Real Chaos, Today – Randomised controlled trials in economics

Real Chaos, Today – Randomised controlled trials in economics(someunpleasant.substack.com)

31 points by stephenflanders 1 year ago | 12 comments

tomrod 1 year ago |

Informationally, this post lands on information regarding randomized controlled trials (RCTs). However, there isn't a specific thesis I can nail down outside of introducing a general audience to the history and philosophy of RCTs. There is some discussion regarding how they aren't a panacea, but its more of an afterthought.

ordu 1 year ago | |

I really liked Judea Pearl take on RCTs in his "The Book of Why". It is not much of the history or philosophy, but he show why RCTs works, why randomness do the trick, he does it from the standpoint of a wider theory of a causation that can explain not just RCT but also other methods of untangling causes and effects, and therefore it becomes easy to him to talk why RCTs may not work.

tomrod 1 year ago | | |

Aye, Pearl's explanation and textbook Causality are fantastic.

thriftwy 1 year ago |

At this point you could just simulate complete economies on a PC.

A medium sized country has 10M people doing ten transactions per day on average. Why have mathematical models when you only ever need to add and subtract.

hnhg 1 year ago | |

Adding and subtracting are insufficient for models that are probabilistic and have interconnected dependencies (graphs).

thriftwy 1 year ago | | |

It may be probabilistic in tx matching but you can simulate every transaction precisely.

Not sure I understand the latter argument.