Build the Argument

An argument is a connected set of statements offered to justify a conclusion. Its strength depends not only on having facts, but on whether the reasons, evidence, and assumptions actually support the claim.

This lesson uses a simplified modern synthesis: Toulmin’s model informs the claim, evidence, warrant, qualifier, and rebuttal layout, while Ciceronian invention and Heinrichs inform the wider search for persuasive material and audience-facing choices. Cicero did not originate the modern technical categories of warrants and qualifiers.

Use these parts: claim, reasons, evidence, warrant, qualifier, counterclaim, and rebuttal condition. The claim is what you want accepted. Reasons explain why. Evidence gives observable support. A warrant is the rule or assumption connecting a reason to the claim. A qualifier marks strength or conditions. A counterclaim is a serious competing conclusion, while a Toulmin-style rebuttal states the exception or condition under which the claim would not hold.

Example: “Run a two-week pilot” is the claim. “We need evidence from real users” is a reason. Sign-up data and support logs are evidence. “A reversible pilot is the safest way to learn under uncertainty” is the warrant. “A pilot may create false expectations” is a counterclaim.

Write each part on a separate line. If one line is empty, you have identified the next piece of work instead of disguising it with confident prose.

A claim is the exact proposition an audience is asked to accept. Precision matters because a vague claim can absorb any evidence and escape any objection.

Claims can describe reality, evaluate it, or propose action. State one main claim in a complete sentence, then define its scope and degree of confidence. “The policy failed” is unclear. “During its first quarter, the policy did not reduce average response time enough to justify its staffing cost” supplies period, measure, threshold, and standard.

Test a claim with four questions:

• What would acceptance commit someone to believing or doing?
• Which words require definitions?
• Under what conditions does it apply?
• What observation could weaken it?

Do not make the conclusion larger than the support. Evidence that one intervention helped one team supports a local claim, not “this method always works.” Modest qualifiers such as probably, under these conditions, and for this decision are signs of calibration when they reflect uncertainty.

Claim template: “Given ___, we should believe/do ___ for ___ because ___, with confidence limited by ___.”

A reason is a proposition that counts in favor of the claim. A logical spine is the short sequence of essential reasons that lets a listener see why the conclusion follows.

Prefer two or three independent reasons over a pile of loosely related points. Each reason should pass three tests: it is relevant to the claim, acceptable on available support, and sufficient together with the other reasons. Remove material that is interesting but does not change the conclusion.

Watch for dependent reasons presented as independent. “Customers asked for it” and “support tickets mention it” may come from the same small group. Conversely, distinct evidence—behavioral data, operational feasibility, and a controlled comparison—can strengthen different links.

Spine drill: explain the argument in three sentences: claim, strongest reason, decisive evidence. Then add only the minimum second reason needed to address a different concern.

Evidence is information used to support or challenge a factual premise. Good evidence is relevant, sufficiently reliable, and proportionate to the decision at hand.

Match method to question. Records establish what was logged; direct observation shows behavior in a setting; experiments help estimate causal effects; representative surveys describe reported views; expert judgment interprets technical material; examples make possibilities concrete but rarely establish frequency alone.

Question	Useful evidence	Limitation to state
Did it happen?	Timestamped records, corroborated accounts	Missing or biased records
How common is it?	Representative sample, complete counts	Sampling and measurement error
Did X cause Y?	Experiment or credible comparison	Confounders and transfer limits
Can it work?	Demonstration or case	A possibility is not a typical result
Is it feasible here?	Local pilot, resource estimate	Short pilots miss long-term effects

Use sources transparently: name origin, date, method, and relevant limitation. Never inflate an example into a trend or a correlation into a cause. Honest limitations strengthen trust because they tell the audience exactly what the evidence can bear.

A warrant is the connecting principle that makes a reason relevant to a claim; an assumption is something the argument relies on without establishing inside the argument. These hidden bridges are often where reasonable disagreement lives.

“Sales fell, so replace the price” assumes price drove the decline and replacement will improve the outcome. The evidence may be accurate while the conclusion remains unsupported. State the bridge: “When demand drops after a price change while comparable factors remain stable, price is a plausible cause.” Now the audience can test whether those conditions hold.

Warrants often draw on causal rules, definitions, values, or practical norms. Ask, “Even if this reason is true, why does the claim follow?” Repeat until the answer becomes an explicit rule. Then ask whether the audience accepts that rule and whether exceptions matter here.

Assumption audit: list every condition that must be true for the proposed action to work. Mark each as supported, testable, contestable, or unknown.

A counterclaim is the strongest plausible conclusion that competes with yours. Including it improves reasoning because it forces comparison rather than pretending only one path exists.

State the counterclaim so its supporter would recognize it. Give its best reason and evidence before answering. Avoid a straw person—a deliberately weak version designed for easy defeat. The goal is not ceremonial balance; it is to locate the objection most capable of changing the decision.

Use a four-line card:

1. Some reasonable people conclude ___.
2. Their strongest reason is ___, supported by ___.
3. This matters because ___.
4. My claim still holds / should narrow because ___.

Sometimes the counterclaim wins. If it reveals that your remedy is infeasible, revise the proposal. If it exposes a subgroup harmed by your policy, add protection or limit scope. A resilient argument is one that can change shape when criticism reveals truth.

A qualifier states the strength or conditions of a claim, while a rebuttal condition identifies when it would not hold. Both make an argument more useful by showing its operating boundaries.

Compare “This will work” with “A four-week pilot is likely to reduce queue time for routine requests, unless demand shifts toward complex cases.” The second claim is less dramatic but more decision-ready. It defines result, population, time, confidence, and exception.

Choose confidence language consistently: possible means credible but unmeasured; likely means more probable than not; strongly supported means several reliable lines converge. Avoid numbers that imply precision you do not have.

Record reversal conditions before debate: “I would change my mind if the pilot increased severe errors, if adoption stayed below 40 percent, or if the comparison group improved equally.” This makes openness observable rather than performative.

Calibration checklist: scope named; uncertainty named; important exception named; evidence strong enough for consequence; reversal condition written.

A reasoning failure is a pattern that makes support look stronger than it is. Naming the pattern is less important than showing the broken link in plain language.

Failure	Broken link	Repair
Personal attack	Speaker traits do not settle the claim	Evaluate reasons independently
False choice	More than two options exist	List realistic alternatives
Hasty generalization	Sample cannot bear the scope	Narrow claim or gather data
Circular reasoning	Conclusion is restated as support	Add independent premise
Moving standard	Required evidence changes after it appears	Agree threshold beforehand
Cause confusion	Sequence or correlation is not causation	Test alternatives and mechanism

Do not turn fallacy labels into weapons. “That is ad hominem” often creates a side fight. Say, “Even if the criticism of me is true, which premise does it change?” The repair keeps attention on reasoning.

Run a red-team pass: try to falsify the main premise, break the warrant, find an omitted alternative, and identify who bears costs not represented in the evidence.

An argument canvas is a compact template for drafting and reviewing a complete case. It turns an opinion into inspectable parts and gives collaborators a shared editing surface.

Complete these fields:

• Decision and audience: what must be accepted or done, by whom?
• Main claim: one scoped, calibrated sentence.
• Reasons: two or three distinct supports.
• Evidence: source, method, date, and limitation for each reason.
• Warrants: connecting rules and values.
• Assumptions: supported, testable, contestable, unknown.
• Counterclaim: strongest competing view in fair terms.
• Response: concede, distinguish, rebut, or revise.
• Reversal conditions: what would change the conclusion?
• Next test: smallest ethical action that reduces decisive uncertainty.

Pair exercise: Partner A explains the canvas in two minutes. Partner B may ask only “How does that support the claim?” and “What would change your mind?” Swap roles, then revise one link rather than adding more decoration.

The finished canvas should let a skeptical beginner trace every step from evidence to action. If the reasoning cannot survive the page, a longer presentation will only conceal the gap.