What works to reduce prejudice and discrimination? - A review of the evidence

Section Five: Reflection, Limitations, And Applicability - What Can We Learn?

As pointed out in the introduction, prejudice is primarily a social problem and prejudice-reduction requires social as well as individual change. In attempting to tackle prejudice, we want to challenge and influence societal norms where entrenched prejudice exists. So the approach has to be about more than just the individual. Of course, criminal justice approaches may be appropriate in cases of hate crime, the extreme manifestation of prejudice. There may also be evidence on the deterrent effects of criminal sanctions, though this is beyond the scope of this paper. Returning to the earlier theoretical discussion, this report has focused on the third 'level' of Duckitt's model: "mechanisms of social influence that operate in group and interpersonal interactions". The intention is to learn from some of the key messages from the theories and the empirical interventions outlined in the last two sections.

The fundamental message is that, unfortunately, the evidence on 'what works' is very limited. We can make some general comments on the fact that developing relationships, encouraging close contact, and aiming interventions at young children where possible may be most effective, but Paluck and Green's 2009 large-scale review of studies on prejudice-reduction interventions concludes that "of the hundreds of studies we examine, a small fraction speak convincingly to the questions of whether, why, and under what conditions a given type of intervention works" (339). Moreover, the majority of studies have taken place in laboratories, not in real-life. We must consider who participates in these 'experimental' lab-based studies: these tend to be dominated by Western undergraduate students. So even those studies that show promise in terms of certain prejudice-reduction initiatives have to be treated with caution, as those with perhaps the most entrenched prejudiced attitudes might not be the ones participating in studies. Indeed, this is also highly likely to be the case in real-life interventions. There is also a lack of evidence from field experimental literature; many interventions which show potential in laboratory settings have never been tested. For instance, Aboud et al (2012) comment that:

"Conditions known to enhance the benefits of contact are not always implemented in community and school settings. Furthermore, the rigorous designs used in small- scale laboratory research are rarely found in evaluations of school programs" (309).

Of course it is unrealistic to expect that in the real world interventions can be controlled and evaluated so precisely, so the intention is not to undervalue these important studies. There is value in drawing upon some of the key messages/themes that might help to inform future practice, whilst always keeping in mind that the question of application is crucial.

The rest of the report will reiterate the key messages and questions that emerged from the previous sections, namely: whether targeting specific prejudices or employing a general prejudice-reduction approach works best; practically, what types of activity are considered to work best (including where, with who, and how often); and finally the importance of measurement and evaluation.

Broad prejudice-reduction approach or one targeted at specific types of prejudice?

An important question is whether prejudices should be directly challenged, or whether encouraging positive contact between groups would be more effective. The question of who are victims of and who are perpetrators of prejudice is not always clear. Most of the studies refer to 'minority' versus 'majority' communities, reflecting the fact that prejudice is most often directed towards minority groups who have less power. Sometimes this relates to numerical majorities but often (for instance women, working class people, ethnic groups in some cases) this is not the case. Furthermore, power is not static, and it is important to consider the intersectionality of prejudice: a black working-class woman will inevitably experience processes of exclusion and discrimination differently to a black middle-class man. Abrams et al (2015) discuss the notion of 'equality hypocrisy', whereby people professing values of equality often still tend to differentiate between which groups are 'deserving', because "social identities, power hierarchies, and intergroup norms come into play, all of which might place greater value on some groups than others" (30). Sutton et al (2007) warn of the need to be careful about prejudice-reduction interventions because prejudiced attitudes vary greatly by factors such as age, education level and region. Even within individuals prejudiced attitudes will adapt over time and/or in different circumstances. A popular example of this is the idea of people behaving in what could be deemed a 'racist' or 'sectarian' at a football match, yet would never consider themselves racist or sectarian, and would be unlikely to display similar attitudes or behaviours in other contexts.

This raises the question of whether specific anti-prejudice initiatives (e.g. anti-sectarian, anti-racist, anti-homophobic etc.) would work better than a 'catch-all' broader focus on prejudice, and/or placing in the broader equalities framework. The evidence review undertaken has been unable to find clear guidance on this matter. Abrams (2010) points out that this has not yet been subject to testing, but suggests that "given that prejudices towards different groups appear to have different developmental trajectories, it seems likely that the latter approach (targeting specific prejudices separately) may work better" (76). Yet consideration of the intersectionality of prejudice may lend support to idea of a broader equalities/respect approach, perhaps with flexibility for targeting where appropriate. In their report for Stonewall, Valentine and McDonald (2004) found strong similarities in the levels and types of prejudice across the country. As a result, the authors stressed that policies should be national, but also "regional and locally sensitive" (21). So it appears that quite a delicate balance has to be struck.

Practical suggestions: Key messages about activity

• Setting and participants: Where prejudice-reduction initiatives take place is important. Interventions might function in schools, workplaces, or in the community. You can encourage or even compel people to participate in the first two of these but the third is less obvious - there has to be choice, self-selection is a factor. Given that we have already established that most prejudice takes place in the home, it is important to consider how positive messages might be transmitted to other settings. This also applies to intended participants. Are there target groups that we would want to participate in certain initiatives? How can we reach out to these people? This may also include long-term thinking, such as enabling tolerance among future parents.
• Clear objectives: It should be clear from the outset of an intervention who it is aimed at and what is intended. Clear objectives, a clear structure, and definable outcomes are essential. What behaviours/attitudes is the intervention trying to change? Are there risks of counter-productivity? How will success be measured?
• A range of activities in group settings: In terms of initiatives, group activities work best (except in the case of entrenched/aggressive attitudes, which may mean one-to-one approach is better). Initiatives should prioritise "learning through doing and experiencing, not just listening and talking" (JRF report). Although it should be a given that young people especially are less receptive to 'instruction', many interventions are still based on this. Recreational activities such as sport and arts-based activities are particularly good (especially for young people). It is also worth considering how best to utilise the media in interventions.
• Discussing sensitive topics and personal experiences: Discussing historical and political events has been shown to affect change in attitudes, and JRF review suggests that interventions should focus on 'encouraging reflection' and "on personal attitudes and experiences".
• Importance of facilitators: The case studies continually emphasised the importance of facilitators in the effectiveness of interventions. Correct training, consistency, and regular feedback will help ensure best effectiveness in this respect. Peer engagement is a key aspect of this: involving participants (e.g. youth) in the design, implementation, and review of interventions had positive results.
• Length of interventions: If projects are short (which they often are), it is questionable how much impact they can actually have, and how can they be feasibly evaluated in that time. Unpredictable, short-term funding has been shown to be less likely to work: on-going support/funding required. JRF report states that one-off activities make less impact; better results come from "sustained activities over a period of time".

Evaluation of prejudice- reduction interventions

As Paluck and Green (2009) point out, evaluating changes in attitudes is extremely complex. For example, if someone chooses to attend a diversity course, or voluntarily participates in a 'sectarianism awareness' workshop, it is highly like that they will have "more positive attitudes towards diversity" already (344). Furthermore, Sutton et al (2007) emphasise that "there is a need to implement effective initiatives, rather than those based on idealised notions of what works, or what should work" (30). Evaluation of interventions should be considered at design stage, not as an afterthought. Even when evaluations of projects do take place, there are issues. Measuring prejudice is difficult partly because of social desirability concerns, so measuring improvement in prejudiced attitudes will be particularly difficult. Paluck (2006) reiterates the issue of self-reporting, self-presentation bias, and social desirability bias. Paluck recommends that future research is needed to establish causal effects of interventions, using unobtrusive measurements that go beyond self-reporting (585), as it is vital not to confuse 'feedback' with evidence of impact. Post-event surveys do not provide enough information to determine whether an activity has been effective or not: it may be that participants simply particularly enjoy the programme, feel like they should be supportive and positive, or they might even have increased awareness of bias - but these do not necessarily translate into long-term attitude change. Post-intervention surveys in particular will have to consider these issues, and might best be complemented by qualitative methods such as interviews, focus groups, and observations, and follow-up measurement. The Joseph Rowntree Foundation report suggests that programmes should be assessed for effectiveness by participants themselves, which would support the argument for peer-based learning (e.g. involving previous participants in the running of projects). Sutton et al (2007) include a useful checklist (appendix 1) which may be a useful example for initiatives to help develop their own. Finally, there may be an argument for retrospective analysis in areas where there is an absence of prejudice where you might have expected to see it. This is potentially an area worthy of further research, to determine what factors impact on positive attitudes and tolerance of diversity, as opposed to simply focusing on where prejudice exists and how to get rid of it.