Finding the causes of complex problems. is the single, most skilful part of the evidence-based manager’s job.  Asking ‘Why?’ is also a prime motivator of human behaviour.  Many of man’s greatest endeavours are driven by this most primal need – why are we here, what has caused me to feel this way?  If we fail to identify the right causes, using evidence-based analysis, we fail to find the right solutions.   When organisational  demands for activity – ‘do anything but do something!’ – take precedence over intelligent thought and proper analysis then we are bound to see short cuts being taken, however erroneous.

Academia is not put under the same immediate, operational duress so we expect academics to go to the trouble of providing better evidence with more scientifically rigorous methods.  The LSE (London School of Economics and Political Science) even takes its Latin motto ‘Rerum cognoscere causas’ – ‘to understand the causes of things’ – from Virgil’s dictum “happy is he who can discover the causes of things”.

Indeed, great happiness will stem from knowing the causes of what ails us and one LSE academic in particular has followed this line of enquiry – Professor Richard Layard – who is well known for his work on happiness and wellbeing.  These are both matters of great interest to those who have devoted their entire careers to the search for the causes of employee happiness and fulfilment at work. So has Professor Layard followed the LSE’s creed in this endeavour?

Anyone dealing with people issues at work will be reminded, every day, that HR management is not a ‘science’ in any meaningful sense.  It cannot be tested in laboratory conditions and neither does it respond well to conventional, statistical analysis. I am referring of course to correlations and regression analysis.  This is an arcane, statistical technique used by social scientists to pretend that they have identified the causes of complex human problems when, in fact, they have just resorted to stacking one correlation on top of another – which is not the same thing at all.  If this is all a bit too academic for you let me make my point much simpler.

Every HR department I know wants to believe that employee engagement causes organisational performance.  This notion appears obvious and very valid until you consider that the buyer of your office furniture could equally make the same claim about their role.  There is probably as much of a correlation between office furniture costs and profit as there is between engagement and profit but no one in their right mind (apart from the procurement department) would suggest that this denotes a significant, causal relationship with performance.  Evidence-based managers do not rely on correlations that can so easily make them look stupid.

This logic does not stop highly respected academics passing off correlations as causation though.  Take Professor Layard’s own paper  – ‘Good Jobs and Bad Jobs’ where he tells us in “Annex A: Evidence on Happiness” that

“Happiness research has confirmed that happiness is a single dimension of all experience, measurable by psychometric or neurological measures (both highly correlated).”

They may well be highly correlated but that will never reveal what causes what.  Or has the LSE dropped its motto now and replaced it with a much lower standard of evidence – “to find correlations between things”.  Would they be just as happy to pass off a bottle of cheap Cava as Dom Perignon at receptions for their generous alumni?  Probably, if they thought  no one would notice.

Surreptitiously and knowingly substituting correlation for causation is a serious crime and extremely dangerous.  There is a correlation between skin colour and prison population in the US and no doubt this is irresponsibly seized upon, by those with their own agendas, to infer that the colour of someone’s skin can actually cause crime.

This is the prime reason why evidence-based HR is now so high on the organisational agenda – because the development of HR management thinking over the last 30 years has been predicated on spurious correlations (e.g. happy employees are productive employees, diversity targets improve diversity) rather than causation.

There can be no better example of this than the “Best companies to work for” (*but see Update below) schemes. In the UK the image shown above was printed in The Sunday Times supplement announcing the ‘winners’ of this award in 2005.  The graphs purport to show a causal connection between being a ‘best company’ and superior performance, when compared to the FTSE 100.  Except that the small print had to admit that -

“in the last year … the FTSE 100 moved ahead with a 14.3% return against an 8.3% return for Best Companies”.

So not only had they failed to support their causal claim with the evidence on show, the correlation they used had been broken and they continued to refer to those with the worst performance as the ‘Best companies’.  One could infer from the ‘Best Companies’ own data that perhaps it should be called the ‘Most Stupid Companies’ awards?

This is why lies and ‘damn lies’ often masquerade as meaningful statistics. Coming up with a solution first and then concocting the ‘evidence’ (sic) later to justify it will never be able to compete with the professionalism of the evidence-based.  There will only ever be one winner in this competition.

For personal development linked to this topic visit the Consummate Professional Series Module 2 and Module 4

*Update 17th April 2012

This similar chart was taken from the i4cp site. Even if these graphs are true it is highly unlikely that they have anything to do with being (or not being) an i4cp member. Trying to mislead innumerate HR people is a sure sign of a non-evidence-based approach – interestingly Professor John Boudreau is on their Board of Directors.