One thing I'd add is that cross-sectional evidence is highly prone (even more than usual) to being misleading in cases where one variable is a deliberate response to the other (feedback).
For example, statins are drugs commonly used by people to prevent strokes, but strokes are more common among people who use…
One thing I'd add is that cross-sectional evidence is highly prone (even more than usual) to being misleading in cases where one variable is a deliberate response to the other (feedback).
For example, statins are drugs commonly used by people to prevent strokes, but strokes are more common among people who use statins. Obviously this does not mean that statins cause stroke, or even that they don't reduce stroke probability.
It's due to the underlying health conditions that the statins are used in the first place and, even if they do help, they are not miracle drugs that make one as healthy as a person who never needed the drugs in the first place.
Similarly, incarceration is a response to crime. So there is nothing surprising about high crime and high incarceration rates co-occurring, even for people who believe that incarceration causally reduces crime.
Other factors than incarceration clearly contribute to the high American homicide rate. But, as you note, what matters is whether crime would be lower or higher in a counter-factual where the incarceration rate was changed and all else held equal.
Exactly right on every point.
One thing I'd add is that cross-sectional evidence is highly prone (even more than usual) to being misleading in cases where one variable is a deliberate response to the other (feedback).
For example, statins are drugs commonly used by people to prevent strokes, but strokes are more common among people who use statins. Obviously this does not mean that statins cause stroke, or even that they don't reduce stroke probability.
It's due to the underlying health conditions that the statins are used in the first place and, even if they do help, they are not miracle drugs that make one as healthy as a person who never needed the drugs in the first place.
Similarly, incarceration is a response to crime. So there is nothing surprising about high crime and high incarceration rates co-occurring, even for people who believe that incarceration causally reduces crime.
Other factors than incarceration clearly contribute to the high American homicide rate. But, as you note, what matters is whether crime would be lower or higher in a counter-factual where the incarceration rate was changed and all else held equal.