A better way to study (and possibly address) discrimination

We often read reports about studies on the under-representation of groups such as women and ethnic minorities in the top echelons of organisations, including private firms and government.

In many cases these research findings imply that the under-representation is due to discrimination, and go on to suggest remedial measures aimed at increasing the numbers of these ‘minorities’ in the upper ranks of organisations, including boards of directors and senior executives.

To be sure, the underlying premise of these reports is valid. There is indeed much to be gained by having diversity at every level of an organisation, especially at the top.

After all, the implications of each decision tend to be heavier as one goes up in a hierarchy, and since the consequences of an incorrect decision will be greater at the top echelons, it is important that organisations get these decisions right—through a healthy debate within a diverse team, for instance.

However, there is reason to be concerned about the rigour of many of these studies, evident in measures such as percentage representation of minorities, which do not even begin to inform us about the extent of the problem.

Leap of faith

Consider a recent UN report which calculated the proportion of women parliamentarians across different countries. The “leader” in this regard was – hold your breath – Rwanda, where women held 63.8 per cent of seats in the lower house.

stop280It may be a leap of faith to conclude based on this statistic that women enjoy very good status and equal opportunities in Rwandan society and politics. In fact, an article by Alexandra Topping in The Guardian suggested that a large proportion of Rwandan women are subject to physical and sexual violence, most often from their partners.

Similar issues would come up if one were to count the number of heads of government across different countries. Many countries in the Indian sub-continent including Bangladesh, India, Pakistan, and Sri Lanka, would figure close to the top on this dimension.

In fact, Sri Lanka had a female prime minister (Sirimavo Bandarnaike) for ten of its first 29 years of independent existence and India for 11 of its first 33 years (Indira Gandhi). The USA is yet to have its first female president although Barack Obama became the first person of African descent to become a president – after 240 years of independence.

Despite the many female heads of government in the Indian subcontinent, a 2009 study by Beaman and co-authors in the Quarterly Journal of Economics noted that Indian voters perceived women leaders to be less effective suggesting that rise of women leaders is not because of positive broad-based attitudes towards women in these countries.

In statistical analysis, we would say that omitted variables are one of the key reasons behind these apparently anomalous results. While we are not aware of the political scenario in Rwanda, we can confidently say that the incidence of women leaders in the Indian sub-continent had a lot to do with membership of powerful political families – an omitted variable when one looks at simple and aggregate statistics such as proportion of female MPs or prime ministers.

Another report in Forbes based on the Grant Thornton Annual Survey released in March 2014 noted that Indonesia, the Philippines and Thailand all have high numbers of women in senior roles (41%, 40% and 38%, respectively), in fact much higher than 21% for the seven industrialised countries. The report suggested that these percentages may have been helped by the tendency for families to live together or nearby, offering women free, built-in childcare.

To stop a recurring wound, we need to understand its causes and what the effective way to treat it is

We believe, however, that the other factor pointed out by the report, high number of family owned and run businesses in these countries may be equally, if not more, important.

Prescriptions – implicit or explicit – made by some of these studies have been even more problematic.

Let us consider, for instance, the suggestion that we should appoint more people from under-represented groups to the top echelons of firms or to top political positions, so as to improve ‘minority’ representation at the top.

Since the reality is that only a very small proportion of people make it to the top, the overall impact of any such appointments on the broader population of under-represented groups is likely to be minimal. While ‘minority’ members in the top echelons certainly can serve as role models or help the career paths of others from under-represented groups, the beneficial effect is going to be indirect and slow.

Appointing more minority members to senior positions is also akin to trying to strengthen the top floors of a building without doing much to its foundation.

So how should the problem of under-representation of minorities be studied and rectified?

Root cause

First of all, we need robust statistical analyses that try to minimise the omitted variables and really understand the root cause of the problem after controlling for most relevant factors.

board280For instance, an analysis of why some people make it to top management but others don’t will include variables such as whether a person had undertaken higher education such as a graduate degree, taken on an international assignment, his/her major field of study, whether the person has taken a break in career and, of course, membership of a minority group.

It is only when we hold these other factors constant that we can conclusively arrive at the impact of minority group membership on career advancement. It is possible that the obstacles to minority recruitment and advancement differ based on contextual characteristics of the broader environment (e.g., the degree of competition) as well as the specific company (high performing or otherwise) and these need to be accounted for as well.

Some academics doing studies using simple and aggregate measures may point to the magnitude of the task (and resources) involved in doing a broad-based statistical analysis. Interestingly, the Grant Thornton study mentioned above had a sample size of more than 6,700, implying a large allocation of resources. Thus it is not the lack of resources, but our lack of desire to really understand the root cause of the problem.

The field of medicine employs a methodology called case-control under which a matched pair of patients (similar in some respects such as age or other demographic factors but different in some key characteristics such as having a disease) is compared, and sometimes tracked over time, to understand the key factors influencing the likelihood of a disease.

Research on minority discrimination could adopt this approach, which allows holding a number of factors constant (through statistical analysis) to determine the exact factors hindering the careers of minority members.

Our suspicion is that minorities are well-represented at the lower levels of organisations, at least in developed countries such as Singapore, but this percentage drops even at middle management levels, let alone top management levels.

Appropriate incentives

A 2010 working paper by two Harvard Business School scholars, Andrew Hill and David Thomas, made exactly this same observation about minority coaches in the National Football League (NFL). They found that better performing teams are less likely to hire minorities to fill lower-level and mid-level coaching positions, but that such teams are more likely to hire minorities into leadership positions.

If we can improve minority representation at middle management levels through appropriate incentives or other policies (e.g., encouraging them to pursue higher education or taking on an international assignment), the problem of representation at the top echelons will ease simply because there are more candidates from minority groups coming up hierarchy, in other words the queue is longer and stronger.

TA6 no borderOur intent is not to say whether there is or isn’t discrimination. Our point is that poorly-conducted studies tell us very little about the extent of the problem and even less about what needs to be done to address the problems, if they exist.

Meaningful solutions to address such problems can be devised only when we understand the problems correctly.

The 2010 Harvard study cited above, noted that the NFL has mentoring programs for minority coaches, connecting younger coaches with more experienced colleagues. It also provides professional training in interviewing and presentation skills to coaches, on the assumption that minority coaches have less background in these areas.

The governments can also do their part, by going well beyond mandating quotas. Governments can provide incentives such as tax breaks or grants to companies that work towards upgrading the skills (or career prospects) of their minority employees through higher education or international assignments, for instance.

The Singapore government has some schemes for improving the skill level of employees of SMEs and it is not difficult to envision similar schemes for minority employees.

In conclusion, to stop a recurring wound, we need to understand its causes and what the effective way to treat it is. Sticking a band aid on a recurring wound only makes the wound less visible, without doing anything to eliminate its root cause.

We need to stop adopting the band aid approach to solving the problems of discrimination and address the root causes of the issue.

Nitin Pangarkar is associate professor of strategy & policy at NUS Business School and Natasha Pangarkar is a junior at Williams College in Williamstown, Massachusetts. A shorter version of this article appeared in the Business Times.

Sign up to our newsletter