Generalizing from A Specific
For decades, politicians have used a slick trick to try to support their positions. They’ll say, “Just yesterday, a druggist in Charlotte, North Carolina asked me when we’ll change health care reimbursement, because he’s going to have to close his business this year if things don’t change.”
Even if the story is true (always questionable), it’s a single incident, not representing “the American people,” and could be the result of the pharmacist’s poor business practices or the big pharmacy chain opening down the block.
Consultants come to me with bizarre plans, justifying them with this same practice: generalizing from a specific. “I want to offer a dancing class for chief executives which will combine strategy work with physical fitness,” I’ll be told. And then: “I was talking to a retired small business owner on an airplane, and she told me after a few drinks that it was a great idea, so I think I’m on to something.”
Conversely, someone will complete a highly successful workshop delivery and be depressed because one person in 20 said on a feedback form, “I didn’t like your sense of humor,” or, “What are you thinking, wearing those shoes?”
Never take action or change your behavior based on a single scrap of feedback. (Most unsolicited feedback is meant to benefit the sender, not the recipient.) Never scrap or implement an idea based on a lonely, single sample. And by all means, elicit feedback from people who know how to do what you want to do and have done it successfully and repeatedly.
I’m very confident about this advice, because I ran it by a wrong number on the phone this morning, and he said it made a lot of sense just before he hung up.
© Alan Weiss 2010. All rights reserved.
Bala Subramanian
Process changes should be done only after rigorous testing, no doubt. Simple changes made based on untested hypothesis can and do lead to “out of control” processes and large variations in the end product. Systems thinking and systems analysis can attest to this fact. I suggest that the world build a model of its operational framework and update that model constantly and make it available “live” for the world citizenry to test through their own simulations to understand how various suggestions affect the outcome before making any proposal to change the system. I believe the future of management is in building and using simulation models and techniques, rather than plunging into “a solution” or procrastinating any decision waiting for studies to be completed.
David Ormesher
Bala, maybe you write your own blog and title it “Specifizing from a General.” I have a hunch it will create as many problems as generalizing from a specific.
Alan Weiss
I’m a pretty decent consultant and don’t understand a thing you said, which sounds grandiose and obtuse. We’re talking about people and behavior. It’s that simple and that difficult.
Joseph Ratliff
Ok Alan, from your observations and experience of course, when do you start to determine feedback “statistically relevant” enough to begin examining whether or not to justify or alter a decision?
One bit of feedback isn’t enough, so what level of feedback gets your attention?
Alan Weiss
Observed behavior and evidence. I don’t pay attention to anything other than that when I make decisions in business. Statistics do not interest me. My shortcoming.
Joseph Ratliff
I don’t call it a shortcoming at all Alan (obviously your results speak for themselves).
I think (my opinion) is statistics could comprise a part of the evidence to make a decision though.
Alan Weiss
No argument about that. I just trust what I see and can validate. The amount of that varies with the issue—one case of physical abuse is too many, but losing one’s temper a couple of times a month isn’t all that much to worry about.
David Ormesher
Someone said to me a few weeks ago that the most important people to test an idea with are people who can write you a check. Your receptionist (God bless him/her) or your neighbor or even a consultant will have an opinion, but the the best source of feedback on an idea or performance is a customer. I like that.