Saturday, September 28, 2013

Beer and Big Bad Data

We all know data is what you make of it.  For example, year-to-date traffic deaths in Illinois shows that where restraints (seat belts) were used properly there were 215 deaths, whereas where restraints were not used, or not used properly there were 176 deaths.  From the data, you could conclude that wearing a seat belt makes you more likely to die in a car accident than not wearing one.

I attended a talk on Big Data earlier this week.  During the networking prior to the talk I was chatting with a gentleman who works for a consulting company that does organizational development work, which is something I do as well.  We got to talking about the value of strong personal relationships at work and he scoffed at the idea and said “We actually have data that shows having a best friend at work increases safety risks.”    This was surprising and I was curious how you go from correlation to causation in this case.  He said, “It’s probably because when its a best friend they’re probably less likely to point out a safety oversight.”  I thought to myself, ‘Better not become too friendly with this fellow.’

What was the data?  It turns out the data comes from a single source, a large brewery in the UK.  The data showed that delivery truck drivers were more accident prone when they scored higher on the question: “I have a best friend at work”.  Now you can imagine the situation.  “At work” could include all the pubs where the drivers deliver their load.  A very friendly driver might well be friends with quite a few pub managers.  Now what do you suppose is likely to happen when they do get together once or twice a week?  Might they have a pint?  You don’t think brewery drivers might drink while making deliveries?  Read this snippet from the Telegraph newspaper just a couple of years ago:

“Supplies were hit on Friday after angry warehouse staff walked out on Thursday following a Carlsberg decision to end a custom giving workers three bottled beers a day on the house.  Denmark's most famous brewery has implemented a new workplace alcohol policy that limits employees to one bottle of beer a day at lunchtime.
Carlsberg's distribution drivers are still allowed to drink three beers during the course of their shifts.

Don’t start crying in your beer for the poor Carlsberg warehouse staff.  There’s a bit of a silver lining.  The article went on:

Jens Bekke, Carlsberg's head of communications, tried to reassure workers. "We 
did remove the extra beers from all refrigerators around the worksite," he said.
"But we have also set up taps from which staff can drink freely. So the employees can certainly manage to drink more than one beer during their lunchtime breaks."

The point is, we can take data and look at it in ways that serve our purposes.  I suppose we are all guilty of this.  It’s not a new discovery of course; the Bhagavad Gita (400 BCE – 400 CE) includes the following: “...the skilled theologian can wrest from any scripture that which will serve his purpose.”  

So the only question remaining is: “How does one become a brewery delivery driver?...”