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

 

   

The Sex Appeal of SPC

Why do health service managers find statistical process control so seductive?

The rise and rise of SPC

Over the last ten years or so, statistical process control (SPC) has become the lingua franca of measurement for improvement in the NHS. Run charts and control charts used to be graphs that were drawn by engineers in factories; now they’ve become features so familiar in the health service landscape that Kurtosis offers a one-day course for analysts that shows how to draw, interpret and explain these graphs. In fact, they are more than just features in the landscape: run charts and control charts often get evangelised about by health service managers.

From an information analyst’s perspective, you'd think this would be a good thing. It’s normally quite difficult getting NHS managers to take data seriously. So if we’re now in a situation where they’re actually getting enthused about data, then we surely have to welcome that with open arms. But it’s not quite as simple as that. When it’s the managers telling the analysts which type of chart to use to plot the data, we might want to conclude that something's afoot. We have to ask why.

Why does SPC “work” as a way of visualising data? How does SPC make visible the complex problems and realities of the health service that other forms of data analysis and presentation do not? What is it about SPC that managers find so seductive?

Large blobs and Myers-Briggs

SPC is strong on formatting conventions and one of the conventions of SPC is that lines on graphs representing actual observations (in the chart below, for example, it’s the individual waiting times of people attending the clinic), these lines should have blobs on them. Whereas lines on your graph representing derived or calculated numbers (in the chart below these are the middle line—the average waiting time of people attending clinic—and the dashed lines: the upper and lower control limits), these lines should be blob-free.

NHS ANYWHERE | ANYTOWN GENERAL HOSPITAL | A RANDOM CARDIOLOGY OUTPATIENT CLINIC

Waiting times (days) of outpatients referred in July 2011

Source: Patient Management System

It’s as if the blobs are accentuating the hard “reality-ness” of the thing they are representing. They are saying: “See this line with blobs on it? The blobs are showing something concrete, something tangible, something real. Something so real that you can practically touch it.” Whereas the blob-free lines, they are just abstract theoretical constructs based on formulae.

And here’s another thing. SPC’s formatting conventions even stretch to making the blobs on the lines really quite large. Most data analysts would flinch from making their blobs (they would probably prefer to call them “markers” rather than “blobs”) of a size that uses so much ink. But take a look at an SPC textbook and you’ll see large blobs.

I think this “blob factor” is a substantial part of the appeal of SPC. Managers see the blobs and realise instinctively that they represent concrete, measurable things. Things that they recognise from real life. And this immediately makes the graphs more understandable, more explainable, more sexy.

Here’s why.

If you’ve ever been Myers-Briggsed you’ll know that one of the end results of the Myers-Briggs process is that you are assigned four letters that basically sum you up.

The second of these four letters will be either an N or an S. If your second Myers-Briggs letter is an N, then you prefer to take information in intuitively. (Yes, that’s right, the N stands for “intuition”, which is a spectacularly unintuitive thing for Myers-Briggs to do, but that’s a whole different story for another day.) N preference people are more comfortable with abstract theoretical concepts. They are more attracted to the dotted lines based on formulae than they are to the blobs that represent the actual observations.

If, on the other hand, you have an S preference, this means that you like to take information in through your senses. The S stands for “sensing”. You literally “make sense” of the world by seeing, hearing, touching, smelling and tasting things. Whenever you have to disentangle a complicated theory or concept, you probably prefer to disentangle it by working through some practical, real life examples. Ss prefer blobs.

NHS managers are mainly S preference folk. The formatting conventions of SPC emphasise the blobs. It’s a match made in heaven.

Keeping to the straight and narrow

Here’s another reason for SPC’s appeal. The ‘C’ in SPC stands for “control” and managers need to be in control of things. Now control charts—with their upper and lower control limits—obviously engender a sense of control quite explicitly. But I think they also do it in a more subtle way that affects all of us. By way of explanation, I noticed some time ago that new Citroen C5 cars now come equipped with a "lane departure warning system" that lets you know if you start to wander out of your lane when you’re driving along the motorway. According to the Citroen website, the system “detects any unintentional lane changes at speeds of 50mph and over and vibrates the driver’s seat on the side of the drift, prompting them to get back on course.”

Reading about this safety feature in a glossy brochure whilst waiting for my car to be fixed, my mind naturally enough turned to control charts. Control charts, when you think about them, they do sort of resemble motorway lanes, with their parallel horizontal dashed lines exhorting you to stick to the space between them, subliminally admonishing those who dare to transgress them. I wondered whether we all have some kind of innate, unfulfilled desire to keep to the straight and narrow. Is this “lane departure warning system” also part of the appeal of control charts?

Reinforcing the illusion of numeracy

When I did A-level Economics at school, one of the teachers we had—let’s call him Mr Burns—seemed to know pretty much nothing at all about the subject. But he was the Head of the Economics department so the problem wasn’t really addressed, and what actually happened was that the other two Economics teachers just tacitly accepted the fact of their boss's incompetence by making sure they left enough time to teach us his bits of the syllabus as well as their own.

Here’s the thing I’ll always remember about Mr Burns. In the first class he taught us, he drew a graph of a monopoly on the blackboard. This graph might be familiar to those of you who’ve studied Economics, and it looks something like this:

This graph had an almost iconic significance for Mr Burns. Here is what I think his reasoning was:

This is a self-evidently complex graph. So you must be self-evidently intelligent in order to be able to draw it on a blackboard. Therefore you must be in command of your subject. Even though you’re actually not, but it’s important that you need to be able to project that kind of image because, let’s face it, you’re the Head of Economics and people will expect you to be able to draw and understand graphs like this. So if you can just remember how to draw this one graph, that ought to be enough to maintain the illusion of competence.

I sometimes worry that control charts are a bit like that for some NHS managers. A bit. Only a bit. Control charts do kind of look like the monopoly graph: lots of lines, some of them solid, some of them dotted. Numbers. Labels. Axes. If you draw a control chart on an office whiteboard and say something like: “We need to start looking at our KPIs using charts like these!” then people will assume you’re tremendously numerate even though you’re perhaps—how can I put this?—perhaps not quite so tremendously numerate.

The Kandy-Kolored Tangerine-Flake Streamline Baby

Forty-six years ago Tom Wolfe introduced the idea of what he called “totem newspapers”.

A totem newspaper is the kind people don't really buy to read but just to have, physically, because they know it supports their own outlook on life. They're just like the buffalo tongues the Omaha Indians used to carry around or the dog ears the Mahili clan carried around in Bengal.

So here’s a controversial, Tom Wolfe-inspired, and somewhat zany conclusion for you. How about we think of control charts as totem charts? Managers don’t ask for them to study them but just to have them, physically. They support their own outlook on life. They reinforce an image that they are highly numerate. They demonstrate that they are in “control” of their department and that their indicators are sticking to the “straight and narrow”. Oh yes, and the charts have blobs on them, which makes at least that part of the graph easy to explain to others.

What's not to like about all of that? No wonder SPC has sex appeal.

[6 October 2011]

 

Make a comment

Name:

Job Title:

Organisation:

Email address:

     
© Kurtosis 2011. All Rights Reserved.