It’s been a
while since I blogged. So what better topic to get back into the swing of
things than Statistical Significance? (I know! How have I gone this long without
one?!)
I like to
think of statistical significance as one of those things that I can get excited
about, so you don’t have to. And, if I’m really honest, few of my clients do
get all that excited about it. That’s their loss. But they do often appreciate
the insight it provides. I consider it part of my overall strategy for turning
data into stories. After all, if you’ve got both then you’ve got more ways to
convince more people of your truth.
So, if I’ve
got a load of results from:
- a survey
- a series of 360 reviews
- focus groups who have scored some qualities of where they work
Then I can rank
those scores top to bottom. And I can produce some nice tables and charts. I
can tell you what the top 5 and bottom 5 are. Or top 10 and bottom 10. Or, well
you get the idea. But what’s to say the top 5, 10, whatever are the most
important?
If people
have grasped any statistical concept, then it tends to be that of the bell curve. You
can picture the shape, and you probably understand that in a “normal
distribution” most results will be in the main body of the curve – the bulgy
bit. They may be high or low, but if they’re in the middle, then that highness
or lowness isn’t yet significant. It’s only when they move to the left or right
that they get more interesting.
So I’ll look
at your results and tell you what are the significant
top x results and the significant
bottom y results. Those that are apart for the rest of the results; those that
people feel most strongly about. Of course, I’m conscious of “lies, damn lies
and statistics”. Sometimes I’ll use a standard deviation, sometimes an average
deviation – it’s got to be what throws most light on the data.
That gives me
– and my clients – more confidence that we’re looking at the truly important,
symbolic or compelling results. And we can build that into a more powerful narrative
– of what’s being said, what needs changing and what should be done.
