Lies, Damn Lies and Evidence – @deeveebee @bengoldacre @senseaboutsci #askforevidence

I’m one of those advocating caution alongside the otherwise laudable Sense About Science Ask For Evidence campaign. You can have too much of a good thing.

Getting overly focussed on seemingly objective evidence is OK so long as we understand what really counts as evidence when it gets communicated transparently (if it’s really intelligible to us) or is presented as newsworthy (if it’s mediated for us). You can never escape some element of trust, dare I say faith, in your sources and channels. There can be no shortage of conspiracy theories, but even the well-intentioned can accidentally mislead and a meme is a meme once it’s off and running.

Several interesting pieces recently.

Beware (crusade against!) multiple regression analyses. Look out for self-selection effects in correlations chosen for possible causal analysis. The psychology may be a bigger factor than the arithmetic. The downside to transparency. The emotional impact of misleading news. And, where there is intention to mislead, even well-intentioned white-lies or ironic cruelty-to-be-kind, it gets all the more complicated. The knowledge deficit model – Comment is free, but … some things are sacred.

====

Post Note :

At the extreme ends of science – the fundamentals and the massively complex – objective evidence is even more precious, and cognitive bias by scientists and their social circles even more of a “crisis”. Here Sabine Hossenfelder talking on the risks to objective evidence when it is hardest to come by and therefore matters most in theory assessment. Several points where I differ with Sabine – the crisis is not so much about the pace of scientific progress, more the opposite, the increasing risk of scientific regress. Weinberg is right on beauty, simplicity and elegance. They are not fundamentally aesthetic when used by a thoughtful scientist – they are merely shorthand for a lot of experience – but they are nevertheless not objectively or fundamentally tested axioms. But she’s right. Consistency is indeed overrated when you are lost in the maths – it’s self-reinforcing. And multiverses can be a hack to cover up the lack of constraining axioms which make anything possible.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.