"It is better to be approximately right than to be precisely wrong."People aren't skeptical enough about the data they receive sometimes. What exactly is this need to appear to know things or to draw a certain conclusion even when it isn't backed by solid data? Why does there seems to be a pervasive over-emphasis on 'knowing' even when one can only suspect (but doesn't really know), while the ability to recognize own ignorance or to doubt oneself is somehow regarded with disdain.... as a character flaw. A political candidate who tells the truth and says that he doesn't have a sure answer is likely to lose to a politician who lies and pretends to know the solution (even when everyone, himself included, knows that he doesn't).
- Warren Buffet
Why is this so? What is wrong with admitting to not knowing something for sure because there simply isn't enough credible data to justify an informed answer? What is served, really, when we pretend to know something that we really can't back up with evidence or a solid chain of logic for?
My teenage niece was doing a paper on the conservation of cougar in the wild, for her first college biology class a while back, and I got to sit in the back of the room as she gave her presentation (I was giving her the ride home). Toward the end she was asked about the current state of the cougars' wild population (to see whether the conservation techniques were working). So the dear girl, armed with a vague lay-language fact sheet from a website that states that the number of sightings of the animal 'had increased in the past 10 yrs', confidently proclaimed that the conservation techniques being used are working and the cougar populations are recovering simply because of the growth in the number of reported sighting. And what did her professor do? She smiled and nodded her approval without offering any correction or thought.
Is that reasonable to you? Think about it a bit... Just how good and solid is that data that my niece based her very positive conclusion on? What else, aside from the actual increase in the population size of the cougars, can cause the number of sighting to grow? Is the 'data' even discriminative enough to support any conclusion? The fact is... She doesn't have any idea exactly what sort of data was used to make up the 'number of sightings' that the fact sheet based its conclusion on. If the data says '10 sightings in a week', can she tell if that means that:
A. 10 different people had each seen a different cougar?
B. 10 different people had all seen the same cougar once and reported their sighting separately?
C. one same person had seen the same cougar 10 times and reporting each encounter?
D. one same person saw 10 different cougars in a week?
Or if it was anything else in between?
With that sort of data, the actual number of live cougars sighted can range anywhere from 1 to 10! The only thing you can say for sure is that there is at least one out there being seen and reported on. There is nothing discriminative enough in this piece of data to infer more out of it.
Also... even if the most optimistic presentation of the data is true, what else, aside from the actual increase in the number of the things being sighted, can cause the number of sightings to increase? How about the increase in the number of the people doing the sighting? How about the closer proximity between the populations of the sighted and the sighters? And how about the improvement in reporting percentage (more people who have seen a live cougar in the wild know where and how to report the sighting to the authority) now from before? (this is the same sort of problems associated with the idea that certain diseases like lupus is more wide spread than before... Has there really been an increase in the incidents of lupus or has the modern diagnostic techniques and education enabled the milder cases to be diagnosed that would have been missed before? and is this reflected in the lower fatality rate?)
It is not enough to have a hypothesis and then to search for the data that would back it up. You must also search (even harder) for the data that may contradict your hypothesis!I don't know about the state of the cougar's population, but I do know for sure that the human population in the United States has increased each year. I do know that our towns and cities are expanding, that we're building more roads, and developing further into what used to be the wilderness. Wouldn't we then expect to run into wild animals more as we encroach further into their home range? In fact, the number of live cougars in the wild can be decreasing while the number of sightings increases... as long as the rate of human encroachment on the cougar's home range out-paces the rate of decline in the animal's population.So what is the proper answer to the question that my niece was asked in class? Insufficient evidence to support a conclusion! There is NOTHING wrong with admitting that you don't know the answer when you really DON'T know the answer. When you pretend to know the answer even though you really haven't got a good supporting evidence for it, you have boxed yourself in and closed the door to further investigation.
One must be aware of one's assumptions when one looks at a scientific data! And even when the data is demonstrably good, it isn't enough to seize on the most favorable (to your hypothesis) interpretation of a piece of data. You must never lose sight of the validity of the interpretations that don't fit the result you wanted. The biology professor missed one heck of a teachable moment in class that day. The lesson goes right to the heart of what separates a scientist from someone dabbling on some science stuff. Scientists aren't in the business of gaining absolute certainty. They are in the business of reducing uncertainty. It is imperative that they are intellectually humble enough to know that they don't know absolute truth, and so must work harder to get as close to it as possible.