You can’t control quality without understanding some of the concepts behind statistical quality control. Here are 7 concepts that are important for managing quality on projects.
Whether it’s widgets, people or processes, population refers to the lot of them. It’s the whole of what you want the information about.
It’s easiest to think of this in terms of physical deliverables. If your project is to make 1000 steering wheels for cars, the population is 1000.
When your population is big, you won’t want to test quality on all of them. That would take too long and cost too much. Take a sample when that’s more practical: a smaller group or subset that represents the whole.
Probability is the likelihood that something will happen. You can express it as a fraction (between 0 and 1) but it’s more commonly seen as a percentage (“There’s an 80% chance that we’ll hit the deadline”).
This is the average of whatever it is that you are calculating. If you had to say what the expected value was for a variable, then you’d say you expected it to be this.
5. Normal Distribution (The Bell Curve)
Every process has variation. That means some of your quality control values will be high, others low, and most fall somewhere around the mean. When you plot those values on a graph you get a line that looks like a bell. This is normal distribution: the most common distribution of values that you should expect from a process.
6. Standard Deviation
This took me a while to get my head around. What it expresses is how close all the values are to the mean. Standard deviation is measured in terms of ‘sigma’. It’s just the name given to the unit, like ‘centimetre’ or ‘dollar’. It’s a statistical term that tells you the spread of the results. A high sigma means the values are spread out from the mean. A low sigma tells you that there is less variation and that the results are all bunched up together.
Without knowing your upper and lower quality specifications all you’ll find out from standard deviation is how bunched up your results are. You need to plot your quality control targets on there too in order to see if your results fall within the target. Otherwise you could be celebrating having a small standard deviation (which is good) only to find out that it is wildly outside your control limits (which is bad).
7. 6 Sigma
The final term it’s worth discussing is Six Sigma. Also the name of a process improvement method, it’s a way of describing what good looks like. First, you need to do your standard deviation work. Know what your quality specifications are. Take the standard deviation output that you’ve created and work out your sigma spread.
Six Sigma is where your results fall +/- 3 sigma from your mean specification limit. In other words, 99.73% of the values in your data set fall between the mean and +/- 3 sigma. There’s little variation in your process and your results consistently hit your quality targets.
You might also see six sigma expressed a +/- 6 sigma. That gives you a breadth of 12 sigma in total (6 each side of the mean of your distribution curve) and that equates to your results falling inside your target 99.99 and a bit% times. All but 3.4 times in every million your process, deliverable, widget or whatever will be on target.
Read next: 3 Levels of Quality
Rest assured that you personally don’t have to know the details of all this. You just need someone on the team who understands it and can apply it. If you are in the kind of company that measures quality in a statistical way, then you probably have a QA team or an analyst who lives and breathes this stuff.
Talk to them; set expectations and work out how you can collaborate to get the best quality control and reporting possible on your project.
You might not need your projects to deliver such as focused, quality result, but regardless of the type of work you are doing it helps to understand what quality means to you and the tools you can use to prove that it exists on your project.