Lies, damn lies and statistics

This morning, ex-Apprentice contestant Katie Hopkins appeared on BBC Breakfast deriding the public sector for striking when the private sector works harder and gets paid less. I almost spat out my pretentious organic muesli all over my Habitat coffee table. Let’s hope I never need the guarantee.

Katie is apparently not alone in her opinion. A brief “Google” returned the following sample quotes:

Public sector pay has exploded out of control” David Frost, Director General, British Chambers of Commerce.

“I can understand the unease many private sector workers feel when they see their contemporaries in the public sector not only getting better conditions and pensions, but also better pay.” John Philpott, Chief Economist, Chartered Institute of Personnel and Development

“The figures illustrate a sharp divide between the public and private sectors” Corin Taylor, Policy Director, Institute of Directors.

In this, my first blog post, I want to warn against falling into the same trap as these auspicious business commentators. I want to get some of Ben Goldacre into all of you.

Let’s take the first point: public sector wages are higher. According to ONS data, the Average Weekly Earnings (AWE) of private sector employees in the first 3 months of this year was £422, compared to £471 in the public sector. Those lazy public servants earn over £2,500 a year more! Case closed.

Hold on now. Surely the correct measure is the AWE including bonuses; we know how much the private sector loves performance-related pay. Including these, the AWE is £490 for the private sector and £471 for the public sector. Moreover, if you remove the “nationalized” banks from the public sector, the public sector AWE falls to £462. So perhaps the private sector earns about £1,500 per annum more.

I’m not arguing that public sector employees earn less, work harder or have better pensions than the private sector. The point is we don’t know, and these comparisons are misleading. The public sector does very different jobs to the private sector, has more females, more professionals and a better-educated workforce. Over recent decades, the lower paid public sector jobs have been outsourced to the private sector.

But the private sector definitely works harder right? On the face of it, yes. ONS data on Annual Survey of Hours & Earnings for 2010 show that, on average, a private sector employee works 2 hours a week more. That’s almost 3 weeks extra per year. Again, however, these averages are misleading. The public sector employs more part time workers, 31% of the workforce compare to 23% in the private sector.

This blog post isn’t a rallying cry for the public sector. My message is simply, be careful of the tyranny of aggregation. Don’t just believe the high-level, aggregate numbers but drill down. This is equally true in business and society.

Consider this example. You’re Lord Sugar and you’ve got two potential new salesmen in the boardroom. Given their past performance, who would you hire?

Ed David
SME Sales Conversions 93% (81/87) 87% (234/270)
MNC Sales Conversions 73% (192/263) 69% (55/80)
Total Sales Conversions 78% (273/350) 83% (289/350)

If you go off their overall, average conversion rate then you’d pick David. However, when considered individually Ed converts both more SME customers and more MNC pitches. This effect is called Simpson’s Paradox and is a form of aggregation bias.

Always drill down to the most appropriate level of analysis. Class over.


1 comment for “Lies, damn lies and statistics

Please leave a comment - we all like them