Check whether the conclusion follows from the evidence
Critical thinking can be practised by identifying a claim, inspecting its support and checking whether the conclusion goes further than that support permits. A clear objection addresses a particular reasoning gap or missing fact. It does not require judging the person making the claim or treating disagreement as proof of poor judgement.
Editorial review: 9 October 2026. The documents, people and card records below form an original fictional argument exercise. They are not a real organisation's approval process, confidential records or a psychological assessment. No IQ, hiring, salary or workplace-performance result is inferred.
In actual work, the relevant policy, responsible owner and decision authority matter. Reasoning about a proposed rule does not grant permission to release a document, change a policy or decide another person's status.
Separate the supplied rule from a stronger proposal
In this invented exercise, coordinator Maya owns release decisions for internal session cards. Sharada is checking a draft instruction but has no release authority. The agreed source instruction states: “Every card approved for release carries an identifier.” It does not state that assigning an identifier approves a card.
A colleague proposes a shortcut: “Any card with an identifier can be treated as approved for release.” The shortcut reverses the relationship in the source instruction. To assess it, Sharada needs to inspect that exact inference rather than repeat a general statement that identifiers are useful.
The source instruction supports a requirement for approved cards. The proposal turns that requirement into a sufficient basis for approval. Those are different claims. Neither the word “every” nor a tidy-looking identifier supplies the missing approval decision.
Inspect the three supplied records
The original exercise supplies three records. X1 has an identifier, and Maya explicitly says it is approved for release. X2 also has an identifier, and Maya explicitly says it is a draft that has not been approved for release. X3 has no identifier in the supplied record, and its approval status is not supplied.
| Record | Facts actually supplied |
|---|---|
| X1 | Identifier present; Maya confirms release approval |
| X2 | Identifier present; Maya confirms it is not release-approved |
| X3 | Identifier absent from this record; approval status unspecified |
X2 is a concrete counterexample to the proposed shortcut: an identified card is expressly not approved. It is not merely a card whose approval is unknown. That distinction matters because a missing answer and a supplied negative answer support different descriptions.
X3 should stay unresolved on the evidence given. The absence of an identifier from this supplied record is not a new decision by Sharada about its wider status. She should ask for the relevant information rather than fill the gap with an assumed approval or rejection.
Explain the gap without changing the source rule
An original bounded explanation is:
The source says that approved cards carry identifiers. It does not say that identifiers create approval. X2 has an identifier, but Maya explicitly says it is not approved for release. Therefore the proposed shortcut does not hold for all supplied records. X1 has confirmed approval, and X3's approval status remains unspecified.
The explanation rejects the stronger shortcut using a supplied counterexample. It does not say identifiers are useless or that the original requirement is wrong. It also does not establish the complete approval procedure or certify every other card.
An unhelpful alternative would be “the colleague does not understand the work.” That assigns a personal judgement without examining the proposal. Sharada can identify the exact inference and request its correction without guessing the colleague's motive or competence.
Inspect where an instruction came from
The fictional packet also contains an unsigned practice worksheet with the sentence “Skip the approval check if an identifier exists.” It gives no authorised adoption or release decision. Its wording alone does not show that Maya's source instruction has changed.
Sharada records the worksheet as an unverified instruction source and asks Maya whether it has been adopted. The exercise supplies no adoption response. The worksheet's authority therefore remains unresolved; it should not silently replace the agreed rule.
The Harvard Library guidance on evaluating sources discusses publication date, relevance, authority, supporting accuracy and purpose in assessing sources. That academic guidance supports asking who makes a claim and what evidence fits the question. It is not a validated thinking score or a universal workplace approval policy.
For a real instruction, check the issuing organisation, intended audience, version, purpose and applicable adoption or amendment. A recent date may be relevant, but it does not alone establish authority. An older source may also remain applicable until replaced through the relevant process.
Ask a question that resolves the specific gap
Sharada's original request could be:
The agreed rule requires identifiers on approved cards. The proposed shortcut treats an identifier as approval, but X2 is identified and expressly not approved. Should the draft instruction retain a separate confirmation from you? The unsigned worksheet's adoption is also unverified. Please confirm the applicable instruction and the approval information needed for X3.
The request identifies the gap and the decisions needed. It does not assert that Maya has already replied, that the worksheet is fraudulent or that X3 is rejected. It also keeps the proposed correction separate from a confirmed instruction change.
In actual work, use the appropriate owner and permitted communication route. Do not send private card records to an external service to guess whether they are approved. A public invented exercise can demonstrate the reasoning without exposing real participant or employer information.
Keep an unknown separate from a counterexample
Consider a separate original variation in which X2's approval status is unknown instead of explicitly not approved. It still has an identifier, and the source instruction is unchanged. This is a new exercise, not a later event in the first record.
In the variation, X2 no longer supplies the same concrete counterexample. You can still say that the shortcut is not established by the original one-way instruction, and you need approval evidence. You cannot report that Maya rejected X2, because that answer is not supplied in this variation.
This demonstrates why wording must track the actual evidence. “Not confirmed” is different from “confirmed not approved.” Both may require an unresolved record in a particular workflow, but neither gives Sharada new release authority.
Produce a narrower usable conclusion
For the first exercise, an original final check note is:
The identified-card shortcut fails on the supplied X2 record. Keep identifier presence and release confirmation as separate fields in the draft. X1 has supplied approval; X2 is expressly not approved; X3 remains unknown. The worksheet's adoption is also unknown. Maya retains release and instruction decisions, and no wider outcome is supplied.
This note distinguishes a demonstrated reasoning error from the remaining questions. Keeping separate fields is a proposal for the draft unless the responsible owner approves the applicable change. The note does not claim that an operational system was updated or that all release errors were prevented.
For another kind of evidence gap, see analytical skills, which separates a checking count from a time-based rate and an unsupported causal explanation. The argument here concerns a reversed rule and a counterexample, rather than numerical productivity comparisons.
Practise with a claim small enough to inspect
Choose an invented claim with a clear premise and conclusion. Write the supplied facts separately, identify what the conclusion adds and test it against an expressly supplied case. Ask a willing reader to explain the gap in their own words. Record their actual response rather than treating agreement as proof of general competence.
A practice partner may help you spot that a status is unknown or that a conclusion reverses a requirement. That is a useful observation about the exercise. It does not diagnose a thinking style, validate a psychometric measure or predict an interview rating.
Questions before reporting the result
Does a required marker prove approval?
Not in this exercise. The source says approved cards require identifiers, while X2 shows that an identifier can also appear on an unapproved draft. Use the applicable real instruction before interpreting a marker elsewhere.
Can I call an unknown a counterexample?
Not in the same way as a supplied contradictory fact. Preserve the distinction between missing approval information and an explicit statement that approval has not been given.
Should I challenge the person or the claim?
Address the exact claim, its source and the reasoning gap. The supplied records provide no basis for a judgement about the colleague's personality or motive.
What has been established?
The stronger shortcut fails on the first exercise's X2 record. The status of X3, the worksheet's adoption and any later operational change remain unspecified. A useful conclusion includes both the supported correction and those remaining limits.