Tuesday, May 28, 2013

Do You Trust Science?

Or do you trust scientists?

There's a pretty big difference.  I can illustrate it fairly simply: What percentage of research is replicable?

Less than half.  As little as 30%, according to some of Bayer's research.

What percentage of research is downright falsified in follow-up research?

Perhaps 33%, according to the same research.

Given that 5% statistical significance (that is, 5% of all studies -should- be false, according to the statistics used) is the bare minimum for most fields, this means somewhere around 27% of all published research is not only false, but unaccountably false.  Either there's a systematic bias that promotes publishing false papers, -or- scientists are consistently mishandling data in a manner than promotes false positives over false negatives.

Incompetence is one explanation.  The other is that a quarter of all scientific research involves some deliberate fabrication or mishandling of data.

Either way, the odds of something being published in a scientific journal being true is less than 50%.  The best bet is always to disbelieve what you read in a scientific journal; you'll be right more often than wrong.  Meaning that anti-scientific nutjobs are right about research, on average, more often than pro-science nutjobs.

Pretty sad state of affairs, that.

2 comments:

  1. Never attribute to malice that which can be
    explained by ordinary stupidity and incompetence. You DO know what they call the
    guy who graduated dead last in his class at the
    worst medical school in the world????? DOCTOR

    ReplyDelete
  2. a source please

    ReplyDelete