One Engineer’s Statistical Experience

“Data trumps intuition.” But is intuition the same thing as “engineering judgment” or do different levels of logical control become applied to individuals and are these intended to make examples for the group? If you can prove something with statistics does this satisfy the expectation that it will become true? How can we exist in the dream world of go/no go, in/out, up/down, on/off, guilty/not guilty—opposites, when purportedly every line crossing is explainable, intentional, and irreversible?

*Bias Alert! If we learn…we might be able to guess the reason for study and we might rethink our opinion of the data. “Who is made to hold a smoking gun” should make us suspicious when there is no crime. Knowing the context of the data protects us from misinterpretation.

Lulled into Complicity.

Illustration: Consider that a particular motto “Tough on Flaws” gets adopted by the quality control manager at a plastic bag manufacturing company. Although the bags are remarkably versatile, have excellent capacity, and are long-lasting, nowhere in the product description is the claim “completely waterproof” because in fact the bags do leak. The R&D department is testing different polymer combinations which could maintain existing benefits as well as provide additional benefits (such as increased leak-resistance) but to date these investigations have not been conclusive.

Who? What? Why? If bags are under-filled, the contents only slightly liquid (i.e. mostly solids), or the stress is not repetitive…voilà! The claim is made that the bags are now leak free (albeit no longer simultaneously dry while demonstrating the other desired traits). By changing the usual and accepted definition of goodness the authority has contaminated our value system and we suffer demoralization.

If a sample is filled with fluid to over-capacity and gets repeatedly stressed, the bag fails to endure the trial. When QC or any such power imparts a battery of examinations designed to “expose flaws” and is (thereby) generating proof of claims, the basic measuring system is compromised. Such is the method some charging decisions follow.

Imagine the shock when consumers discover they were duped. The horror is that “bait and switch” tactics produced an end which does not justify the means. Dismay is an understatement of the disappointment that follows exposition of flawed logic. Recalcitrant authorities must set right the disturbed integrity of samples else all will be lost!

Back to our example. If you are perfectly satisfied with outcomes—say the resealable one gallon bag will hold up to one gallon of your produce—and suddenly the bag manufacturer switches their product, claiming “New and Improved—Now Leakproof” then what consequence are you willing to accept? Yes it is new but No it is not improved. Now a ten pound bag won’t hold ten pounds. Its goodness has been diminished.

Rejection of the argument “unacceptable loss” because “the old product was flaw-laden” is not agreeable. Forget the fact you went through learning curves, paid your dues, caused a few mistakes but made a lot of excellent contributions too, and gained efficiencies in areas outside your chosen field of expertise. You yourself are recovering from system vulnerability (it takes 20 pounds of effort to achieve 10 pounds of fit). Advocate for mutual respect.

Causality.

To establish causality, you need to run controlled experiments. There are many books on controlled experiments in multiple domains, including health.

That’s where you should search for related materials.

--Ronny Kohavi Director of Data Mining and Personalization, Amazon.com

If I caused you to trip and you caused the next guy to stumble and they caused the one after them to fall and so on, when does blame enter the scope? Say that I developed an amazing fertilizer which increased the outcome of production ten-fold, a distributor incorporated a binding agent to prevent storage decay making the fertilizer 1/10^th the previous cost (due to reduction in losses), and a farmer applied the fertilizer to a crop which was in remarkably short supply, thereby increasing availability. Furthermore the law of Supply and Demand happened to not keep up with this particular consumable. Now it turns out that all who came within six degrees of contact with the commodity are stricken with cancer. But forget trying to make sense of this toilet-bowl spiral in shame…

Instead say that I recognized an amazing cure for an undeniable trend in the population’s behavior. The authorities said “not interested” but I looked for a way to develop interest anyway. A second batch of rejections came from the establishment. Undaunted I tried again. A third layer of naysayers shot me down. And so on until I could go no further. Finally someone recognizes the goodness of the proposition however by this time changes must be made and modifications are required in order for the agenda to proceed. What caused the population to refuse healing? When does complicity in rejection of solution become involvement for advancing success?

Credit? Blame? Promised improvement does not supersede viable deliverables. The Customer/Supplier chain puts us all in the same bed; if one is guilty then the other is guilty too—we all lose when the population distances itself from compassion and care for the individual! Perhaps the greatest criminals are those who sway public opinion without getting caught.

*I am concerned this article has gotten too philosophical. It is not too intellectual for anyone to recognize the argument that there will never be proof of which came first: chicken or egg. However when present day authorities cite evidential observations as clues to justifiy the advancement of one agenda it necessarily restricts other advancements. The public is known for catching on to whatever version of the truth pleases them. For example, "Let them eat cake" (see http://urbanlegends.about.com/od/dubiousquotes/a/antoinette.htm) could be a misquote which lives on in infamy.