Are You Deluding Yourself With Data?

Data can be misleading if you’re not careful.

It’s important to think critically about the source and type of data and avoid jumping to conclusions. An article on the Manufacturing Business Technology website warns manufacturers against being deluded by data. It’s easy to read too much into data if you don’t question it.

As an example, the author outlines how sales data could be misunderstood when setting color specifications for a new product line of an electronic device. Suppose a retailer provides annual sales data for similar devices that shows 50 percent of the devices sold were black, 20 percent white, 15 percent blue and 15 percent red.

Does the data show that 50 percent of the new product line should be black? That’s a bit of a reach with the given information. What if the sales ratio merely reflects the ratio of colors put on the sales floor and not actual customer preference? That’s just one possible explanation demonstrating the need to dig deeper in data.

When conducting data analysis, manufacturers need to ask themselves what objectives they’re trying to solve and what questions they should be asking. What data do you need and where can you find reliable data to answer your questions? Any study is able to provide some interesting observations, but if it’s a small sample size, then you’re not really using statistically sound data.

Trying to prove something can lead to leaps of logic. Instead, ask what you’re trying to disprove as opposed to what hypothesis you’re trying to prove.

It’s often very easy to over-engineer an analysis. Remember, data needs to be relevant and believable. There’s a lot of information out there, and it can be overwhelming. Rather than trying to be perfect, figure out an error rate (plus or minus) that you’re comfortable with. For example, could you live with a 5 percent error rate? If so, don’t spend time aiming for perfection.

To successfully analyze data, define what measurements you’re going to use, determine the data integrity and decide whether the data is relevant to the decision you’re going to make. That’s the key to using data correctly and not allowing data to delude your business.