Click to learn more about Sharad Varshney.

“We don’t see things as they are. We see things as we are.” –Anais Nin, French-Cuban-American novelist

Cognitive bias damages research and, in turn, analytics. However, you might be surprised to learn that bias is fundamental to the function of the human brain.

In one cognitive bias study at Stanford University, “Psychological Dimensions of the Israeli Settlements Issue: Endowments and Identities,” researchers showed participants specific video footage documenting Arab-Israeli conflicts. The test group, which included both Arab and Israeli volunteers, was asked to comment on the balance of the reporting. Despite watching the same footage, the Arab volunteers perceived more anti-Arab references, while the Israeli volunteers felt the video was more anti-Israeli. 

So, how did this happen? There are several theories as to why our brains work in this manner. One theory is that bias occurs because the human brain has two hemispheres. The right hemisphere is tasked with collecting information, while the left hemisphere is required to process and interpret it – much like how data is moved from various sources to systems better suited to data analysis.

With so much information available to process, the left hemisphere picks select pieces and develops a theory based on this data. The trouble is, once the theory is established, the brain will only accept information that supports it and disregard everything else.  

In this article, we will:

• Discuss bias in interpreting data
• Explain how curiosity reduces bias
• Encourage data-driven curiosity 

Cognitive Bias in Data Interpretation

Cognitive bias happens when there is a deviation from rationality in judgement. In these situations, the brain makes logical shortcuts to process the world around it more efficiently. In some cases, these shortcuts can prove time-saving, but what about situations where the outcome is detrimental?

Decision-making in the workplace is the bread and butter of the business world. Every aspect of a company requires users to make operationally unique decisions necessary for success. Yet, from hiring new employees to deciding on production plans, bias-based decisions are easily overlooked. 

The example we cited at the beginning of this article falls under the category of confirmation bias, where only information that supports your predetermined stance on an issue is processed. Large-scale data interpretation is vulnerable to a different type of cognitive bias called availability bias. Availability bias takes into account recent information when making a decision rather than the full spectrum of relevant data.

For example, there is currently a shortage of computer chips called semiconductor or integrated circuit (IC) chips, which are used in most technological products from cars to kitchen appliances. IC chips are made by only a few companies due to the specialized conditions required for the manufacturing process.

During the early stages of the pandemic, the automotive industry closed its doors for safety reasons and in anticipation of a reduction in sales in line with unfavorable economic conditions. They projected a significant decline in car buyers, and a repeat of the downturn experienced during the Great Recession of 2008. 

This is a prime example of availability bias because the analytics did not take into account the full implications of the pandemic, a different crisis to the Great Recession.

Conversely, with a large proportion of the world working from home, many technology companies predicted a potential boom in sales. Students would need remote access to education services, adults would have to navigate the business world from home offices, and household income would be spent on consumer goods instead of annual vacations. 

While the automotive industry purchased fewer IC chips, many technology companies began to stockpile them. By the time automobiles went back into production, it was already too late. The limited IC chip manufacturers could not keep up with the pressing demand. Now, IC chips are experiencing a global shortage, and the automotive industry is behind. And all this can be attributed to availability bias.

Ford predicts that the IC chip shortage will impact their profit in 2021 by approximately $2.5 billion. Consequently, consumers can expect price increases.

In order to combat bias, such as availability bias, you must access the full scope of relevant data and understand how the connections and implications reduce the possibility of biased decisions. Essentially, more information provides a clearer picture for decision making. 

Data literacy and access to big data pave the road for not only reduced bias, but data-driven curiosity. This curiosity can be used to enhance the decision-making process of a business.

Why Does Curiosity Reduce Bias? 

With so many discussions about the business potential of big data, the smoking question is, “How can big data influence decision making if the mind instinctively chooses a predetermined outcome?”

Let’s examine an industry where risk assessment and data decisions directly impact personal wealth: the stock market.

When the pandemic spread, the stock market experienced its fastest decline in history. However, within six months, the markets recovered and surpassed previous peak records.

An investor will examine many sources of data to determine investment potential and possible returns. Even when the market plunged, investors looked to the future and invested accordingly. With more people working from home, more technology would be needed. As the vaccine developed, health care stocks would go up. In fact, many new investors entered the market for the first time.

Now that people were at home and had less incentive to spend money on nights out, many turned to first-time investment in stocks. Stock trading apps such as Robinhood and Public saw extreme growth. 

Even these newcomers to investment knew to do their research before committing any money. These new users received a great deal of information, enabling them to weigh the positives and negatives of a company before committing to purchasing stock. This is the definition of data-driven curiosity. 

How to Encourage Data-Driven Curiosity?

Nowadays, lots of companies are striving to achieve data-driven growth. However, the process of collecting and distributing data for decision-making is difficult primarily because it involves many people who will undoubtedly bring with them their own biases. 

Data owners are invested in individual gains, but curiosity in the workplace leads to innovation and growth at both a personal and corporate level. To enable users to be curious about the scope of the data in their organization, employers can provide tools that enable them to ask questions about it. 

A Data Governance tool organizes, manages, and presents data in an accessible way. This makes accessing data easier and fosters data-driven curiosity. Deeper and more complex questions can be asked, and answered. Decisions can be made with higher confidence and there is less potential for falling into the trap of complacency and bias.

High-quality data, by its very nature, is unbiased. It is factual. So, when users are provided access to it, they can ask independent questions about it, and the information they receive is honest and impartial. Employees can also follow recommendations about related data sets and further increase their knowledge to make better business decisions. Any errors can be tracked back through data lineage. 

Data lineage is also a mechanism for trust. It shows where the data originated and how it moved over time. When using trustworthy data, the likelihood of cherry picking data to show the desired results is significantly reduced. 

Essentially, when carefully managed, data-driven curiosity counteracts preexisting cognitive bias.