Synopsis

How can academics who provide policy prescriptions and policymakers who implement policy determine whether a given strategy works? Controlled Randomised Trials are a useful way to identify racial and religious discrimination, to evaluate policy, and to bridge academic research and policy outcomes.

Commentary

THE SINGAPORE Government has identified social cohesion as an important policy pillar to support the goal of nation-building. This has arisen out of awareness that societal differences in Singapore are an issue that needs to be managed with care. Whether it concerns race, religion, or nationality, these differences are perceived to have the potential to create or increase tensions between different groups, destabilise society and even result in violent conflict.

Besides the ethnic-religious riots in 1964, the Jemaah Islamiah terrorist arrests in 2001, the Amy Cheong rant against Malay weddings or the illegal bus strike last year, were events which tested ethnic, religious or national ties.

What is social cohesion?

 Social cohesion is defined by scholars as a situation where Singaporeans believe they belong to society, have a stake in it and are involved in its development. High levels of cohesion involve strong social bonds, trust in institutions, and norms that promote reciprocity. Towards this end, the government has pursued a multi-pronged strategy to manage racial and religious harmony and relations. They include the management of society through the Ethnic Integration Policy (EIP) in public housing; National Service (NS) for male Singaporeans; self-help groups for the respective ethnic communities; celebration of Racial Harmony Day; and the creation of institutions such as the Inter-Racial and Religious Confidence Circle (IRCC).

These government policies have, in sum, worked in building social cohesion in Singapore. But which policy worked? Was it the management of society along the ethnic policy categorisations of Chinese, Malay, Indian, Other (CMIO); EIP for public housing; or was it a combination of bothCMIO and EIP? What precise effect does NS have on social cohesion in Singapore today? Or did all the above policies serve to achieve the level of cohesion we observe today? While something worked we cannot precisely identify what has worked.

How then can scholars who provide policy prescriptions and policymakers who implement policy determine whether a given strategy works? How can we know whether a given policy has had the desired effect on people’s attitudes and behaviour? To know whether a policy has worked, we need to imagine what academics call a counterfactual: a world where this particular policy was not implemented but everything else remains the same.

For example, consider a Singapore where EIP was not implemented but everything else stayed the same. Depending on whether we imagine a Singapore with ethnic ghettos and skewed housing prices, or a Singapore where most things remain the same, we would have a different estimation of the effectiveness of this particular policy. Indeed, most debates over the effectiveness of a policy are basically debates about counterfactuals.

Controlled Randomised Trials

Controlled Randomised Trials (CRTs) are a useful way to identify racial and religious discrimination, to evaluate policy, and to bridge academic research and policy outcomes. The roots of CRTs lie in the medical sciences where human trials are mandatory. For example, consider a new drug that claims to cure fever. The drug company is required to conduct a CRT where they would randomly allocate people to two groups: a Treatment group which receives the new drug, and a Control group which does not receive the new drug. Any difference in measures of fever between the two groups is attributable only to the new drug. Replace ‘drug’ with ‘policy’, and replace ‘cure fever’ with ‘change attitudes/behaviour’ in the above example and we can see the usefulness of CRTs to evaluate policy. Its main advantage is that the Control group described above is the true counterfactual.

CRTs have been used to identify areas where discrimination could take place along lines of race and religion. For instance, two US scholars conducted a study in 2004 to identify racial discrimination in the labour market. They sent out resumes in response to help-wanted advertisements in Chicago and Boston newspapers and measured the call-back rate for an interview for each resume. They used a fictitious job applicant and randomly manipulated the name with White American names like ‘Emily Walsh’ or ‘Greg Baker’, and African-American sounding names like ‘Lakisha Washington’ and ‘Jamal Jones’.

They found that African Americans had to send, on average, 50 percent more resumes than White Americans for an interview callback. This study was a CRT since they randomly manipulated the name of the job seeker but ensured that all other aspects of the resumes were identical. This allowed them to conclude that the 50 percent difference can be precisely attributed to racial discrimination and not to any other extraneous factor.

Possible CRTs in Singapore

In Singapore, we could conduct similar studies to identify potential discrimination along race, religion, age, gender and nationality in the labour market. Ideally we would want to eliminate any systematic discrimination along these lines since it can affect social cohesion, and CRTs can help detect such cases in the job market. For instance, CRTs could identify instances where employers may favour people of certain races and religion, or perhaps prefer younger to older employees, or perhaps choose a foreigner over a Singaporean.

It would involve performing a study similar to the one described above – keeping everything the same in the resume but randomly manipulating the race, religion, age, gender or nationality of the respondent. In a similar vein, this could also be expanded to areas such as the food and rental markets. For example, a CRT could help identify if some food stalls overcharge tourists or some house owners discriminate against tenants from a certain country of origin.

Once a problem is identified, CRTs could also assist in evaluating policy. For instance, CRTs can be used to identify the effect of inter-racial/religious groups in schools and workplace, building social space with foreigners, or tweaking EIP to include foreigners. We can also envision evaluating programmes like National Education and the Speak English/Mandarin campaigns. In some instances, we may have formulated the right policy but the mode of public communication could dilute its effect.

CRTs can help identify the right type of communication strategy that will deliver the desired change in attitudes and behaviour of the mass public. Further, CRTs can also be used to evaluate the precise effect of social media on shaping the attitudes of Singaporeans, especially younger minds. Each CRT listed above would involve randomly allocating individuals to treatment and control groups, and ensuring that only the treatment group gets the policy that we wish to evaluate.

In sum, CRTs provide us with a way to evaluate policy before it is implemented in the broader society. If we have a rule in place that no new drug will be disseminated to the public without clear control randomised trials to show that it works, why should it be any different for public policy?

About the Author

S.P.Harish is a PhD Candidate in the Department of Politics at New York University. He is currently a Visiting Research Associate at the Centre of Excellence for National Security (CENS), a constituent unit of the S. Rajaratnam School of International Studies (RSIS), Nanyang Technological University.