results not supporting the original hypothesis, especially when the survey asked participants to imagine interaction with a stranger or a It is possible that the imagination conditions were not powerful enough. This study was conducted online; researchers could not ensure that participants were paying absolute attention to the imagination instructions, even if they spent time imagining the interaction. Additionally, the instructions did not specify a positive interaction; West and Greenland found that positive, promotion-based interaction decreased prejudice towards an outgroup, while imagining negative interactions may actually increase prejudice (2016).

Imagined contact has a greater likelihood of decreasing prejudiced when participants are prompted to imagine that their outgroup person of contact has positive traits (Quattrone & Jones 1980). When the prompt is left more open, participants are more likely to imagine negative traits or encounters. Future studies should continue to explore the possibility that imagining positive contact will decrease prejudice, specifically concerning Muslim people as the outgroup.

The phenomenon of imagined contact is less powerful when participants have high levels of bias towards outgroups (Quattrone & Jones 1980). It is possible that the political and media climate of today’s world has made the population as a whole more bias against Muslim people. If that is the case, an imagined contact intervention towards this outgroup was more likely to be ineffective.

As imagination of outgroups is supposed to increase projection of positive self traits to the target group (Stathi & Crisp, 2008)—projection of positive self traits usually is more common with in-group individuals—the imagined Muslim should be projected with more positive self traits of the imaginer. However, when participants imagined negative interactions, they likely projected their own negative traits onto the outgroup.

While imagined contact may actually increase intergroup anxiety if participants were asked to imagine the targeted in a negative light, we are asking participants to imagine Muslims in a positive light, anxiety should be reduced (Turner, Crisp, & Lambert, 2007).

Although controlling for demand characteristics may be more challenging given the medium of our experiment (online surveys), Crisp, Turner, and Lambert (2007) found that very few participants were able to guess the hypothesis of the experiment, and implicit tests conducted by Crisp and Stathi (2008) have confirmed that for this type of experiment, demand characteristics have minimal effects.

There are also statistical explanations for the inconclusiveness of this study’s result. The Cronbach’s alpha for our dependent variable was only 0.20, indicating low validity (see Appendix J).

Future studies should address the limitation that our study suffered. Many participants would had negative contacts when imagining a Muslim, as described from their responses. Imagining contact might make prejudice higher if negative traits are reflected and if interaction is negative (Quattrone & Jones, 1980). Future studies could take into account the work of West & Greenland (2016). They hypothesized that imagined contact would have different effects depending on whether participants were placed in prevention or promotion focused conditions. Prevention focused conditions meant researchers asked participants questions about social desirability in regards to their prejudice. Promotion focused conditions focused on the motivation to achieve a desired and positive goal. Researchers found that when the goal of participants was to not come off as prejudiced (prevention-focused) as opposed to coming off as friendly or kind (promotion-focused), imagined contact actually increased anxiety towards the outgroup. This means that when applying imagined contact as a prejudice reduction technique, both the stated purpose of the exercise as well as the intrinsic motivation of the individual is important to define and can influence the efficacy of the intervention. This study did not prime our participants with promotion or prevention but this effect could prevent the negative interactions that many participants in this study were imagining.

The operationalization of the dependent variable should also be altered to achieve a higher validity as measured by Cronbach’s alpha.

The subject pool might have been another limitation as it is not representative of the target audience for this technique of imagining contact. Imagined scenarios should be used when better methods of prejudice reduction (i.e. direct contact) are not possible. Claremont Colleges are in a relatively diverse environment at their institution. Additional contact through imagination may spark negative contact that would not be normally present.

Additionally, the topic of this study is extremely politically charged at this moment. It could be that the stereotypes associated with Muslims and the motivations behind Islamophobia are so visceral and so deeply ingrained that imagined contact was not strong enough to counteract.