Home   About Kurtosis   Courses   Course Calendar   Booking Information   Ideas   Technique   Clients   Contact

Adding a time dimension to control charts

What happens when we try to show spacing and clustering on a control chart?

When we plot individual observations on a control chart (using an X chart), we always take care to plot the values in strict sequential order. But the chart as conventionally drawn fails to show the extent to which the observations might be spaced or clustered. So although we might label the arrival times of each individual patient (as we have done here, along the horizontal axis), the chart design itself gives no obvious indication of the patient arrival times:

One way of incorporating spacing and clustering into the chart design would be to re-draw the X chart as a scatterplot as opposed to a line chart.

[ See the Excel spreadsheet Scatter_Control.xls for a worked example]

We don't really lose any information from the resulting chart but we do gain a clearer visual understanding of the times that each patient arrived in the emergency department:

The clustering of arrivals between 9am and 11am and between 3:30pm and 5pm is now clearly visible.

I'm not trying to argue here that times of arrival are an important cause-and-effect factor in terms of how long AMU admissions spend in the ED before being admitted. What I am suggesting is that if you draw the chart like this you stand to gain additional knowledge and understanding without having to compromise anything else.

It's entirely possible that people do this already, but I've not seen examples of it.

[6 August 2013]


Make a comment on this blogpost:


Job Title:


Email address:


Kurtosis 2013. All Rights Reserved.