When you know too much to see what’s happening


Today, I tried to print an invoice and for the first time in eight years, it didn’t print.  I tried a second time but still no luck. As I attempted to print, a few subtle clues emerged. First, the print dialog box closed way too fast (compared to usual).  In addition, my program generally adds the word “printed” next to the invoice when I print but that wasn’t showing up. Almost immediately, an idea popped into my head as to what might be wrong.

About three weeks ago, I switched over to using a Mac.  The invoicing program that I use is only available on Windows.  Luckily, I found a program that creates what’s called a virtual machine on my computer. This allows me to run Windows programs on my Mac.  Running one operating system within another would seem to be a breeding ground for problems.  Clearly there was an issue with the printer interface between the Mac and Windows.

I looked on-line to find a solution.  Others had problems getting their virtual Windows programs to print.  There were a lot of recommendations on how to fix it.  I tried them all but still had no luck.

Then I tried an experiment.  I printed a test page from the printer driver in Windows.  It printed perfectly.  That, plus the strange printing behavior I noticed earlier should have been a red flag that my theory was incorrect.  Yet, I was still convinced that the problem had to do with the virtual machine program.

After 45 minutes with tech support (who also though it was a virtual machine problem), it finally occurred to me that the problem wasn’t the virtual machine program after all.  It was the invoicing program. A few quick searches found evidence of a quirky bug in the program.  Sometimes it randomly stops reading the graphic file used in the logo. When that happens, it simply doesn’t print.  As soon as I removed the logo graphic from the invoice, it printed perfectly.  How could I miss such an obvious answer for over two hours?  The data were right in front of me.

As often happens with data, I was blinded by my experience.  I’d printed invoices successfully for years.  I never had an issue.  However, in my transition from Windows to Mac, I’d been experiencing continual compatibility issues. As a result, when the printer didn’t work, my past experience led me to believe that it was most likely a compatibility issue between the two operating systems rather than an issue with my invoicing program.

I fell victim to two of the most common biases that impact effective data-driven decision making: confirmation bias and availability bias.  Both of these biases are discussed extensively by Daniel Kahneman.

Confirmation bias occurs when you generate a belief or conclusion about a situation and unconsciously seek and focus only on data that confirms that expectation. The small amount of data that supported my virtual machine theory was enough to outweigh the data that disproved it.  Our unconscious brains work very hard to filter out data that do not support their foregone conclusions.

Availability bias is when you unconsciously place greater weight on things that you more readily remember.  In my case, the recent compatibility problems were at the forefront of my mind.  Together, confirmation bias and availability bias prevented me from seeing a pretty simple and obvious problem.

What expectations do you bring to your data?  Do you have expectations about high or low performing products or people?  Do you anticipate which metrics are performing well and which are performing poorly?  How might those expectations be influencing how you see or interpret that data?

You might be thinking that you would never miss something so obvious. If so, you are falling into what I call the rationality trap. The rationality trap is the belief that your brain works rationally, examines all available data objectively, and comes to a logical decision.  The rationality trap is an illusion.  There is a pretty large body of research that proves that we are much less rationale than we think.

It’s impossible to proactively eliminate your bias.  The most significant biases take place before you even become conscious of the data.  Instead, focus on correcting for those biases.  Slow down a bit.  Look for alternative explanations of your current situation.  Don’t let your experience stop you from seeing what’s right in front of you.

Print Friendly, PDF & Email