Tuesday, October 18, 2016

Swapping Cause and Effect

It is interesting sometimes to read the websites of self-proclaimed skeptics, such as CSICOP. While they often have a host of apolitical, reasonable arguments, they also seem prone to dabbling in politics, and it seems inevitably when they do so, they veer hard to the left. From the assumption skeptics must be atheists, to the belief that anyone who does not accept every premise of the latest AGW theory is a "denier" and conspiracy theorist, they seem strangely prone to left wing beliefs. (I noted this before in "My Irritation with Supposed Skeptics", "A Thought on Intelligence" and "A Bit Disappointed in CSICOP - The Difference Between God and UFOs".) Perhaps it has something to do with the popular, yet mistaken, belief that "smart people are liberals" (cf "Intellect and Politics", "A Thought on Intelligence") or maybe the tendency for academics to lean left. (cf "Some Thoughts on the Media", "The Press Versus the Nation", "The Path of Least Resistance") Whatever the reason, it is a strange phenomenon, yet one that I think would be obvious to any but the most partisan readers, for some reason, skeptics tend to unquestioning acceptance of left wing premises, often failing to view them with the same skepticism they would other beliefs1,2.

I mention this because I sometimes forget how far left skeptics tend to skew. Yes, I know they are completely in the most extreme AGW camp, but other than that, I often forget their political bias. And thus I am sometimes surprised when, reading through skeptical sites, I come upon evidence of skeptics completely accepting at face value the most dubious partisan hogwash without a second thought. As, for example, CSICOP accepting an essay on guns and the risk of homicide, which completely fails to take into account a host of confounding factors3.

The study in question made two claims which, though superficially damning, can actually be explained quite simply by some common sense argument.

First, there is the statement that suicide is more common among household with firearms. The assumption behind this, I suppose, is that firearms, by making it so easy to kill oneself, make it more likely someone will successfully kill himself rather than give time for others to talk him out of it. But this seems somewhat implausible. After all, poisons are also commonly available in most homes, as are knives and other sharp objects. Granted, knife wounds and poisons can be treated more easily than gunshots, but they still require someone to find and identify the suicide as such, and get help in time. This seems unlikely, as most suicides do not commit the act while others are present.

Is it not far more likely this is confusing, at least in a number of cases, cause and effect. That those who are contemplating suicide are more likely to purchase a firearm, making the number of those committing suicide who also own firearms higher? After all, despite the claims of the article that suicide is a spontaneous and impulsive act, many who commit suicide have tried more than once, and a large number have planned the deed in advance. So is it that unlikely they would buy a gun so that they would be sure of a quick demise?

Well, whatever one thinks of that argument, the other one is even less plausible. That is the claim that somehow guns cause homicides. That is, the claim that households with guns are also more likely to suffer a murder.

Now, this one is obviously a case where cause and effect can easily been switched. First, and most obviously, people who own guns for nefarious reasons are also often those who get murdered with guns. So, ignoring anything else, criminals who own guns dying of homicide is hardly a surprise. Second, and only slightly less obvious, many people who feel threatened, either because someone is threatening them, or because they live in a dangerous area or travel to dangerous places or work in a risky field, would also be more likely to both buy guns and suffer a death by homicide. In short, is it not possible that guns do not increase the risk of homicide, but rather those at an elevated risk of being a victim of homicide are also more likely to purchase guns?

These are the sort of questions one would expect skeptics to ask. However, oddly enough, they seem to be the ones that are rarely if ever even considered by professional skeptics. Instead, they present a very partisan picture, accepting the study at face value, never asking questions that are terribly obvious.

So much for self proclaimed skeptics.


1. You can see this in part by looking at the credentials of writers on skeptic sites. A surprising number come from openly partisan journals, such as Mother Jones.

2. A perfect example is their attachment to atheism. As I have argued in "Atheism's Circular Reasoning", "Is The Flying Spaghetti Monster From Canada?", "Materialist Arrogance" and "A Bit Disappointed in CSICOP - The Difference Between God and UFOs", there is not rational proof for or against religion, and thus, if one is being purely rational, religion is a non-issue. You can no more disprove God than prove him, and thus, strictly speaking, there should be no official skeptical position on religious belief as such. Yes, one can dispute doctrinal points, such as young earth beliefs, or creationism, but not the belief in God itself. However, despite that, skeptics tend, by and large, to assume every skeptic must be an atheist.

3. This is hardly unusual in politicized science. Many times I have seen a study on the latest health scare which either fails to take into account autocorrelation (eg the fact smokers also tend toward other behaviors which can exacerbate health effects) or which use sample groups which are not comparable (eg the US lifespan effect of trying to save premature children rather than writing them off as stillborn, giving a number of infantile deaths which are nothing but a statistical artifact). Other than exaggerating the significance of small statistical changes, this seems to be one of the most common ways to skew data.

No comments:

Post a Comment