Numbers don’t lie. This is a long-held belief that many analysts, scientists and marketers use to prove a point. While there is truth to this statement, there are also a variety of factors that can lead to the misinterpretation of data analyses.
Care must be taken to ensure that the numbers being examined are appropriate for the intended purpose. For instance, revenue may not be an accurate factor for comparing the performance of several retail stores if location is not taken into account. Stores in densely populated areas have far more customer opportunities than those located in less populated locations. For the comparison to be accurate and useful, it is important to use data that incorporates the differences in expected revenue by location.
On a grander scale, remember that just because two things may seem to be statistically related, it does not mean they really are. This is the old warning that “correlation does not imply causation.” Sometimes hidden factors are the underlying cause. Other times, it is random chance that a relationship is seen. A quick tour of the Internet provides a host of examples highlighting this problem, such as ‘the decrease in pirates is related to the increase in global warming.’ In the case of big data, it is dangerously easy to believe you have found a solid link when there is so much data at your disposal.
The problems inherent in data interpretation are compounded simply because we are human beings. We are wired to seek and find connections; it’s how we make sense of the world. However, sometimes the desire to find links can lead us astray and cause us to see connections even when they don’t exist. (By the way, this is the same mechanism that leads some athletes to end up with a lucky pair of socks!) People are also at risk for Confirmation Bias. In other words, we tend to look for information and interpret results that fit our pre-existing expectations and beliefs, rather than seeing things how they really are.
So what can be done to avoid drawing inaccurate conclusions from data? Here are a couple of tips:
- Use healthy skepticism and a scientific approach.
- Work with people who have expertise to know the proper procedures and limitations of the analytics being used.
Data is a valuable tool. It can be very useful when examined in a thoughtful and proper manner. Just be sure that as a distributor of data statistics, and as a consumer, you are aware of the intricacies involved in this not-so-precise science.