People are making lots of arguments about this graph in the past few days:
As my friend posted, "Read this and tell me there's not a correlation!" I've written before on this common fallacy, this assumption that correlation equals causation.
There are many examples of the problems of this kind of reasoning, including graphs like this one:
What makes a cause and effect relationship is that there is a logical, not a spurious, connection between the two things. In the real world there are piles or causes and effects all interacting together, so it is unlikely that we can always disentangle these things anyway.
But that logical connection idea is tricky. Perelman argues that we adapt our arguments to the values of our particular audiences. Evans' notion of selective scrutiny, that we ignore given reasoning and simply add our own logic to get to a plausible conclusion seems increasingly true in the arguments on social media about Baltimore.
And since a chart like this simply provides two data lines on the same graphical space, we are asked to provide our own reasoning. This, in argument-speak, is called an enthymeme, an incomplete argument.
The original article from The Atlantic suggested all sorts of reasons for this correlation, including that cities run by Democrats tend to be in the north or on the coasts, where for reasons of geography you can't simply expand and take over new empty lands for new housing. See, we simply backfill the data with speculation that fits our existing ideological biases.
The National Review suggests that this must mean all liberal policies caused these problems: the argument slides down the greased rail of its own abstraction as the word liberal then is adapted to mean anything the critic wants it to mean, regardless of its connection to the original data.
But how do we know what happened first, the housing trouble or the Democrat leadership? The poverty or the liberal policies?
It would be tempting to think that if we just look that up, we would have the answer!
I tried my version of that. Here's what I said on Facebook about it:
Additionally, affordable housing is always a casualty of population density, which is only possible when there is enough economic opportunity to allow for that density. So it moves outside of town, in which case we need mass transit to get the low wage workers into town. Or we need to establish rent controls to cordon off sections of affordable housing. Neither of those are liberal policies, inherently. They sound liberal from the perspective of post-Goldwater conservatism, but they are attempts to solve the problem of housing affordability, which have nothing to do with the liberals in the mayors offices, using what we now call liberal political solutions.
There is a correlation, but it is the opposite of what you are espousing here. A free(ish) housing an development market inevitably increases rents in high-density urban centers. City governments usually like this kind of development and provide tax breaks to encourage it. This is conservative social engineering. Gentrification displaces people. There is an eventual liberal backlash which now has limited means of solving what is always a market problem.
I guess we can blame clean air and environment restrictions, too. But people complained about liberals making laws against throwing raw human waste on the streets from second floor windows 200 years ago, so our definitions of acceptable environmental restrictions shifts. But these are very small things have have little to do with the vast weight of the problem.
I am suggesting that The National Review has the causal order wrong. But this is how we get sucked down into Internet-argument rabbit-holes:
But THAT WOULD MISS THE POINT! :)
I would now have two lines on a graph that I connect with my ideology. One happened before the other, but that fallacy, post hoc, ergo propter hoc, is just as wrong as cum hoc, ergo protper hoc. The time difference doesn't magically make my argument correct.
You might suggest that if I did reallllllly gooood research I could find the TRUTH! And I would have reasoning and evidence that would prove beyond a reas
I kind of wrote a book with some chapters on how to approach that very tricky thing, so, you know, a blog post probably isn't quite enough space to fully answer that question. But, because you will ask, here's a bit of the approach: You've got to complete the argument! No more incomplete enthymemes. And you've got to focus on the warrant (I've explained that before), usually the logic and/or values that connect your evidence to your conclusion.
The focus you want is designed entirely to not let your audience fall into Evans' selective scrutiny. Here's a quick example of that from The Zombie Guide to Public Speaking book: