Common Logical Flaws & Fallacies
What You Need to Know
Logical Flaw questions (and many Strengthen/Weaken/NA/Sufficient Assumption questions) hinge on spotting how an argument goes wrong. On the LSAT, most wrong arguments fail in a small set of repeatable ways: bad conditional reasoning, bad causal reasoning, bad sampling, shifting terms/quantifiers, or confusing what was proven.
Your core job: separate premises (support) from conclusion (claim), then name the gap: what must be true for the premises to justify the conclusion?
The “flaw vocabulary” you must recognize
- Necessary vs. sufficient confusion (especially with conditionals)
- Correlation vs. causation (and its cousins: reverse causation, confounders)
- Quantifier shifts (some/most/all; “not all” vs “none”)
- Scope shifts / new term introduced (conclusion is about something broader/different)
- Representativeness & sampling (small, biased, self-selected)
- Attacking the source instead of the claim (ad hominem)
- Misdescribing an opponent’s view (straw man)
- Bad comparisons/analogies (ignoring relevant differences)
Critical reminder: a “flaw” answer choice can be worded abstractly. If you can translate it into what happened in the stimulus, it’s a contender.
Step-by-Step Breakdown
Use this for Flaw, Weaken, Strengthen, Necessary Assumption, and Parallel Flaw.
Find the conclusion (the author’s main point).
- Look for: therefore, thus, hence, so, consequently, clearly.
- Or ask: “What is the author trying to prove?”
List the premises (the support).
- Evidence, facts, studies, statistics, comparisons, expert statements.
Put the reasoning in a simple form.
- Conditional? Causal? Statistical? Analogy? Plan/proposal? Elimination?
Articulate the gap in one sentence.
- “This assumes ____.”
- “This confuses ____ with ____.”
- “This rules out ____ without justification.”
Pre-phrase the flaw (predict it before reading answers).
- Even a rough prediction helps you avoid trap answers that sound logical.
Match your pre-phrase to an answer choice.
- Correct answers often use standard phrasing like “takes for granted,” “presumes,” “fails to consider,” “confuses,” “infers… solely from,” etc.
Micro-worked example (conditional)
Stimulus: “If a product is unsafe, it will be recalled. This product was recalled. Therefore it is unsafe.”
- Form:
- Flaw: Affirming the consequent (treating a necessary condition as sufficient).
Key Formulas, Rules & Facts
A. Conditional reasoning fallacies (super high-yield)
Valid moves
- Modus Ponens:
- Modus Tollens:
Invalid moves (classic flaws)
- Affirming the consequent:
- Denying the antecedent:
B. Fallacy map (what it is, LSAT tells, how to attack)
| Flaw / Fallacy | What it is | Common LSAT “tells” | Best attack (what would fix/undercut it) |
|---|---|---|---|
| Necessary vs. sufficient | Treats a requirement as a guarantee (or vice versa) | “Only if,” “requires,” “must,” “ensures,” “guarantees” misused | Identify which condition is required vs enough; look for missing link |
| Affirming consequent | Concludes from and | “Since we observed B, A happened” | Show other ways to get without |
| Denying antecedent | Concludes from and | “Not A, so not B” | Show can occur without |
| Causation from correlation | Treats correlation as proof of causation | “Associated with,” “linked to,” “goes with” → “therefore causes” | Alternative cause, reverse cause, coincidence, confounder |
| Post hoc | Because B followed A, A caused B | “After,” “since then,” “following” | Show other changes occurred; timing alone isn’t causation |
| Confounding variable | Ignores a third factor causing both | “People who do X have Y” | Introduce plausible common cause |
| Reverse causation | Mixes up direction | “X is correlated with Y, so X causes Y” | Show Y could cause X |
| Sampling bias / unrepresentative | Sample doesn’t represent population | “Survey of website visitors,” “volunteers,” “small group” | Demand random/representative sample; increase size/diversity |
| Small sample / anecdote | Generalizes from too little | “In my experience,” “one town,” “three cases” | Ask for broader data; show variability |
| Hasty generalization | Overgeneralizes from limited evidence | “Therefore all/most/always” from “some” | Point out exceptions; require stronger quantifier support |
| Quantifier shift | Illegitimately shifts some/most/all | “Some” → “Most,” “Not all” → “None” | Track quantifiers precisely; require support matching scope |
| Scope shift | Conclusion broader/different topic | Premises about A, conclusion about A + B | Highlight new element; require premise connecting it |
| Equivocation | Key term changes meaning mid-argument | Same word used differently | Clarify definition; show the switch |
| Circular reasoning | Conclusion restates a premise | “X is true because X” | Demand independent support |
| Straw man | Misrepresents opposing view | Opponent said “some,” author attacks “all” | Compare what was said vs attacked |
| Ad hominem | Attacks person/source not claim | “She’s biased/immoral, so false” | Separate credibility from truth; require evidence about claim |
| Appeal to authority (weak) | Treats authority as decisive when not qualified or context missing | “Expert says…” with no field relevance | Ask: qualified? consensus? evidence? |
| False dilemma | Only two options when more exist | “Either A or B” | Provide third alternative; show coexistence |
| Slippery slope | Unjustified chain of escalating outcomes | “If we allow A, soon Z” | Demand links between steps; show stopping points |
| False analogy | Assumes because two things share some traits, they share the key trait | “Just like…” | Identify relevant disanalogy |
| Composition | Parts have property ⇒ whole has it | “Each part is light, so the machine is light” | Whole can differ due to combination |
| Division | Whole has property ⇒ each part has it | “Company is rich, so each employee is rich” | Parts can differ from whole |
| Percent vs number / relative vs absolute | Confuses rate with raw count, or % change with total | “Higher percentage, so more people” | Demand base rates, denominators |
| Self-selection | People choose to be in sample/treatment | “Call-in poll,” “opt-in study” | Show participants differ systematically |
Examples & Applications
Example 1: Causation (classic weaken targets)
Stimulus: “In neighborhoods where more parks were built, crime fell. Therefore, building parks reduces crime.”
- Type: Correlation → causation.
- Key insight: Need to consider:
- Confounder: city also increased policing or economic investment.
- Reverse causation: city built parks because crime was already falling.
- Selection: parks built in neighborhoods already trending safer.
- Best weaken answer shapes: introduces alternative cause, shows timing mismatch, or shows crime fell equally where no parks were built.
Example 2: Necessary vs. sufficient (common flaw answer wording)
Stimulus: “To get licensed, a contractor must pass an exam. Rivera passed the exam, so Rivera is licensed.”
- Form: (passing is necessary, not sufficient)
- Flaw: Treats a necessary condition as sufficient.
- What would fix: add that passing the exam is enough, or that Rivera completed all other licensing requirements.
Example 3: Quantifier shift + sampling
Stimulus: “A survey of 50 customers found that some prefer Brand A to Brand B. Therefore, most customers prefer Brand A.”
- Flaws:
- Some → most (quantifier shift)
- Small/possibly unrepresentative sample
- Trap answers: choices that only say “survey size is small” might be incomplete if the bigger problem is the quantifier leap (depends on wording).
Example 4: Equivocation
Stimulus: “A ‘theory’ is just a guess. Evolution is a theory. So evolution is just a guess.”
- Flaw: “Theory” shifts meaning (scientific theory vs casual guess).
- Correct flaw answer often says: uses a key term in an inconsistent or ambiguous way.
Common Mistakes & Traps
Boldly labeling a fallacy without matching the stimulus.
- What goes wrong: You see “correlation” and scream “causation flaw,” but the argument only claims “this is evidence of,” not “this proves.”
- Fix: Match the strength of the conclusion: “suggests” vs “establishes.”
Mixing up conditional indicators (“only if” vs “if”).
- What goes wrong: You reverse the arrow.
- Rule: “ only if ” means (B is necessary).
- Fix: Translate slowly, then look for mistaken reversal.
Choosing an answer that’s true but not the flaw.
- What goes wrong: You pick a criticism of the evidence that doesn’t explain why the conclusion doesn’t follow.
- Fix: Ask: “If I fixed this issue, would the reasoning become valid?” If no, it’s not the core flaw.
Falling for “criticizes the author” trap language.
- What goes wrong: You pick harsh-sounding answers (e.g., “ignores all evidence”) when the stimulus does something milder.
- Fix: Calibrate: LSAT flaw answers must be accurate, not just negative.
Confusing “attacks source” with “attacks argument.”
- What goes wrong: You call it ad hominem whenever someone mentions bias.
- Fix: It’s ad hominem only if the argument goes: “Source is bad → claim is false,” without addressing the claim’s support.
Missing the “new term” in the conclusion (scope shift).
- What goes wrong: Premises discuss “reducing emissions,” conclusion recommends “banning cars.”
- Fix: Circle new concepts in the conclusion; ask where they were supported.
Overlooking denominator/base-rate issues in stats.
- What goes wrong: “More incidents” is inferred from “higher rate” (or vice versa).
- Fix: Always ask: “Out of how many?” “Compared to what total?”
On Parallel Flaw, matching topic instead of structure.
- What goes wrong: You pick the choice with similar subject matter rather than the same logical mistake.
- Fix: Abstract it: replace content with letters, then match the pattern.
Memory Aids & Quick Tricks
| Trick / Mnemonic | What it helps you remember | When to use it |
|---|---|---|
| CARS = Correlation, Alternative cause, Reverse causation, Statistical fluke/Selection | The main ways to attack causal claims | Any time you see “X is linked with Y, so X causes Y” |
| NA = “Needs + Assumes” | Necessary Assumption answers state what the argument must assume | On NA questions and when pre-phrasing a flaw |
| “Only if” = arrow points to the “only” side | Translate conditionals correctly | Any conditional stimulus |
| Some/Most/All ladder | Quantifier discipline: (but not backward) | Any generalization or survey |
| New Word = New Work | If conclusion adds a new concept, you need a premise bridging it | Great for quick flaw spotting |
| Parts ≠ Whole | Composition/division errors | Any argument moving between group and members |
Quick Review Checklist
- Can you underline the conclusion and list premises in under seconds?
- Did the argument switch from correlation to causation (or ignore reverse/third-variable explanations)?
- Did it confuse necessary and sufficient (especially via “only if,” “must,” “requires”)?
- Did it make an illegal conditional move (affirming consequent / denying antecedent)?
- Did it shift quantifiers (some → most/all; not all → none)?
- Did it broaden scope or add a new term in the conclusion?
- Did it rely on a bad sample (small, biased, self-selected) or an anecdote?
- Did it equivocate (same word, different meaning) or argue in a circle?
- For Parallel Flaw: did you match the structure rather than the topic?
You don’t need to memorize every name—just get fast at spotting the pattern and stating the gap clearly.