Welcome back to "Doggerel," where I ramble on about words and phrases that are misused, abused, or just plain meaningless.One of the words that'll always come up, especially if there's anything tangentally related to political parties, is "biased," as if that alone defeats the opponent. It's usually an irrelevant ad homenim when it comes to science and logic: The characteristics of an arguer are irrelevant to the argument. That makes it one of the many predictable subject changes woos are so fond of.
Everyone is biased. I'm biased. You're biased. Molly is biased. Despite common opinion, Lt. Commander Data is biased. If bias alone was enough to dismiss an argument, you might as well just head right over to Subjectivism Land on your way to solipsism. What matters is the data. (With a lower case 'd', not the fictional android.)
One of the wonderful things about science is that it strives to eliminate bias. Good experiments are designed to produce unambiguous results. Double-blinding prevents people from forcing their expectations onto the experiment. If you want your medicine to pass the clinical trial, you're either too distant from the experiment to influence the result, or if you're involved, you're kept in the dark about which subjects to influence.
That's the sort of thing science does. If you want to dismiss a study, you have to point out how bias could have slipped in and altered the numbers. That's what the peer-review process is about. What's more is that dissenters usually have access to the experimental protocols so that they can replicate them and see them for themselves. Shy of the administrative nightmare of a world-spanning conspiracy out to falsify the results, I don't see many ways for bias to slip into the process.
4 comments:
"Reality has a well-known liberal bias." - Colbert
Your post just can't be complete without that quote.
I'm biased to think I'm not biased...
I do know that it's important for the power output tubes on a Fender Amp to be properly biased.
But then I could just be biased.
There are ways for bias to slip into even peer reviewed studies. Mark Hoofnagle over at denialism blog is fond of citing the study "Why most published research findings are false", in which he finds that bias can and does affect certain findings, even in peer review. http://medicine.plosjournals.org/perlserv/?request=get-document&doi=10.1371%2Fjournal.pmed.0020124
The mistake, however, would be to dismiss peer reviewed literature in toto based on the mere possibility of bias. That's a hasty generalization fallacy. Though bias is still possible in peer reviewed studies, the chances of bias affecting the results are still far less than in non peer reviewed literature. Peer review is good at reducing bias, but nothing is 100% effective, of course. Peer review is an excellent tool, but it's not a panacea.
Post a Comment