"Follow the science" is the premier attention-grabbing meme used by popular media to inspire confidence in whatever is presented next. The phrase amounts to a slightly more nuanced version of the renowned "evidenced-based science" appellation. How might those characterizations affect you and your decisions? You probably do sit-up and take notice, especially if the information promotes your preexisting confirmation bias. But perhaps that is understandable, since, after all, the media reports proceed from "science" and/or "evidence-based science."
Let's, for this posting, restrict our considerations to research that is reasonable science and that truly is evidence-based. I specify that because much widely disseminated media-promoted research eventually is revealed to be illegitimate research. For instance, many of us vividly remember reading or hearing about bogus Cornell University research suggesting that you will eat less food if it is served on a smaller vs a larger plate. If you are benevolent enough to forgive that single mistake by Professor Brian Wansink, the responsible researcher and director off Cornell’s Food and Brand Lab, good for you. But what if you learn-- as is true -- that after the aforementioned revelations, Dr. Waansink retracted a total of 15 study results and then voluntarily retired from Cornell.
Okay, okay, but how about other meticulously structured research guided by well-vetted computerized numerical algorithms? They must be reliable and valid. Well, not necessarily.
Since I am neither a computer programmer nor a mathematician, I defer to and quote Dr .David A. W. Soergel’s “Rampant software errors may undermine scientific results.” (2014)
“… people show a level of trust in computed outputs that is completely at odds with the reality that nearly zero provably error-free computer programs have ever been written …even the most careful software engineering practices in industry rarely achieve an error rate better than 1 per 1000 lines. Since software programs commonly have many thousands of lines of code, it follows that many defects remain in delivered code–even after all testing and debugging is complete.”
Soergel then gives numerical examples and states” Multiplying these, we expect that two errors changed the output of this program run, so the probability of a wrong output is effectively 100%. All bets are off regarding scientific conclusions drawn from such an analysis. `”
I conclude that we must never accept any single study, especially one that reinforces our preconceptions. What do you think?
No comments:
Post a Comment