The most common misuse of statistics is when based on the numbers, a judgment is made that something is increasing or decreasing. What is generally unknown is that there are standards pioneered in America and accepted internationally for assessing trends.
Generally, you need 20-25 data points that can then be placed on a run chart along with the median (center-line). Depending on the standard, the following trends would indicate an increase or decrease and if it was sustained, would indicate a systemic change: 6 Consecutive Points in a row rising or falling; 8 Consecutive Points in a row above or below the centerline. The simplest standard is the rule of 7 – 7 data points in a row increasing or decreasing or 7 in a row above or below the centerline. (PQ Systems Trends)
LinkedIn – Covid – Misuse of Control Charts by Dr. Tony Burns
Background Info – overview of quality management concepts, principles, methods and tools – Success through Quality
Misuse of Process Behavior Charts is rife. An example of such misuse is the many attempts to use Process Behavior Charts for Covid data.
My recent posts, such as “Down the Rabbit Hole” https://lnkd.in/gzk5bEPq have shown how the right approach is Professor Deming’s “Operational Definition”, or what Dr Shewhart called “Operational Meaning”:
1. What do you want to accomplish?
2. By what method will you accomplish your objective?
3. How will you know when you have accomplished your objective?
The first step is to ask, what are you trying to achieve? Professor Deming wanted to know if the processes he was studying were running at their maximum potential. He also asked whether these processes would continue to run at their maximum potential into the future. The tool he chose was the Shewhart Chart.
The first step with Covid data should be to ask “What do you want to accomplish?” Folk trying to draw control charts for Covid data have not asked this question.
What we want to know about Covid is whether it is getting better or worse. What we want to accomplish is a downward trend. The best method to observe such trends is the Run Chart, or Bar Chart.
Control Charts cannot help answer this question. All that a control chart can show is that the data is non-homogeneous. However we already know this because of the way the disease is spread from one person to many more people.
Control charts are ideal for bringing a process to its full potential and ensuring good quality into the future. They are worthless for COVID-19 data. Use a run/bar chart as shown. Run charts are ideal for showing trends.
“How will you know when you have accomplished your objective?” We will know this when the Run Chart trends to zero.
The root cause for the ubiquitous misuse of Process Behavior Charts is firstly courses failing to teach the fundamentals of Quality. The second cause is statistical software which encourages the blind pressing of buttons without thinking. When you have bought an expensive hammer, don’t spend your time looking for things to use it on. Use common sense instead. THINK.
KEEP IT SIMPLE.
“But, as we have seen, an epidemic is anything but a steady-state system. It grows, changes, and evolves. We do not need to ask the question “Has a change occurred?” because we know that by its very nature every epidemic is constantly changing. This is why any attempt to use a process behavior chart to analyze the daily Covid values is a misapplication of the technique. It is conceptually equivalent to someone computing the average for a list of telephone numbers.” – Dr Wheeler.