How Does Confirmation Bias Interfere with Data?

Analytics convert a large data set into meaningful information that drives our decisions. But the insights we derive is heavily influenced by our inherent tendency to confirm our existing beliefs. How can we overcome this?

How Does Confirmation Bias Interfere with Data?

2 Jun 2017 by  Jacob Tan

There has never been a more exciting time to access and analyse large chunks of data and convert them into meaningful insights for different groups of professionals across myriad industries. However, as Harvard University Professor and renowned social scientist Gary King asserts, “big data is not about the data.” Indeed, it is the analytics that convert a large data set into meaningful information that drives our decisions. However, the process of selecting relevant data points, analysing them, and deriving meaningful insights from them is heavily influenced by our inherent tendency to interpret new evidence as a confirmation of our existing beliefs—also known as confirmation bias.

This explains why the same data points can give rise to very divergent statistics and insights. While confirmation bias is inherent in all of us, we must be more conscious and cautious when key decisions are at stake. In the HR world, some of these decisions involve reward strategies and how much we decide to pay CEOs and other top executives, in order to drive optimal performance.

Gender pay gap: Fact or fiction?

Recently, a sensational headline generated much interest and debate in the media and among industry experts: Do female directors earn less than male directors in general? It does sound plausible, given what we have seen and heard in media and society over the past decade or so. 

To be precise, the reported statistic was that females on Boards earn 43.2% less than male directors.  Upon conducting a preliminary analysis based on data from our executive compensation database, this statistic seemed sound and valid. As we can see in Table 1, the median pay of males is indeed greater than that of females, among executive and non-executive directors alike. Looking at the numbers, one would be inclined to agree that it is shocking that “Salaries of female directors at listed companies on the Singapore Exchange (SGX) are just over half of what men in similar positions earn” (TODAY Online, 2017) and that the nation should “address deep-seated inequalities including remuneration” (The Business Times, 2017).

However upon conducting a regression analysis, we discovered that gender was neither a significant determinant of Executive Directors’ total compensation nor NEDs’ director fees. Board Committee membership and Chairmanship, number of board committees the director sat on, as well as the firm’s market capitalisation were the significant factors in determining total compensation  and director fees earned. Further secondary research confirmed that structural reasons such as board committee membership are the main drivers of the apparent gender pay gaps.

It may come as a shock to many of us that the fundamental ‘facts’ that we take for truth could merely be strongly bolstered myths. However, with a better understanding of the root contributing factors we can then channel more efforts into solving the real problem at hand. An example would be to increase female involvement in Board Committees as well as Board Chairmanship. The next question we need to answer is around how we can consciously act to minimise our susceptibility to confirmation bias. 

How to correct bias?

Warren Buffet notes that Charles Darwin used to write down all dissonant information within the first 30 minutes to which he is exposed to it, so as to avoid disregarding it. While none of us can imagine this to be practical amidst our hectic schedules, such confirmation checks could take different physical forms within the organisation.

Firstly, we can ensure that Executive Compensation committees and /or HR teams consist of devils’ advocates from different designations and disciplines within the firm, who would raise contrasting views. Secondly, there must be evidence that these alternate perspectives have been thoroughly considered in decision-making (assuming logic and accuracy). Thirdly, the data obtained to support these perspectives and the final decisions made, should be derived from multiple sources, audited, and triangulated within.

Start a conversation with us

Need help with your executive compensation plans? Get in touch with us today.

Jacob Tan

Jacob leads the Executive Compensation Solution for South East Asia. His experience includes benchmarking for executives and board of directors’ compensation, board evaluations, employee share plan design and implementation, short-term incentives design, Economic Value Added (EVA) incentive plans, productivity benchmarking and executive contracts.

Get in touch
Jacob Tan