1) How often does correlation equal causation? One neat example is Wilde and Hollister (2002): In four of eleven non-experimental studies, the sign is the opposite of the true effect. This seems like an extremely important topic. Suppose you come across some econ study that shows a clear correlation between X and Y and includes all the controls you can think of, other than
Plan for the UK: Government is exceptionally bad at picking winners and losers. Under no circumstances should government be encouraged to set up an investment fund for new technologies. The applicants would have to pass the government sniff test to get the funding, which is almost a guarantee that the technologies are failures, deep holes into which endless amounts of money can be poured with nothing coming out. Government can however create environmental incentives for private capital to fund these ventures. VCs tend to be a little more hard-nosed. The UK has perhaps one-tenth the research lab capacity of the US, entirely owing to idiot stuff like building regulations, council taxes, government inspectors, seven year wait to get approval, that disincentivize doing research.
On 2 (the inequality analysis), a closer examination shower that Auten-Splinter used a very aggressive assumption regarding unreported business income (basically evenly distributed suggesting that poorer individuals people have major unreported business income). This heavily skews the result downward. Generally, however, inequality has increased but I would also add the question whether the levels of inequality in 1960 was a 'good' level i.e. the question should be not only is inequality rising/falling but what should be the optimal level of inequality.
On the Swedish lottery and crime, I saw some econ discussions suggesting the $20,000 win is not a significantly large number to alter behavior.
Maybe you are centre-right adjacent? I would be interested to know how strong a correlational study needs to be to update your priors. I suppose at one end of the spectrum is Sir Richard Doll and Austin Bradford's 1950 BMJ study, which found that smokers were 50 times more likely to develop lung cancer than non-smokers. But what about Nate Silver's post suggesting that vaccines are much more effective than lockdowns in the US? I certainly thought Republican states would have massively higher excess deaths pre-vaccines. What was hte last correlational study that caused you to significantly update your priors?
Plan for the UK: Government is exceptionally bad at picking winners and losers. Under no circumstances should government be encouraged to set up an investment fund for new technologies. The applicants would have to pass the government sniff test to get the funding, which is almost a guarantee that the technologies are failures, deep holes into which endless amounts of money can be poured with nothing coming out. Government can however create environmental incentives for private capital to fund these ventures. VCs tend to be a little more hard-nosed. The UK has perhaps one-tenth the research lab capacity of the US, entirely owing to idiot stuff like building regulations, council taxes, government inspectors, seven year wait to get approval, that disincentivize doing research.
On 2 (the inequality analysis), a closer examination shower that Auten-Splinter used a very aggressive assumption regarding unreported business income (basically evenly distributed suggesting that poorer individuals people have major unreported business income). This heavily skews the result downward. Generally, however, inequality has increased but I would also add the question whether the levels of inequality in 1960 was a 'good' level i.e. the question should be not only is inequality rising/falling but what should be the optimal level of inequality.
On the Swedish lottery and crime, I saw some econ discussions suggesting the $20,000 win is not a significantly large number to alter behavior.
Thank you for the digest! Happy New Year!
Maybe you are centre-right adjacent? I would be interested to know how strong a correlational study needs to be to update your priors. I suppose at one end of the spectrum is Sir Richard Doll and Austin Bradford's 1950 BMJ study, which found that smokers were 50 times more likely to develop lung cancer than non-smokers. But what about Nate Silver's post suggesting that vaccines are much more effective than lockdowns in the US? I certainly thought Republican states would have massively higher excess deaths pre-vaccines. What was hte last correlational study that caused you to significantly update your priors?