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

DCMR: a SQVID for analysts

Introducing a new analytical imagination activation tool

I've been reading The Back of the Napkin. Ninety-eight pages into this book, we get introduced to a thing called SQVID, which is Dan Roam’s “visual imagination activation tool”. SQVID is a way of getting us to be better at imagining things. Given that the subtitle of his book is “solving problems and selling ideas with pictures”, it’s easy to see why Dan Roam might think it’d be a good thing for us to find better ways of imagining things.

SQVID shows you how to look at the same thing in different ways. In his book he uses the example of an apple. Suppose you are trying to explain what an apple is to someone who’s never seen an apple. And suppose you can’t describe it in words because you don’t speak the same language as this other person. You’d have to draw an apple. Which is easy enough, but there are different ways of drawing an apple. And SQVID helps you navigate your way through the different ways of drawing an apple.

The S in SQVID stands for “singular”. Singular as opposed to plural. In fact, all the letters in the SQVID acronym are to do with dichotomies. The Q stands for quality as opposed to quantity. The V stands for vision as opposed to execution. The I stands for individual attributes as opposed to comparison. And the D (which is actually an upper case Greek Δ – or Delta) stands for change as opposed to status quo.

So, when applied to this apple-drawing scenario, you might choose to draw just one apple or you might choose to draw many apples. That would be the S. You might choose to draw an apple that’s nicely lit, that looks good enough to eat, that shows all of its different shades of green, or you might instead choose to draw a graph that shows the different quantities of apples grown in the different counties of England. That would be the Q. And so on.

The thing I find most appealing about Dan Roam’s SQVID imagination activation toolkit is that it encourages us to look at the same thing in different ways. It forces us to imagine different ways of showing. It’s a systematic way of getting you to use your imagination. And it’s this attribute of SQVID that has been sticking in my mind in recent weeks. It’s been bothering me because there’s an analogy with the way that data analysts should work.

People who are trying to draw things—let’s call them artists—self-evidently need to use their imaginations. But analysts also need to use their imaginations. We need to discipline our minds into thinking of different ways of using data to describe reality. And of course we often don’t. We don’t really think of our jobs as being creative. You rarely see “must have a vivid imagination” in the job specification for an NHS data analyst. But we are wrong to think that. Actually, we do need to be able to imagine stuff.

So, suppose that there is a parallel between drawing pictures to communicate and analysing data to communicate. Is there an equivalent to SQVID for analysts?

Well, in one sense, why bother inventing a new toolkit? Can we not actually just use SQVID straight out of the box.

If we think in terms of the S for example, we could choose to describe something using just one number (the average) or using many numbers (the minimum, the lower quartile, the median, the upper quartile, the maximum).

As for the Q, we could think of whether to describe something using an indicator of quantity (a count of the number of admissions) or—instead—an indicator of quality (for example, the number of readmissions).

And so on.

But I also think we can conjure up a SQVID of our own. And we can conjure it up from statistics, from the academic discipline of statistics.

If you pick up pretty much any introductory stats textbook, I reckon you could take all the material therein and divide it up into four sections: distributions, comparisons, margins of error and relationships. I admit that that is a bit of a broad-brush approach to a distinguished and complex academic discipline, but let’s face it, it’s not that far off the mark, is it? And it’s these headings—distributions, comparisons, margins of error and relationships—that are the analyst’s equivalent of the SQVID. Let's call it DCMR.

So whenever an analyst is looking at a problem or an issue that needs data to shed light on it, they can first of all say: “What if we displayed the data as a distribution?” And that would lead them to think about drawing a histogram or a box plot or calculating the percentiles instead of just trying to describe it as a single number.

Secondly, they could ask: “What can I compare this with?” In reality, this is what we already do quite well. Most of the time we are doing analysis, we are doing analysis that involves comparison. But we can think about comparison as comparison through time or as comparison in space. And we can borrow from statistics the idea that we might want to compare the experience of one group of patients with another group of patients to see if those experiences are different.

Thirdly, margins of error. I was reading an article the other day that tried to sum up how we should critically read a piece of science. It boiled down to three questions:

  1. Does the article present numerically precise estimates of the quantities of greatest substantive interest?

  2. Does the article include reasonable measures of uncertainty about those estimates?

  3. Does the article require little specialized knowledge to understand?

It's the second of the questions that matters here. Have we included a reasonable measure of uncertainty? Calculating margins of error is something we hardly ever seem to do. Yet it ought to be at the core of what we do. So the M in the analyst’s toolkit stands for margins of error.

And finally: relationships. Analysts should be asking: is there a cause-and-effect relationship we can explore here? What causes long waits? What causes delayed discharges? What causes high levels of readmissions? Is one indicator related to another indicator?

DCMR. I admit. Hardly the most memorable of acronyms. In fact, not even an acronym. Also—confusingly—it ends with the two letters MR which means that you might at first glance think it’s some new fangled way of calculating a mortality ratio.

But it’s actually a toolkit that analysts can borrow from statistics. And it runs right through the middle of the one-day course Visualizing Statistics. There’s an open course scheduled for a week today (Friday 22nd March) in Wallacespace Covent Garden and there are places still available. Please contact me if you are interested in booking a last minute place.

[15 March 2013]


Make a comment


Job Title:


Email address:

© Kurtosis 2013. All Rights Reserved.