Drawing their attention to the histogram, I might say, “You can see on the histogram that 26.5% of the sample waited 3 minutes or less. 74.5% of the sample waited 21 minutes or less, meaning 25.5% waited longer than 21 minutes. There’s a long tail, pulling the mean to the right. What if our next step is focusing on root causes of waits greater than around 20 minutes or more?”
On the run chart I’d say, “Most of the data points are below 20 minutes, but when the data goes up, it seems to stay up in little groups of 3 to 5 patients. We should find out why this happens. It looks like it happens every 7-20 patients. Let’s schedule a gemba walk focused on this pattern.”
Notice I included the mean on both charts so I can say, “Look how poorly the mean represents the dataset as a whole.”
Of course we can’t just ignore the mean. We need it to calculate financial impact and conduct two sample t-tests. But the deceptive duo of mean and standard deviation are put in plain perspective with the honest histogram and reliable run chart. They contain more information and communicate it with greater clarity.
