A report by Or Kashti in Haaretz several months ago stated that the Council for Higher Education had approved a new framework for training kindergarten and primary school teachers, which for the first time is to take into account education for shared existence and the prevention of racial prejudice. This is ostensibly a necessary and welcome development in a socially fragmented country like Israel, where we frequently hear about new surveys that reveal the depths of hatred and prejudice that exist between the different communities here. So, in what ways will the teachers be trained? And what are the best methods for reducing prejudice?
In an attempt to answer these questions, we, at the Hebrew University looked at all 418 of the experiments aimed at reducing prejudice that were conducted around the world over the past dozen years, and carried out a narrative review and meta-analysis. The study was done in conjunction with Betsy Levy Paluck, professor of psychology and public affairs, and Chelsey S. Clark, a graduate student, both from Princeton University, and with Donald P. Green, professor of political science from Columbia University. This has been a thriving field of research in recent years, as evidenced both by the number of studies examining the effectiveness of such initiatives, and from the plethora of methods being studied, emerging from a wide range of disciplines, including psychology, political science, economics and public health.
We found that in 2007 there were fewer than five experiments conducted, whereas in 2019 the number had exceeded 50. We encountered a mélange of approaches to reduce prejudice, new and old. Some of the studies examined classic approaches like dialogue groups, joint sports activities or efforts to alter social norms; others involved new approaches, such as use of virtual reality technology that allows a person from one group to step into the shoes of a member of a different group, or attempts to teach people how to control and regulate their emotions and thereby to enhance empathy or lessen anger and hatred toward members of other groups.
Our joint study, “Prejudice Reduction: Progress and Challenges,” published this past January in the journal Annual Review of Psychology, thoroughly reviewed every experimental study conducted in the past decade in the field. We coded the empirical findings – asking, for example, whether an intervention successfully reduced prejudice – and conducted a meta-analysis to determine the average impact of all the initiatives. A dozen years’ worth of rigorous research shows these types of prejudice-reduction programs have little effect.
Imagine a Jew who harbors some resentment or dislike with regard to Arabs. If we ask them to rank their feelings toward Arabs on a scale ranging from zero (very negative, cold feelings to 50 (neutral) to 100 (very positive, warm feelings), they might say 40. Most of the initiatives we examined would have shifted him or her somewhat in the desired direction. On average, the prejudice-reduction programming we reviewed would move this person from a 40 to a 48 – almost, but not quite neutral.
That may look like an encouraging finding, as it’s a substantive shift. However, when we limited ourselves only to the higher-quality studies in our dataset, the effect was cut in half, from 40 to 44 – a negligible change in viewpoint. The more reliable the results of a study were, the less encouraging its results were, with no standout success for any particular form of intervention.
- This Jewish-Arab school embodies an Israeli utopia. Its Arab principal reveals what's beneath the surface
- When Palestinians are represented positively in history lessons, Israeli students' trust in them grows
- 'There's an undercurrent on the American left that regards Jews as suspect'
One study examined an hour-long online diversity training session administered by a global company that employs more than 10,000 people. Training of this sort is the most widespread practice in the United States to reduce prejudice and to encourage inclusion of minority group members. The great majority of Fortune 500 companies offer diversity training to their employees. The researchers found a small change for the better in the attitudes of those who participated in the program, but discovered that it had no effect on the employees’ actual behavior.
For example, about three weeks after the training, the company encouraged its workers to take part in a mentoring program in which senior employees were asked to tutor junior. The researchers’ expectation was that those who had undergone the diversity training would be more willing to mentor women and racial minorities than those who had not. To their surprise, they discovered that despite the minor improvement in declared attitudes, they were not translated into practical behavior – nor in the willingness to participate in the mentoring program. And whereas the training did nothing to better the behavior of white males, it did have a slightly positive effect on the behavior of women and minorities.
In another study, which was featured on the cover of the prestigious journal Science, Salma Mousa, a political scientist at Stanford, examined another common practice in conflict zones: mixed soccer teams. In 2020 alone, the U.S. Agency for International Development allocated more than $870 million for projects of this kind, to promote “social cohesion” worldwide. Dr. Mousa examined attitudes and behaviors of Christian Iraqi refugees who had fled from ISIS, but remained in Iraq. The players were randomly assigned to train and play for a two-month period on either all-Christian teams or mixed Christian-Muslim teams. Mousa’s study found that having Christians and Muslims train and play together did not affect Christian players’ attitudes toward Muslims, though it did have a small positive effect on their willingness to continue playing together after the league had ended. However, an examination of these positive influences beyond the playing field – for example, whether Christian players playing on a team with Muslims would eat in a restaurant in a Muslim area, or take part in a social event that included Muslims as well – did not reveal a more generalized shift, despite the positive change in behaviors toward teammates.
We tend to treat prejudice as a disease that is responsible for many disparities in Israel. As such, we must treat anti-racism programs with the same rigor used when testing drugs that treat physical and mental illness.
A considerable portion of the studies we examined were conducted in Israel, many of them from the field of psychology. We were surprised to discover that despite the multiplicity of programs and approaches meant to reduce prejudice and promote coexistence, the majority of the research done here was conducted under laboratory conditions – of the sort that examines ideas in their preliminary stages before they are scaled up to a full-blown program that can be implemented in the field – and that most of the participants were either students or online workers who received payment or course credit for their participation. Thus, even though many initiatives to promote shared life can be found in Israel, they are often not subjected to scientific evaluation that can gauge their effectiveness.
Attitudes and behavior
It’s possible to understand the hesitation of Israeli organizations when it comes to cooperating with researchers performing experiments. Significant effort on the part of these organizations is necessary, and they usually have to assign some of their potential participants to a control group, in which those individuals do not immediately receive what’s known in the professional jargon as the “treatment.” Organizations engaged in this realm generally feel that it’s important to reach as many participants as possible, so they are reluctant to subject potential participants to control conditions, whereby treatment will be delayed for the sake of research.
On the other hand, an examination of the practices currently being employed by the organizations to evaluate the effectiveness of their programs is troubling. The absolute majority of these organizations customarily make do with distributing questionnaires that measure attitudes before and after participation in a particular program – and attribute any attitude changes to that activity.
A study reported recently in Haaretz (“When Palestinians are represented positively in history lessons, Israeli students’ trust in them grows”; July 3) involved exposing Jewish high-schoolers to positive and negative representations of Palestinians and measuring their attitudes before and after that exposure. They found that, as opposed to negative representations, exposure to positive portrayals heightened the students’ trust in Palestinians and support for a peace agreement with them. However, the study as carried out did not allow a definitive conclusion to be reached about whether this activity, which involved their materials read in history classes, was the cause of changes in attitudes observed among the participants, as the findings could have been influenced by external factors that occurred in that period.
To better understand why a study design like that cannot provide definitive evidence regarding a program’s effectiveness, let’s take a hypothetical example of a two-month initiative aimed at reducing prejudice by means of Jewish-Arab encounters. Let’s assume the original assessment of participants’ views took place at the beginning of April this year, and the second one at the end of May. If we observed a negative shift in one side’s attitudes toward the other – can we say that the initiative failed? Or could it be that the hostilities in the Gaza Strip in May and the simultaneous intercommunal tensions in Israel precipitated the changes found among the participants?
The accepted scientific means for gauging the effectiveness of any treatment – vaccination, medication or educational endeavor – is to conduct an experiment where participants are randomly assigned to receive the “treatment” or not (the control group). Without random assignment to treatment and control conditions, it cannot be said with certainty that a medication, or an initiative or program, is effective, nor is it possible to rule out alternative explanations such as, in our hypothetical case, the state of relations within Israel between Jews and Arabs.
The paucity of organizations in the field that apply proper scientific methods to examining the effectiveness of their programs is a worldwide problem, and does not characterize the situation in Israel alone. We were surprised to discover that it is precisely the most commonly employed programs (like diversity training or integrated team sports) that are rarely tested experimentally; we found only six studies in the past decade, for example, that experimentally examined the impact of diversity-training programs.
But beyond the surprising dearth of experimental research assessing programs in actual use, we also encountered other research practices that should concern anyone who holds this subject dear. First and foremost is the absence of research focusing on actual behavioral phenomena as distinguished from self-reported attitudes and emotions that serve as frequent measures for determining a program’s success. The focus on attitudes and emotions – i.e., how much empathy I feel toward Arabs, the degree to which I would be ready to live in a building with an Arab neighbor – is problematic at the very least, because the reduction of prejudiced and discriminatory behaviors is no less important.
Fewer than 10 percent of the studies looked at concrete behavior, and the few that examined both that and attitudes found that they were not always correlated. In other words, a change in one’s views and feelings did not always translate into a behavioral change, and vice versa. A good example of this is the study involving the Iraqi soccer players, whose willingness to train with Muslim players was not accompanied by a sympathetic attitude toward Muslims in general.
Instead of our children learning about prejudice from Jewish teachers, they could encounter an Arab principal in their school and change their negative stereotype.
Also of concern is the fact that only a very small proportion of the over 400 studies examined the long-term effects of various interventions. While a form of measurement is generally carried out immediately upon the conclusion of the activity or initiative in question, no tests are conducted to detect whether the effect survives after a few days, weeks or months.
Reformatting the system
How can we reconcile the paradox by which research about reducing racism is flourishing, yet their findings are meager? A possible explanation lies in what we found to be the most widespread approach to reducing prejudice. We found that more than 70 percent of the initiatives sought to influence participants by training them to think differently, feel differently, perceive the other side differently.
It makes for a logical thesis: Prejudice is a psychological problem rooted in individuals’ minds, hence its treatment is perceived as something that should attack that problem using individualized mental approaches. If the problem is that Jews hate Arabs, the proposed treatment is to teach individual Jews to control and regulate that emotion. The assumption is that if we succeed in influencing the attitudes and emotions of individual people and render them less prejudiced, this will lead to greater changes in the social institutions and organizations in which they operate.
But our findings, as well as the worrisome data that arise from various surveys being conducted in Israel and elsewhere, suggest that this approach is failing us. We can’t train prejudice away. Moving forward, our first recommendation is directed at researchers and activists working in the realm of prejudice reduction in Israel. We tend to treat prejudice as a disease that is responsible for many of the economic, health care-related and educational disparities in this country. As such, we must start to examine programs and activities that have been proposed for treating it with the same seriousness and rigor used when testing drugs for treating physical and mental diseases.
The best scientific approach is to examine these initiatives through randomized controlled trials, which are considered the gold standard in the scientific community. Instead of the current common practice, in which questionnaires are distributed before a particular treatment and at its conclusion, participants in an RCT must be assigned randomly, so that they will either receive the proposed treatment or will be subject to control conditions in which the treatment is postponed until after the experiment’s conclusion. Only thus will we able to deduce the effectiveness of a proposed initiative.
The second proposal is bolder and more radical, and is directed at decision makers and individuals in key positions in large public organizations, the media or the private sector: a change of approach in an attempt to reduce prejudice. Instead of trying to influence the attitudes and feelings of individuals, in the hope that they will influence the system within which they operate, we propose focusing on changing the system. The point of departure of this approach lies in classic studies that show that human social behavior is influenced by one’s immediate environment, and not just one’s cognition. If we want to influence people, the most effective way is likely to be by changing the environment in which they operate. Initiatives of this sort will not aim to affect the attitudes of individuals per se, but the way organizations work, and the social norms and the shared experiences of the members of the organization.
Let’s take the example of the Council for Higher Education’s proposal, mentioned above, to promote teacher training that will deal with the phenomenon of prejudice, in the hope that within a number of years we will see a decline in indices of such bias among pupils. A systemic approach will aspire to bring about a broader institutional change in the education system, which within a few years will be manifested in the form of less prejudiced attitudes.
This can be realized, for example, by integrating teachers from minority groups in senior management of administrative positions in local schools. Instead of our children learning about prejudice from Jewish teachers who were trained to deal with this subject, they will encounter an Arab principal in their school and will, perhaps that way, change their negative stereotype regarding this minority group. In the world of medicine, for example, rather than training doctors to be less biased — an activity with minimal impact if any — we could offer them the use of an algorithmic aid that assists health care providers in making diagnosis and decisions that are not biased toward minority populations.
Approaches like these are studied far less in the realm of reducing prejudice, but the little evidence that has accumulated gives rise to hope. A breakthrough study conducted in India found that a policy of affirmative action for women in communal political leadership roles led within a few years to less bias toward women. In Israel, a study conducted by Chagai Weiss, a doctoral student at the University of Wisconsin, Madison, showed that Jews who received medical treatment from Arab physicians expressed less prejudiced attitudes after receiving the treatment.
Because there is a dearth of research that examines such institutional and social approaches, we can’t say with certainty that they would lead to the much-desired change. But our evidence supports a mandate for scientists and practitioners alike to be more ambitious and test new approaches. An extensive and comprehensive change such as the one we are suggesting is challenging and can be expected to stir opposition from various quarters. But precisely now, when prejudice in Israel is on the rise, it is possible to hope that the new “government of change” will also attack that phenomenon systemically and institutionally.
Dr. Roni Porat is a behavioral scientist who teaches in the departments of political science and international relations at the Hebrew University of Jerusalem.