In God We Trust; all others must bring valid data
Before he retired as CEO of Sara Lee Bakery in 2003, Barry Beracha apparently kept a sign on his desk reading: “In God We
Trust; all others must bring data. Widely attributed to W. Edwards Deming, but never definitively verified, it is probably one of the
most repeatedly parroted phrases in the business transformation and continuous process improvement field, and one that is just
as equally misinterpreted.
The fact that Demming repeatedly eschewed total reliance of management decision-making to the blind-eye, button-counting
approach to statistical probability for which he is so often credited, is ironical, to say the least.
To quickly build up to a counter-point, let me use another quote whose authorship is also in question, and often incorrectly
attributed to Mark Twain. Twain himself attributes the quote to Benjamin Disraeli. However, just where the original quote is to
be actually found has still not been firmly established in the passing century that has followed since its first incontrovertibly
appearance in the pages of the North American Review (No. DCXVIII, July 5, 1907), and its subsequent, almost mantra popular
The Twain version of the quote, sadly, is often presented in truncated form -- presumably out of laziness, inexactitude or some
other hidden agenda that individually is now shrouded by deliberate coyness and time -- so that the richness of Twain’s own full
meaning is invariably obscured or subverted:
“Figures often beguile me, particularly when I have the arranging of them myself; in which case the remark attributed to
Disraeli would often apply with justice and force:
‘There are three kinds of lies: lies, damned lies, and statistics’."
As Twain cleverly suggests: statistics don’t lie, people do.
That the sly argument that Twain himself was attempting to make has been blithely sabotaged over the passing hundred years
would have surely appealed to the author's well-established sense of understated mockery.
Just prior to his death in 1998, I had the extraordinary good fortune to work professionally with my father. He had recently retired
as a professor from the University of Wisconsin School of Social Work, and was noodling around in dignified semi-retirement
doing research projects on whatever caught his fancy. I had just recently made the jump to full-time self-employment as a
consultant, and had been approached to do a small study for LOMA (formerly the Life Office Management Association), a trade
association for the insurance and financial services industries.
At that point in my life, my father and I had been estranged for over three decades.
From a business perspective, I needed someone with deep, incontrovertible professional experience in statistical methodology
and statistical instruments, who would be willing to work on the cheap as a statistical consultant to my company. Without it, my
own glaring gap in actual statistical knowledge and research experience would have surely taken us out of contention for landing
the contract we so desperately wanted. I suppose, my father needed a final shot at somehow repairing the stormy and now
shattered relationship with his eldest son.
We never discussed it. It was a straight-forward business deal that turned out to be oddly mutually-beneficial.
In any event, my father agreed to the business relationship, at an insanely low hourly rate that I would have gladly doubled, and
the contract was -- against all probability -- won.
Therein, began my true statistical and methods education, not necessarily in statistical knowledge and interpretation, but rather
in professional standards and instrument validation, well beyond the realm of books and theory.
We began our professional association by discussing the study's methodology and survey instrument. As the junior partner of a
two-man questionnaire writing team, and to mitigate billable hours, it was suggested that I would write the first draft of the LOMA
questionnaire for Professor Lefcowitz’ initial review. I must have rewritten that damn questionnaire at least a half-dozen times
before my father was finally satisfied that it was good enough for testing.
Testing? Dear God, we have a deadline. Testing, I was firmly told.
In a series of interviews with local executives in the industry, the draft instrument was tested and retested, so that its reliability,
its internal and external validity, its sensitivity, and its specificity were as certain as it could possibly be made on a tight budget,
with little wiggle room.
In the last analysis, I am absolutely certain that the Professor was not completely satisfied with either the instrument or its
ultimate delivery method. But we were working within the constraints of the contract, and the customer’s own non-negotiable
As for me, I was just glad I had the opportunity to test my own hard-won skill sets against the standards of a true business
professional, and somehow successfully -- but painfully -- held down my own end of the log in the doing of it.
I learned a lot of personal, professional, and business lessons in a relatively short period of time, for which I am eternally grateful
for the very good fortune that was so undeserved. Of these, with all due respect to W. Edwards Deming and his minions,
perhaps the most important is:
In God We Trust; all others must bring valid data.
© Mark Lefcowitz 2014 -- 2015
All Rights Reserved