A Magical Journey through the Land of Logical Fallacies, Part 1

by Brian Dunning
November 6, 2007

If you've ever had a conversation with anyone about their supernatural or pseudoscientific beliefs, you've almost certainly been slapped in the face with a logical fallacy or two. Non-scientific belief systems cannot be defended or supported by the scientific method, by definition, and so their advocates turn elsewhere for their support. In this episode, we're going to examine a whole bunch of the most common logical fallacies that you hear in reference to various pseudosciences. When you hear one that you recognize, be sure to wave and say hello.

Let's begin with:

The Straw Man Argument

We're starting with this one because it's the most common and also one of the easiest to spot. This is where you state your position, and your opponent replies not to what you said, but to an exaggerated and distorted caricature of what you said that's obviously harder to defend.

Starling says: "People who commit minor offenses should be let out of jail sooner."
Bombo replies: "Emptying out all the jails would create havoc in society."

Well, maybe Bombo's right, but that's not relevant, because "emptying the jails" is not what Starling advocated. In fact Bombo did not refute Starling's point at all — he invented a different point that was easier to argue against. He created a straw man — one of those dummies stuffed with straw that soldiers use for bayonet practice. It's too weak to fight back. And Bombo can then take satisfaction in having made a point that no reasonable person would argue with, and he appears to have successfully defeated Starling's argument, when in fact he dodged it.

Ad Hominem

From the latin for "to the person", an ad hominem is an attack against the arguer rather than the argument. This doesn't mean that you simply call the person a jerk; rather, it means that you use some weakness or characteristic of the arguer to imply a weakness of the argument.

Starling: "I think Volvos are fine automobiles."
Bombo: "Of course you'd say that; you're from Sweden."

Starling's Swedish heritage has nothing to do with the quality of Volvo automobiles, so Bombo's is an attempt to change the subject and is an avoidance of the issue at hand. Bombo is trying to imply that Starling's Swedish heritage biases, and thus invalidates, his statement. In fact, one thing has nothing to do with the other. Ad hominem arguments try to point out fault with the arguer, instead of with the argument.

Appeal to Authority

This type of argument refers to a special authoritative source as validation for the claim being made. Every time you see an advertisement featuring someone wearing a white lab coat, or telling you what 4 out of 5 dentists surveyed said, you're seeing an appeal to authority.

"Acupuncture is valid because it's based on centuries-old Chinese knowledge."
"This article in a peer-reviewed scientific journal says that people are getting fatter."
"A growing number of scientists say that evolution is too improbable."
"Wired Magazine says that Skeptoid is an awesome podcast."

An appeal to authority is the opposite of an ad hominem attack, because here we are referring to some positive characteristic of the source, such as its perceived authority, as support for the argument. But a good authority supports a position because that position has been shown to be otherwise justified or evidenced, not the other way around. If you say that scientists support Theory X, are those scientists claiming that Theory X is true because they believe it? No, good scientists attach no significance at all to their own authority. Theory X needs to stand on its own; an appeal to authority does not provide any useful support.

Special Pleading

An argument by special pleading states that the justification for some claim is on a higher level of knowledge than your opponent can comprehend, and thus he is not qualified to argue against it. The most common case of special pleading refers to God's will, stating that we are not qualified to understand his reasons for doing whatever he does. Special pleadings grant a sort of get-out-of-jail-free exemption to whatever higher power lies behind a claim:

Starling: "Homeopathy should be tested with clinical trials."
Bombo: "Clinical trials are not adequate to test the true nature of homeopathy."

No matter what Starling says, Bombo can claim that there is knowledge outside of Starling's experience or at a level that Starling cannot comprehend, and the argument is therefore ended. Bombo might also point out that Starling lacks some professional qualification to discuss the topic, thus placing the topic out of Starling's reach.

Bombo: "You're not a trained homeopath, so you shouldn't be expected to understand it."

A special pleading makes no attempt to address the opponent's point, it is just another diversionary tactic.

Anecdotal Evidence

One of the most common ways to support just about any non-evidence based phenomenon is through the fallacious misuse of anecdotal evidence. Anecdotal evidence is information that cannot be tested scientifically. In practice this usually refers to personal testimonials and verbal reports. Anecdotal evidence often sounds compelling because it can be more personal and captivating than cold, uninteresting factual evidence.

Anecdotal evidence is not completely useless. You could say "We saw the Bigfoot corpse at such a location", and if that information helps with the recovery of an actual body, then the anecdotal evidence was of tremendous value. But, note that it's the Bigfoot corpse itself that comprises scientific evidence, not the story of where it was seen.

"I know for a fact that ghosts exist. My friend, who is a very reliable person, has seen ghosts on many occasions."

Anecdotal evidence is great for suggesting new directions in research, but by itself it is not evidence. When it is presented as evidence or in place of evidence, you have very good reason to be skeptical.

Observational Selection

Observational selection is the process of keeping the sample of data that agrees with your premise, and ignoring the sample of data that does not. Observational selection is the fallacy behind such phenomena as the Bible Code, psychic readings, the Global Consciousness Project, and faith healing. Observational selection is also a tool used by pollsters to produce desired survey results, by surveying only people who are predisposed to answer the poll the way the pollster wants.

Bombo: "The face of Satan is clearly visible in the smoke billowing from the World Trade Center."
Starling: "And in one of the other 950,000 frames of film, the smoke looks like J. Edgar Hoover; in another, it looks like a Windows XP icon; and in another it looks like a map of Paris."

Remember that one out of every million samples of anything is an incredible one-in-a-million rarity. This is a mere inevitability, but if observational selection compels you to ignore the other 999,999 samples, you're very easily impressed.

Appeal to Ignorance

Argumentum ad ignorantiam considers ignorance of something to be evidence that it does not exist. If I do not understand the mechanism of the Big Bang, that proves that there is no knowledge that supports it as a possibility and it therefore did not happen. Anything that is insufficiently explained or insufficiently understood is thus impossible.

Starling: "It is amazing that life arose through the fortuitous formation of amino acids in the primordial goo."
Bombo: "A little too amazing. I can't imagine how such a thing could happen; creationism is the only possibility."

Using the absence of evidence as evidence of absence is a common appeal to ignorance. People who believe the Phoenix Lights could not have been simple flares generally don't understand, or won't listen to, the thorough evidence of that. Their glib layman's understanding of what a flare might look like is inconsistent with their interpretation of the photographs, so they use an appeal to ignorance as proof that flares were not the cause.

Non-Sequitur

From the Latin for "It does not follow", a non-sequitur is an obvious and stupid attempt to justify one claim using an irrelevant premise. Non-sequiturs work by starting with a reasonable sounding premise that it's hoped you will agree with, and attaching it (like a rider to a bill in Congress) to a conclusion that has nothing to do with it. The sentence is phrased in such a way to make it sound like you have to accept both or neither:

"Corporations are evil, thus acupuncture is good."
"The government is evil, thus UFOs are alien spacecraft."
"Allah is great, thus all Christians should be killed."

When we do science, it takes more than simply connecting two phrases with the word "thus" to draw a valid relationship. Thus, non-sequiturs are not valid devices to prove a point scientifically.

Post Hoc

The idea that some event must have been caused by a given earlier event, simply because it happened later, is post hoc ergo propter hoc — "It happened later so it was caused by". The assumption of cause and effect is the type of pattern that our brains are hardwired to find, and so we find them everywhere. He took a homeopathic remedy, and his cancer was cured — one happened after the other, and so the faulty assumption is that the homeopathy caused the remission.

Starling: "I bought this car from you, and the heater is broken."
Bombo: "It worked before you bought it, so you must have broken it yourself."

Bombo sees that the breakage happened after Starling made the purchase, so he assumes that one caused the other. In fact there are no grounds for such a correlation. Combined with observational selection, faulty post hoc assumptions account almost entirely for the proliferation of alternative therapies and widespread belief in psychic powers.

Confusion of Correlation and Causation

Closely related to post hoc, but a little bit different, is the confusion of correlation and causation. Post hoc assumptions do not necessarily include any correlation between the two observations. When there is a correlation, but still no valid causation, we have a more convincing confusion.

Starling: "Chinese people eat a lot of rice."
Bombo: "Therefore the consumption of rice must cause black hair."

Due to the nature of Chinese agriculture, there is indeed a worldwide correlation between rice consumption and hair color. This is a perfect example of how causation can be invalidly inferred from a simple correlation.

Slippery Slope

A slippery slope argument presumes that some change will inevitably result in extreme exaggerated consequences. If I give you a cookie now, you'll expect a cookie every five minutes, so I shouldn't give you a cookie.

Starling: "It should be illegal to sell alternative therapies that don't work."
Bombo: "If that happened, any minority group could make it illegal to sell anything they don't happen to like."

No matter what Starling suggests, multiplying it by ten or a hundred is probably a poor proposition. Bombo can use a slippery slope argument to exaggerate any suggestion Starling makes into a recipe for disaster.

The slippery slope is probably the most common subset of the larger fallacy, argument from adverse consequences, which is the practice of inventing almost any dire consequences to your opponent's argument:

Starling: "They should remove 'Under God' from the Pledge of Allegiance."
Bombo: "If that happened, all hell would break loose. Students would have sex in the hallways, school shootings would skyrocket, and we would become a nation of Satan worshippers."

That's enough for one day. Any more than this at one sitting would turn anyone into a quivering lump of irrational jelly, just like the one that first took shape in the primordial goo. Next week we'll continue and wrap up our list of logical fallacies. Until then, digest all of these that your system won't reject.

By Brian Dunning
Follow @BrianDunning