Marketing Viewpoint by Ruth Winett
Can You Trust Your Company’s Data?
Questionable scientific data. What about business data?
Authors of an article on a new “room temperature superconductor” asked the journal Nature to retract the authors’ article because the “lead researcher misrepresented data.” 1 In 2022 5,500+ scientific papers were “retracted,” according to Retraction Plus because volunteer reviewers had detected fraudulent data. Fraudulent data is a widespread problem. Data influences decision-making in medicine, the social sciences, economics, politics, and especially in businesses.
Can you trust the data used in your company to make major decisions? Following are some of the factors that contribute to the creation of fraudulent data, as well as tests to detect fraud. 2
Why Fraudulent Data Is Widespread
Fraudulent data may be the result of human error or “’p-hacking,’ cherry-picking data or analyses to make insignificant results look statistically credible.” Among the underlying causes are the pressure among scientists and academics to “publish or perish.” Another reason that fraudulent studies get published is that “volunteer experts” [peers] review articles. Their function is “to ensure the quality of the published work,” not to review the data.
In the business world, the push to attract and retain customers, to retain suppliers, to satisfy regulators, to thwart competitors, and to encourage investors can also result in the publication of fraudulent or inaccurate data designed to make companies look good to stakeholders. And, as in the world of science, it is essential that qualified people review and assess the data.
Data Colada’s Tests for Detecting Fraud 1
Three former Princeton grad students now operate Data Colada which like other “data detectives” exposes fraudulent scientific data. The organization has three tests for detecting fraud. Businesses, social science organizations, and non-profits should also apply these tests. We have added current examples:
1) “Eyeball the data to see if they make sense in the context of the research.” This implies viewing all data with skepticism, including data and charts in the press. Now, people should also view images with skepticism. They may have been generated or altered by AI.
2) Doubt “improbable claims.” Why did smart, worldly people believe Bernie Madoff’s promise of stock market returns that were “too good to be true”?
3) Be suspicious of “perfect data in small data sets. Real world data is random, chaotic.” Demographic data can be misleading when a small number of categories are used to define ethnic origin. One common problem is failing to provide a “mixed race” option.
When making critical business decisions, apply a healthy dose of skepticism about the supporting data. First, apply Data Colada’s three tests. Then consider how the data was derived. If a survey or interview was used, did the questions elicit valuable insights or merely ask throwaway questions? (e.g., ”Would you recommend product ABC?”) Is the sample large enough? Was it a random sample? Did the people who analyzed the data have the right expertise? Did management insist on a pre-determined conclusion? Did time constraints, a desire for sales growth or individual career advancement impact the outcome?
https://atonce.com/blog/jokes-on-statistics . Why statistics are useful. Also, statistics jokes:
“Why did the statistician become a tailor? Because he knew how to fit the data!”
Actionable Business Insights
Copyright ©10/23 Ruth Winett. All rights reserved.
For Email Marketing you can trust
Winett Associates tel: 508-877-1938 email
©2023 Winett Associates. All rights reserved.