Free Courses › Logical Reasoning

Read arguments like a lawyer · Lesson 4 of 14

The flaw patterns behind most wrong answers

After this lesson, you can name the specific flaw in a Logical Reasoning argument by matching it to one of 10 common patterns.

9 min video, plus notes and 1 practice question below

In this lesson

• Recognize a sampling flaw when a conclusion rests on a sample that's too small, too biased, or unqualified to judge the claim
• Spot ad hominem attacks that criticize the arguer instead of addressing the argument
• Tell a necessary condition apart from a sufficient one so you don't mistake a requirement for a guarantee
• Catch circular reasoning and equivocation, where an argument restates itself or shifts a word's meaning midway through
• Identify correlation-causation errors and absence-of-evidence flaws that mistake timing or silence for proof

Bad samples and attacks that dodge the argument

A sampling flaw shows up when an argument draws a big conclusion from a small or biased group. Picture this: someone asks 5 people at a coffee shop whether exercise is the most important part of staying healthy. All 5 agree, so the argument concludes exercise must be the key to good health. That's flawed 3 times over. 5 people is too small a sample, people at a gym already lean toward valuing exercise, and none of them are necessarily qualified to judge what matters most for health. A doctor's opinion would carry more weight than an informal poll at a coffee shop.

An ad hominem flaw attacks the person making an argument instead of the argument itself. If someone says, "don't believe Alex's argument about recycling, Alex doesn't even recycle at home," that's an ad hominem. Whether Alex personally recycles has no bearing on whether Alex's argument for recycling is logically sound. You can call Alex a hypocrite, but that criticism never touches the actual reasoning.

Confusing a requirement with a guarantee, then just repeating yourself

Confusing necessary and sufficient conditions means treating something required for an outcome as if it guarantees that outcome. Example: "if someone is a licensed doctor, they must have completed medical school. Chris completed medical school, so Chris must be a licensed doctor." Finishing medical school is necessary to become a doctor, but it doesn't guarantee licensure. Chris still needs to pass board exams and meet other requirements. Watch for arguments that treat a requirement as if it were a guarantee.

Circular reasoning happens when the conclusion just restates the premise without adding new support. "This book is popular because it's the bestselling book right now" says the same thing twice, since popular and bestselling mean roughly the same thing here. The argument never explains why the book sells, it just swaps one word for another and calls that proof.

Same word, different meaning; one part standing in for the whole

The flaw of equivocation uses the same word with 2 different meanings across an argument. "Lee is fast, so Lee must excel at solving problems quickly" slides from physical speed to mental speed as if they were the same trait. Look up "fast" in a dictionary and you'll find multiple definitions. An argument has to use a key word the same way from start to finish, or the logic breaks down.

Confusing part and whole assumes what's true of one piece of a group must be true of the entire group, or the other way around. "This team has the best player in the league, so it must be the best team" makes exactly this mistake. One great player doesn't guarantee the team as a whole plays well together. The other players might be weak, or a group of stars might not mesh. What's true of a part doesn't automatically transfer to the whole.

Gaps that only look like proof, and appeals that skip proof entirely

The absence of evidence flaw assumes something is true, or false, just because it hasn't been disproven. "There's no evidence this new diet is harmful, so it must be safe" is a clean example. Not having evidence of harm isn't the same as having evidence of safety, it just means nobody has checked yet.

An appeal to emotion or authority swaps evidence for feeling or reputation. Take this line: "this charity must be effective, just think of all the people who would suffer without it." That's emotionally compelling, but it says nothing about whether the charity's programs actually work. The same flaw shows up when an argument leans on someone's authority in a field where that person has no real expertise. Feeling moved or impressed is not the same as being handed evidence.

Arguing with yourself, then mistaking timing for proof

An internal contradiction happens when an argument makes 2 claims that can't both be true. Consider: "this product is the best on the market, but no one should waste their money on it." If the product really is the best, it makes no sense to also say not to buy it. When you spot 2 statements in the same argument that fight each other, you've found an internal contradiction, and the argument falls apart on its own terms.

A correlation-causation flaw treats 2 things happening together as proof that one caused the other. "Sales increased after the new manager was hired, so the manager must be responsible" ignores other explanations: coincidence, a seasonal trend, or some third factor entirely. Timing alone never proves cause.

The short version

• Before you accept a claim built on a handful of people agreeing, ask who was asked and whether they already lean toward that answer, that's the sampling-flaw tell.
• When an argument attacks the person instead of the claim, or leans on feeling instead of fact, cross it off, neither one is evidence.
• Before you accept that X guarantees Y, check whether X is only required for Y, not enough to guarantee it, like a medical degree proving licensure.
• Log which of these 10 flaws you miss most on your next timed set, not just which questions you got wrong. If circular reasoning or equivocation is the one costing you points, drill 5 questions built only on those 2 flaws before you move to a mixed set.

Mark lesson 4 complete