Dakota Beacon

The scientific method is just that--a method. And methods can be corrupted by people.

Today's most notorious illustration of junk science is the human-caused global warming hoax. However, even scientists who genuinely want to perform objective research are capable of practicing flawed science.

In an article covering the rising incidence of flawed science, Wired magazine contributing editor Jonah Lehrer notes that "all sorts of well-established, multiply confirmed findings have started to look increasingly uncertain. ...claims that have been enshrined in textbooks are suddenly unprovable."

One of Lehrer's examples is the story of Jonathan Schooler. In the 1980s, the psychologist thought that he'd showed that people who were required to describe faces are actually less able to later recognize the faces than people who'd simply seen the faces without having to go through the description process. Schooler called the phenomenon "verbal overshadowing."

But repeating these findings grew increasingly unachievable, even by Schooler himself. Psychologist Joseph Rhine documented similar non-reproducibility in the 1930s while investigating extrasensory perception. Rhine labeled the problem the "decline effect." Lehrer suggests that a phenomenon known as "regression to the mean" causes the effect.

Here's how regression works.

Conclusions drawn from small amounts of data stand high chances of being wrong. As more data is taken, truer pictures emerge and accurate conclusions can be drawn.

A flawed dice-throwing experiment illustrates.

An investigator rolls two dice six times. Three rolls each produce a "2," the other three each produce a "12." The investigator concludes that the most likely numbers obtainable on rolling two dice are 2 and 12. The investigator is flat-out wrong. Insufficient data collection is to blame.

Anyone who's played craps knows that the result obtained most often when repeatedly rolling two dice is the number "7." In fact, 2 and 12 are the least likely results. Six possible combinations of the roll of two dice yield the number "7." Only one combination each produces 2 or 12.

So then, regression to the mean explains the Schooler and Rhine problems?

No.

I don't buy the explanation. (Neither does Lehrer, who goes on to raise other possibilities which make his article worth a read.)

My hypothesis as to the decades-old degradation in scientific study requires recognizing a fundamental difference between something as straightforward as shooting dice and observing human behavior or investigating global climate. That difference lies in the vastly disproportionate potentials for investigator bias.

Rhine's "decline effect" afflicts Jonathan Schooler and also afflicted Rhine himself. Rather than explaining away non-reproducibility with a cause as convenient as lack of data, we must look to human nature to understand the untrustworthiness that permeates today's "science."

The instant that an investigator assigns a theory to a phenomenon, that investigator's thinking becomes subconsciously confined within self-imposed boundaries. It's a fact of life that Mother Nature wired Homo sapiens to view the world through subjective lenses.

Scientists, mere Homo sapiens after all, instinctively see what they want to see. Schooler wanted to debunk the "describing improves memory" axiom. In his early experiments, Schooler probably inadvertently made it comparatively easy for test subjects who were not required to describe faces to recall those faces. Then after the work became repetitive and interest waned, his subsequent experiments were designed and conducted in a detached manner--in other words, objectively. Schooler has therefore never duplicated his early "success."

One way to test my hypothesis--and apply the test to any study, really--would be to have investigators watch the investigators. But the extra investigators would also be members of the species Homo sapiens. So wouldn't the added layer of observers simply increase the likelihood of the "decline effect?"

Maybe we could minimize the influence of human bias by using, as our additional observers, utterly disinterested people who are unfamiliar with the field of study in which the experiment is being conducted. We should keep our new observers in the dark as to study's hypothesis.

Ah, but someone would still have to interpret what the observers observe. Again we'd have to deal with human bias.

What a tangled web humans weave when humans practice science.

I shall continue this discussion in an upcoming blog post in which I also tie in the great global warming scam.

Click HERE to receive all posts by email FREE

© 2011 Chuck Rogér