Analyzing Arguments
How to break down, evaluate, and construct arguments: the fundamental unit of reasoning.
What Is an Argument?
An argument is a set of statements where some (premises) are offered as reasons to accept another (the conclusion).
Not an argument:
- A fight ("We had an argument")
- An assertion ("The economy is bad")
- An explanation ("The economy is bad because of inflation"). This explains a fact, it isn't arguing for a conclusion
An argument:
- "Inflation has increased 8% this year [premise 1], wage growth has been only 3% [premise 2], therefore people's purchasing power has declined [conclusion]."
Anatomy of an Argument
Every argument has three components:
| Component | What It Is | Example |
|---|---|---|
| Claim/Conclusion | What the argument tries to establish | "We should hire more engineers" |
| Premises | Reasons offered to support the conclusion | "Our project backlog has tripled" |
| Assumptions | Unstated premises the argument relies on | "More engineers would reduce the backlog" |
Indicator Words
| Conclusion Indicators | Premise Indicators |
|---|---|
| Therefore | Because |
| Thus | Since |
| Consequently | Given that |
| So | As shown by |
| It follows that | For the reason that |
| This means | The evidence shows |
| We can conclude | Due to the fact that |
| Hence | Assuming that |
Warning: These words aren't always present. Many real-world arguments leave the structure implicit.
Identifying Claims
Types of Claims
| Type | Definition | Example | How to Evaluate |
|---|---|---|---|
| Empirical | About observable facts | "Sales increased 20%" | Check the data |
| Definitional | About what words mean | "That's not real socialism" | Agree on definitions first |
| Evaluative | About value or quality | "This is the best approach" | By what criteria? |
| Causal | About cause and effect | "The policy caused unemployment" | Check for confounds |
| Prescriptive | About what should be done | "We should raise taxes" | Requires both facts and values |
Precision Matters
Vague claims are impossible to evaluate. Always pin down exactly what's being claimed.
| Vague Claim | Better Version |
|---|---|
| "Crime is getting worse" | "Violent crime per capita increased 5% in 2024" |
| "This product is natural" | "This product contains no synthetic ingredients" |
| "Exercise is good for you" | "150 minutes of weekly exercise reduces cardiovascular mortality by 30%" |
| "The economy is bad" | "GDP contracted 1.2% last quarter and unemployment rose to 6%" |
Finding Hidden Assumptions
Every argument rests on unstated assumptions. Finding them is one of the most valuable critical thinking skills.
Method: The Gap Test
Look at the premises and conclusion. Ask: "What would need to be true, but isn't explicitly stated, for this conclusion to follow?"
Example argument:
"Our competitor released a similar feature, so we need to rush ours to market."
| Stated Premise | Hidden Assumptions |
|---|---|
| Competitor released a similar feature | Our feature needs to be first to market to succeed |
| Speed is more important than quality | |
| Customers will choose based on who was first | |
| We can't differentiate in other ways |
Once hidden assumptions are visible, you can evaluate them. Often, the hidden assumptions are the weakest part of an argument.
Common Hidden Assumptions
| Category | Example | Hidden Assumption |
|---|---|---|
| Causal | "She studied hard and passed" | Studying caused the passing |
| Normative | "That's unnatural, so it's wrong" | Natural = good, unnatural = bad |
| Generalization | "My friend had a bad experience there" | One experience is representative |
| False equivalence | "Both sides have valid points" | Both sides are equally supported by evidence |
| Status quo | "We shouldn't change the policy" | The current policy is working well enough |
Argument Mapping
Argument mapping makes the structure of reasoning visible. It's the single most effective technique for improving critical thinking.
Basic Structure
[CONCLUSION]
/ \
[Premise 1] [Premise 2]
/ \ |
[Sub-premise] [Sub-premise] [Evidence]
Types of Argument Support
| Type | How Premises Support Conclusion | Analogy |
|---|---|---|
| Convergent | Each premise independently supports the conclusion | Multiple legs of a table. Remove one, it still stands |
| Serial | Premises form a chain; each supports the next | Links in a chain. Break one, the chain breaks |
| Linked | Premises work together; each is needed | Legs of a two-legged stool, both required |
Worked Example: Argument Map
Argument: "Remote work should be the default. Studies show productivity increases, employees report higher satisfaction, companies save on office costs, and it reduces commuting emissions."
[Remote work should be the default] ← CONCLUSION
| | | |
| | | |
[Productivity [Higher [Lower [Reduced
increases] satisfaction] costs] emissions]
| | | |
| | | |
[Studies] [Surveys] [Real [Commuting
estate data]
data]
This is a convergent argument. Each premise independently provides support. Defeating one doesn't defeat the whole argument.
To attack this argument effectively, you'd need to either:
- Challenge multiple premises, or
- Show that the conclusion doesn't follow even if the premises are true (e.g., other factors outweigh these benefits)
Evaluating Argument Strength
For Deductive Arguments
| Criterion | Question |
|---|---|
| Validity | Does the conclusion logically follow from the premises? |
| Soundness | Are all premises actually true? |
| Completeness | Are there important missing premises? |
For Inductive Arguments
| Criterion | Question |
|---|---|
| Sample size | Is the evidence base large enough? |
| Representativeness | Is the sample representative? |
| Strength | How probable does the evidence make the conclusion? |
| Cogency | Are the premises true AND does the reasoning work? |
Universal Evaluation Checklist
Apply these questions to any argument:
| # | Question | Red Flag If... |
|---|---|---|
| 1 | What exactly is the conclusion? | It's vague or shifting |
| 2 | What are the premises? | They're missing or assumed |
| 3 | Are the premises true? | Can't be verified or are contested |
| 4 | Does the conclusion follow? | Logical gap between premises and conclusion |
| 5 | Are there hidden assumptions? | Unstated premises are doing heavy lifting |
| 6 | Is the evidence sufficient? | Anecdote used for sweeping claims |
| 7 | Are there counterarguments? | None considered |
| 8 | Is the argument charitable? | Misrepresents the opposing view |
Steel-Manning vs. Straw-Manning
Straw-Manning
Misrepresenting an argument to make it easier to defeat.
How it works:
- Opponent states position X
- You restate it as weaker position X' (distorted)
- You defeat X'
- You claim to have defeated X
Example:
Opponent: "We should consider reducing military spending." Straw man: "You want to leave us defenseless?"
Steel-Manning
Strengthening an argument before responding to it.
How it works:
- Opponent states position X
- You restate it as the strongest possible version X+ (may be stronger than they stated)
- You respond to X+
- If you can defeat X+, you've genuinely addressed the position
Example:
Opponent: "We should reduce military spending." Steel man: "The strongest version of this argument is that our military budget is larger than the next 10 countries combined, and redirecting even 10% could fund significant domestic infrastructure and social programs while maintaining military superiority. The opportunity cost of current spending is enormous."
Why Steel-Manning Matters
| Benefit | Explanation |
|---|---|
| Finds truth | The best version of an argument might actually be right |
| Builds trust | People engage when they feel understood |
| Strengthens your position | Defeating a strong argument is more meaningful |
| Prevents echo chambers | Forces engagement with opposing ideas |
| Improves your thinking | You might change your mind, and that's good |
Test: Before disagreeing with anyone, ask: "Could you tell me if I'm representing your view fairly?" If they say no, you haven't understood them yet.
Burden of Proof
Who has the obligation to provide evidence?
General Principle
The person making the claim has the burden of proof. It's not your job to disprove claims others haven't proven.
Specific Rules
| Situation | Burden Falls On | Why |
|---|---|---|
| Positive claim ("X exists") | The claimant | You can't prove a negative |
| Challenge to the status quo | The challenger | Existing practice has a track record |
| Extraordinary claim | The claimant (heavy burden) | Extraordinary claims require extraordinary evidence |
| Both sides make claims | Both sides | Each must support their own claims |
Common Burden-Shifting Tactics
| Tactic | Example | Response |
|---|---|---|
| "Prove me wrong" | "Prove that ghosts don't exist" | "You made the claim. What's your evidence?" |
| "Many people believe it" | "Lots of people have seen ghosts" | "Popularity doesn't shift the burden of proof" |
| "You can't explain X" | "You can't explain this photo" | "Not being able to explain something doesn't prove your explanation" |
| "It's just obvious" | "Everyone knows this is true" | "If it's obvious, evidence should be easy to find" |
Constructing Strong Arguments
The Structure of a Good Argument
- State your conclusion clearly. What exactly are you arguing for?
- Provide multiple independent premises. Convergent support is harder to defeat
- Support premises with evidence. Each premise should have backing
- Address the strongest counter-argument. Show you've considered the other side
- Acknowledge limitations. Credibility comes from honesty about weaknesses
- Make hidden assumptions explicit. Don't rely on unstated premises
Argument Quality Checklist
| Criterion | Questions |
|---|---|
| Clarity | Is every term defined? Could it be misunderstood? |
| Relevance | Do all premises actually relate to the conclusion? |
| Sufficiency | Is there enough evidence? |
| Accuracy | Are all factual claims correct? |
| Fairness | Have you represented opposing views accurately? |
| Logic | Does the conclusion follow from the premises? |
| Transparency | Are assumptions stated? Are limitations acknowledged? |
Example: Building an Argument
Conclusion: Our team should adopt automated testing.
| Component | Content |
|---|---|
| Premise 1 | Our bug rate has doubled in the last 6 months (data) |
| Premise 2 | Teams that adopt automated testing see 40-60% bug reduction (industry data) |
| Premise 3 | The setup cost is ~2 weeks of engineering time (estimate) |
| Counter-argument | "It'll slow us down" → Short-term yes; long-term velocity increases as debugging time drops |
| Limitation | This won't catch all bugs, especially UX issues |
| Hidden assumption made explicit | We value code quality and are willing to invest short-term for long-term gain |
Dialectical Thinking
Dialectical thinking means holding two opposing ideas and synthesizing them, rather than simply picking a side.
The Process
THESIS → An initial position or claim
ANTITHESIS → The opposing position
SYNTHESIS → A new position that incorporates valid elements of both
Example
| Stage | Position |
|---|---|
| Thesis | "The free market solves everything" |
| Antithesis | "Government must control the economy" |
| Synthesis | "Markets work well for most goods but need regulation for externalities, public goods, and market failures" |
The synthesis isn't splitting the difference. It's building a position that accounts for the legitimate insights on both sides while resolving the contradictions.
Practical Exercises
Exercise 1: Argument Extraction
Take any opinion piece or editorial. Identify:
- The main conclusion
- All premises (stated and unstated)
- The hidden assumptions
- The weakest link in the argument
Exercise 2: Steel-Man Practice
Pick a position you disagree with. Write the strongest possible argument FOR that position. If a proponent would say "Yes, that's exactly my point," you've succeeded.
Exercise 3: Assumption Hunting
Take a common belief in your field. List every assumption it rests on. How many of those assumptions have you actually verified?
Key Takeaways
- Arguments have structure. Conclusion, premises, and hidden assumptions
- Find the hidden assumptions. They're usually the weakest point
- Argument mapping works. Making structure visible improves evaluation
- Steel-man before attacking. Engage with the strongest version of opposing views
- Burden of proof matters. The claimant must provide evidence
- Good arguments acknowledge limits. Overconfidence is a red flag
- Convergent support is strongest. Multiple independent reasons beat a single chain