STATISTICAL *SIGNIFICANCE*

By Divya Khanna, AVP & Regional Planning Director (South Asia), JWT Bangalore.

“I am the one in ten,

A number on a list.

I am the one in ten,

Even though I don`t exist.

Nobody knows me

Even though I`m always there –

A statistic, a reminder

Of a world that doesn`t care.”

- Ub40¹

Anyone who has ever expressed an opinion has felt the pressure of showing proof to support that feeling in the gut that tells them it’s true. There has to be an example to prove the opinion or at the very least to refute its absolute opposite. And when the opinion will impact the direction of where large sumsof money are to be invested, nothing less than statistical data will do. But why does this data automatically get the air of authority that trumps the most impassioned and gut-felt belief?

I do get it – a belief, by definition, could be different from what’s actually out there.

Similarly, by definition, a probability is not necessarily a certainty.

And extrapolation of data is, evidently, an extrapolation.

Let me explain what I’m trying to say. Just because “only about 15% of C-suite jobs” in America are held by women2 , it doesn’t automatically mean that without the alleged discrimination they face, women would be proportionately represented in these jobs. Such an assumption, and it is an assumption, would not take into account the other factors at play. For example, as Sheryl Sandberg mentions in her book ‘Lean In’, the tendency of women to downplay their achievements and give credit to other people versus the opposite tendency of men. Or the evidence that many women, as they move up the corporate ladder and simultaneously raise their children, start to make career choices that don’t head towards earning them a C-suite job.

When we reduce a person to a number, we feel more certain as numbers give the illusion of certainty.

But there are 2 caveats to keep in mind:

1. Numbers too can be misinterpreted and manipulated. Just ask anyone who’s ever committed or detected an accounting fraud!

2. We are losing the nuances of the person behind the number. Every woman of the 15% in C-suite jobs is not like every other, there are bound to be differences that may not allow a one-size-fits all approach in any decision regarding them as a group.

Don’t get me wrong, I do believe statistical data is useful. But we must always remember its context. The language of data is just as uncertain as the language of gut – the terms probability, extrapolation, assumption, levels of significance, etc. are evidence of this. There is anecdotal evidence to support (but not confirm) every side of the argument between qualitative and quantitative based decisions. This suggests that we can either side with one and take our chances or combine both sides to improve them – but at the end of it, no matter what we do, we are still taking a chance and cannot be absolutely sure of any definite outcome, except in retrospect.

1 One In Ten lyrics © EMI Music Publishing, Sony/ATV Music Publishing LLC, Universal Music Publishing Group

2 Harvard Business Review, September 2013, South Asia edition