Data can be a non-complicated topic. At it’s core, you capture data, you extract data, you process data, and you report on the data. Data is a deceptively simple process, yet we often hear reports of misleading financials or suppressed results. In this week’s edition of “Featured Insights”, we will highlight three case studies that will demonstrate the importance of ethics in data.

Data Ethics Case Study 1: Kraft Heinze

In 2021, two executives from The Kraft Heinze Company were charged by the SEC for falsifying cost-savings between 2015 and 2018. (SEC, 2021)  Kraft Heinze ordered to pay $62 Million to settle the charges and, undoubtedly, the executives were ordered to pay a fine plus were prohibited from serving as an officer or director of a public company. Nonetheless, Kraft Heinze has a significant blemish on its reputation. The cost of being unethical when reporting data comes at a huge cost both financially and reputationally. While this instance is not likely to impact consumers purchasing their product, they are being used as an example of what not to do in a book about data ethics.

Data Ethics Case Study 2: Enron

Our second example is a classic one, Enron. For those who are unfamiliar with the Enron Scandal of 2001, it was one, if not the most, notorious corporate fraud case in the United States. Enron was one of the world’s largest electricity, natural gas, and communications companies with a market capitalization of over $60 billion. (Healy and Palepu, 2023) 

Enron’s Collapse

While there are multiple factors that caused Enron to fail, the one we are focusing on is accounting fraud. Specifically, Enron used complex and deceptive accounting practices called “mark-to-market accounting” which inflated income and asset values. “Mark-to-market accounting” involves adjusting an asset’s value to reflect the asset’s “fair value” based on the current market. This allowed Enron to hide debt off their balance sheets to give the illusion of a highly profitable company. Furthermore, the cost of this unethical data practice for Enron was more than a simple fine. Enron declared bankruptcy resulting in thousands of workers losing their jobs and tens of thousands of investors saw their investment accounts “depleted or destroyed”. (Healy and Palepu, 2023) Without a doubt, this was one of the largest financial collapses of an individual company at the time.

Data Ethics Case Study 3: Purdue Pharma

While the last two examples focused on unethical reporting of financial data, our last, and most important example, shows the consequences of unethical reporting of data during validation to push a product to market. That example is the Purdue Pharma Scandal. Purdue Pharma created the painkiller OxyContin which they touted as non-addictive because of slow-release properties. (Gale, 2022) Evidence emerged showing that OxyContin was addictive and played a critical role in the opioid addiction epidemic that has been plaguing the United States. What makes this especially egregious is that Purdue Pharma created a chart from a study in the mid 90’s.

What Happened?

The study claimed that OxyContin peaks quickly then plateaus in the blood stream to provide a steady dose of pain relief (Edwards, 2011) This chart was on a logarithmic scale, not a linear scale. For those unfamiliar, a logarithmic scale is a method to display data over a large range but keeping the graph compact. This differs from a linear scale where the numbers may not be evenly spaced out like that of a linear graph. 

Dopesick

This was highlighted exceptionally well in the TV show “Dopesick”.  As the data scientist highlights, the logarithmic scale shows a smooth peak and plateau of the medication in the blood. There are three major misleading factors. First, the dosage appears to be in the middle of the graph which is will mislead people who do not carefully read the graph’s scale. Secondly, the concentration is 30 ng/mL in the blood. Again, this is a logarithmic scale, not a linear scale meaning this is not necessarily 30 ng/mL in linear terms. Lastly, Purdue Pharma failed to transform their logarithmic scale back into a linear scale. Figure 2 clearly identifies two peaks. The concentration of OxyContin found in blood on a linear scaled graph contradicts the logarithmic graph. The concentration is not a gradual build, plateau, then steady release, but an instant peak followed by a quick and sustained crash followed by a secondary peak hours later.

Impact

The fallout from this scandal was monumental. From a business perspective, Purdue Pharma declared bankruptcy and paid a large fine. From the human perspective, hundreds of thousands of people became addicted to and died from abusing OxyContin. The Purdue Pharma Scandal demonstrates that not adhering to data ethics can have far reaching consequences.

Conclusion

There is a huge price that must be paid if you do not adhere to strict data ethics with your reporting.  There is a grey area that we, as data professionals, must walk. How do you navigate the grey area? Find out if there is business alignment. Reach out to a mentor or someone you trust. Work to understand the reason behind the ask. You should be able to justify each decision from a data perspective or business perspective. Lastly, any decision that is made should be documented. I would strongly recommend that you document the decision, who authorized the decision, the date the decision was made, and your initials. In addition to being a good reminder, proper documentation will help you should a data decision be questioned. Documentation has saved me and it will save you.  Remember, we are all data stewards. We must uphold the highest ethical standards and let the data tell the full story.