Terms 51–60: causes, measurements and uncertainty

51. Correlation

Correlation describes a statistical relationship in which values vary together. If weeks with more room visits also have more books borrowed, the quantities may be correlated. That relationship does not identify its cause. More visits could lead to more borrowing; a popular book display could increase both; or another factor could be involved. A pattern is the beginning of an explanation, not its automatic completion.

The verb is correlate; use is correlated with when describing an established association. Do not use the word merely because two events happened once in sequence. Tricia asks whether the claim comes from repeated paired observations or a single before-and-after story. She also avoids the opposite overstatement: “Correlation tells us nothing.” Correlations can be useful for prediction and for generating questions. The boundary is that the relationship alone does not establish which variable causes which, or whether a third factor explains the pattern.

52. Causation

Causation concerns one factor making a difference to an outcome. Saying that repainting caused better concentration is stronger than saying that concentration appeared better after repainting. The causal claim asks what would have happened without the change, under an appropriately comparable situation. That alternative cannot simply be assumed from the order of events.

Use causal explanation, causal evidence and the design does not establish causation. Kai Kai proposes measuring performance before and after the room changes. Tricia notes that this improves the evidence but does not automatically isolate colour if seating, timetable and teaching also change. Better causal reasoning identifies competing explanations and asks how the design addresses them. It does not require every school publication to run a professional experiment. It does require the wording of the conclusion to match the investigation actually completed rather than the investigation the writer wishes had occurred.

53. Confounding

Confounding occurs when another factor is entangled with the relationship being investigated, making it difficult to separate effects. In the room example, repainting occurs at the same time as longer opening hours. More visits afterward could reflect the additional access, the colour change, both, or other influences. The timetable is therefore a possible confounding factor in the proposed explanation.

Write a potential confounder or the comparison is confounded by and name the relevant relationship. Listing every imaginable factor is less helpful than identifying those plausibly connected to both the change and the outcome. Alicia writes, “Because opening hours increased during the same period, the records cannot isolate the effect of repainting.” That sentence turns the abstract term into a specific limitation. A confounder need not be a secret or a mistake. It may be an ordinary feature of how events occurred, which the report must acknowledge.

54. Reverse causality

Reverse causality is the possibility that the direction of cause runs opposite to the proposed explanation. A report might say that pupils who use the room read more because the room changes their habits. Another possibility is that pupils who already enjoy reading choose to visit more frequently. The observed relationship can fit both stories unless further evidence distinguishes them.

Use reverse causality is a plausible alternative, not “reverse causality proves the report wrong.” The alternative weakens an unsupported causal conclusion; it does not establish itself automatically. Tricia asks about reading habits before pupils began using the room. That information could help, although its quality and completeness still matter. When evaluating any claim linking behaviour and outcome, draw an arrow in the proposed direction, then reverse it. If the reversed story remains plausible, explain what evidence would be needed to choose between the accounts or recognise a two-way relationship.

55. Counterfactual

A counterfactual considers what would have happened under a different condition from the one observed. The causal question is not only whether visits rose after repainting. It is whether visits would have risen differently without repainting. We cannot directly observe both versions of the same room at the same time, so an investigation needs a defensible comparison or model.

Use the counterfactual comparison and a hypothetical alternative outcome. Kai Kai imagines an identical week without the new paint, but imagination is not evidence. A comparable room, a carefully planned design or other information might help estimate the alternative, each with limitations. In school writing, the term is useful for exposing what a causal claim assumes. “The report has not shown what would have happened under unchanged conditions” is a clear explanation. Do not turn counterfactual reasoning into unrestricted storytelling; the alternative must remain relevant and constrained by what is known.

56. Operational definition

An operational definition states how an idea will be observed or measured in a particular investigation. “Engagement” is broad. The publication might define it as the proportion of a reading session spent on the assigned text, using a specified observation method. Another study might measure voluntary return visits. Those measures do not capture exactly the same thing.

Write operationally defined as when explaining the measurement rule. Alicia learns that naming a concept is not enough to make it measurable. Before reporting “improved concentration,” she asks what behaviour or performance was counted, by whom and under which conditions. The operational definition may be useful while still incomplete. It does not redefine the whole concept for every context. A good report explains what its chosen indicator captures and what it leaves out, so readers do not mistake one measurable aspect for the entire quality being discussed.

57. Measurement validity

Measurement validity concerns whether a measure supports the interpretation intended for it. Counting room visits may be suitable for describing attendance. It is not, by itself, a valid measure of reading comprehension. A measure can be accurately recorded yet fail to answer the research question. The mismatch is between the indicator and the claim, not necessarily between the number and the record.

Use the measure has limited validity for this claim. Tricia accepts the attendance count while rejecting its use as proof of improved learning. That distinction allows a nuanced answer: the data are useful, but for a narrower purpose. To test your own project, write the concept on one side of a page and the measurement on the other. Explain the connection and identify a situation in which the measurement could change without the concept changing. That counterexample often reveals a limitation worth reporting.

58. Reliability

Reliability concerns consistency under comparable conditions. Two observers following the same clear counting rule should not produce wildly different attendance totals. Repeated measurements of a stable situation should be reasonably consistent, allowing for expected variation. Reliability is necessary for many useful measurements, but it does not guarantee validity.

Write reliable measurement, inter-observer agreement or inconsistent recording as appropriate. Alicia and Kai Kai count visits differently: one counts each entry, the other counts each unique pupil. Their disagreement is partly a definition problem. They clarify the rule before comparing results. Even perfect agreement about entries would not measure comprehension. This example separates two common questions: “Can the method be applied consistently?” and “Does the method measure what the conclusion claims?” Strong evaluation can answer yes to the first and no to the second without contradicting itself.

59. Uncertainty

Uncertainty is the condition of not knowing an answer completely or precisely. It can arise from limited data, measurement variation, missing context or competing explanations. Uncertainty does not mean that every conclusion is equally plausible. Some claims may be strongly supported while other parts of the same report remain unresolved.

Useful phrases include reduce uncertainty, the principal uncertainty and uncertainty remains about the cause. Tricia distinguishes confidence in the attendance records from uncertainty about why attendance changed. This makes the report more informative than a vague statement that “nothing is certain.” Name the source of uncertainty and its practical consequence. Does it prevent publication of the entire story, require narrower wording, or justify another question? Advanced reasoning keeps those responses separate. A limitation should guide interpretation or action, not become a decorative disclaimer pasted beneath an overconfident headline.

60. Calibration

Calibration, in judgement, means matching confidence to the strength of the available evidence. It differs from the technical calibration of an instrument, although both involve alignment with an appropriate reference. A calibrated writer is firm about a directly documented event and more cautious about a speculative explanation. Caution should be selective rather than spread evenly across every sentence.

Use calibrate the conclusion and well-calibrated confidence. Alicia writes, “The room opened for two additional sessions,” because the schedule establishes it. She writes, “The added sessions may explain part of the attendance increase,” because that causal contribution is not isolated. Her language communicates different evidential states. To practise, mark sentences as documented, supported interpretation or unresolved possibility, then revise their verbs accordingly. Do not invent numerical confidence percentages without a defensible basis. Clear verbal distinctions often communicate school-level uncertainty more honestly than false precision.

Terms 61–70: platforms, attention and distribution

61. Affordance

An affordance is an action a design makes possible or invites a user to take. A share button makes redistribution easy. A short-video format encourages a different kind of message from a long article. These features do not determine exactly what users will do, but they shape the available actions and their effort.

Write the platform affords rapid sharing or a design affordance. Kai Kai notices that his dramatic headline fits neatly into a small preview card, while its qualifications remain behind a link. The format creates a communication problem: the most widely encountered version may be the least complete. He can respond by making the preview accurate on its own rather than relying on readers to repair it later. When comparing media forms, ask what each makes easy, difficult, visible or hidden. Avoid treating technology as an all-powerful cause that removes human choice.

62. Engagement

In platform reporting, engagement refers to defined interactions with content, such as comments, reactions, shares or viewing behaviour. The exact definition depends on the service and report. Engagement does not necessarily mean agreement, understanding or approval. A critical comment and a supportive comment may both count as interactions.

Use engagement metric and state what was counted. Alicia’s article receives many comments after a confusing headline. Calling this “strong public support” would misinterpret the metric. The comments might reflect disagreement or requests for clarification. The group reads a sample of the actual responses and distinguishes their content from the interaction total. In an advanced answer, explain the indicator’s operational definition before judging success. If the goal was accurate understanding, clicks and reactions are incomplete measures. They describe activity around the message, not necessarily the quality of what the audience learned.

63. Reach

Reach commonly refers to the number of distinct people or accounts exposed to content within a defined reporting system and period. It differs from total displays because one account may encounter the same post several times. Platforms may estimate or define reach differently, so comparisons require attention to the method.

Write reported reach during the campaign rather than treating the figure as an exact census of human readers. Kai Kai sees a reach figure of two hundred and writes “two hundred people read the full article.” The metric does not establish full reading. Exposure may mean only that a preview appeared. A more accurate sentence reports the platform measure and avoids adding an unmeasured behaviour. This distinction helps students interpret publicity claims and their own project analytics. Count what the system actually records, then separately investigate whether the intended audience understood or acted on the message.

64. Impressions

Impressions are recorded displays of content under the platform’s definition. One account seeing a post five times may contribute five impressions, not five unique viewers. Impressions therefore describe delivery events rather than a direct count of people, opinions or completed reading. The timeframe and reporting rules matter.

Use total impressions and impressions per reached account only when the underlying measures are compatible. Tricia compares six hundred impressions with two hundred reached accounts in an invented report. Dividing gives an average of three recorded displays per reached account, not proof that every person saw the post exactly three times. Averages summarise distributions; they do not describe every case. The example connects vocabulary, arithmetic and interpretation. A correct division can still produce a misleading sentence if the writer turns an average or a platform event into an unsupported claim about individual behaviour.

65. Ranking

Ranking is the ordering of results or content according to a system’s criteria. A high position may reflect relevance, predicted interest, recency or other factors, depending on the service. It is not a universal certificate of factual accuracy. The first result deserves evaluation just as later results do.

Write search ranking, ranked results and the ranking criterion. Alicia changes her query from a sensational phrase to the report’s exact title and finds different sources. The result order has changed because the question submitted to the system has changed. She records the query when it matters to reproducing her search. Students should not claim to know the current hidden rules of a platform unless they have appropriate evidence. The transferable skill is to recognise ranking as a selection process and investigate sources, rather than confuse visibility with authority.

66. Recommendation system

A recommendation system selects or orders items for a user based on information available to the system and its objectives. It may use viewing history, item features or patterns from other users, among many possible inputs. A recommendation is a prediction or selection, not evidence that the recommended content is true, important or personally beneficial.

Use recommended content and the system’s selection objective. Kai Kai watches several videos about study-room colour and then encounters more of the same topic. That experience alone does not prove what every platform user sees. Nor does it reveal the complete algorithm. He can report the observation while limiting the inference. In media analysis, distinguish the content’s claims from the distribution mechanism that brought it to the audience. The mechanism may explain repeated exposure, while the claim still needs separate evidence.

67. Amplification

Amplification is an increase in the visibility or spread of a message. Sharing, prominent placement, paid promotion or recommendation can amplify content. The term describes distribution rather than truth. A false claim can be amplified; so can a correction, a useful explanation or an ordinary announcement.

Write amplify the claim or amplification through repeated sharing. Alicia realises that repeating a sensational sentence merely to criticise it may still give it another audience. Her correction therefore leads with the accurate information and includes only the misleading wording necessary for readers to identify the problem. This is an editorial choice in the fictional case, not a guarantee about all audiences. The concept encourages students to consider the consequence of their own communication. Debunking is not outside the media system; a correction also has a headline, audience and pathway of distribution.

68. Feedback loop

A feedback loop occurs when a system’s output influences later inputs or behaviour. A publication notices that dramatic headlines attract more clicks, publishes more dramatic headlines, and then receives further click data shaped by that choice. The observations are partly produced by the system’s own earlier decisions.

Use a reinforcing feedback loop when the process strengthens itself. Not every loop is harmful; a correction process can improve accuracy through reader feedback. Tricia asks what the group is learning from its analytics. If it tests only sensational titles, it may learn little about the performance of accurate, informative alternatives. The advanced question is not merely “What did users do?” but “What choices did we make that shaped what they were able to do?” This brings evaluation back to the design of the system rather than treating its data as a neutral window onto all possible preferences.

69. Selection effect

A selection effect is a pattern produced partly by which cases become visible or included. Reading only the most widely shared posts can create a different impression from reading a broader sample of posts. The visible collection may favour emotional, unusual or heavily promoted material. The exact effect requires evidence; it should not be guessed from one experience.

Write the visible sample may reflect a selection effect. Kai Kai concludes that “everybody is angry” after reading the most commented-on posts. Tricia asks whether calm responses are less likely to appear in that selected set. They cannot infer the whole audience from a collection selected for high interaction. The term overlaps with selection bias but is often used more broadly to describe how inclusion creates a pattern. State the selection process explicitly. That explanation is more useful than treating a feed as a representative survey of public opinion.

70. Incentive

An incentive is a reward, cost or condition that makes an action more or less attractive. A publication rewarded only for click totals may have an incentive to prioritise attention. A team assessed on accuracy and useful corrections has different incentives. Incentives influence behaviour without determining every individual choice.

Use commercial incentive, an incentive to exaggerate and align incentives with the purpose. Do not treat the existence of an incentive as proof that a person acted improperly. Tricia can explain why sponsorship creates a reason for scrutiny while still examining the actual evidence. In their own project, the learners change the success criterion from “most reactions” to “readers can accurately explain the main result and its limitation.” Their chosen metric is imperfect, but it better matches the educational purpose. The vocabulary helps them redesign the activity rather than merely criticise outsiders.

Terms 71–80: photographs, files and synthetic media

The C2PA Content Credentials explainer, version 2.2, distinguishes verifiable provenance information from the truth of what an image or recording depicts. Credentials can support an investigation, but their presence does not certify every claim and their absence does not prove that a file is false.

71. Authenticity

Authenticity concerns whether an item is genuinely what it is presented as being. A photograph may genuinely come from the named photographer and event. That still leaves questions about framing, timing and the claims made from it. Authentic evidence can be used misleadingly when a caption or surrounding story changes its significance.

Use establish the photograph’s authenticity and specify the claim being authenticated. Alicia distinguishes “This is a genuine photograph of the reading room” from “This photograph proves improved learning.” The second proposition does not follow from the first. A genuine image of smiling visitors can show that particular moment; it cannot measure all visitors’ learning. Avoid using authentic as a universal synonym for reliable, complete or unbiased. In digital-source evaluation, ask separately about origin, alteration, context and interpretation. Different evidence may be needed for each question.

72. Integrity

For digital evidence, integrity concerns whether information has remained intact or changed only in documented, expected ways. A preserved original recording can be compared with an edited publication to check what was removed. This is distinct from the personal virtue also called integrity, although responsible handling of evidence may express that virtue.

Write file integrity, preserve the original and an integrity check. Tricia stores a read-only original and edits a working copy. If the final quotation is challenged, the group can return to the earlier record. An unchanged file can still contain a false statement; integrity does not prove the truth of its contents. The concept is valuable because it narrows a question: did the evidence change after collection? That narrower question is answerable in ways that a broad demand to “prove the whole story is true” may not be.

73. Manipulation

Manipulation is alteration of material, often discussed when the alteration changes meaning or deceives an audience. Editing itself is not automatically deceptive. Adjusting volume for audibility, trimming an irrelevant pause or resizing an image can be legitimate. The ethical and evidential questions concern what changed, whether meaning was preserved and whether important changes were disclosed.

Use deceptive manipulation when deception is supported, rather than labelling all edited media false. Alicia removes a long silence from an interview. If that silence conveyed uncertainty relevant to the answer, its removal could change the audience’s interpretation. The group compares the versions and explains the edit. An advanced analysis describes the operation and its consequence: “The cut removes the speaker’s qualification and makes the answer appear unconditional.” That is stronger than saying “The video has been edited,” because almost every produced video has been edited in some way.

74. Synthetic media

Synthetic media are images, audio, video or other content generated or substantially constructed through computational processes rather than simply captured as an unaltered record of an event. Synthetic material can be used openly for illustration, creativity or simulation. The problem arises when it is represented misleadingly as evidence of something that happened.

Write a synthetic illustration or generated media labelled as such. Kai Kai wants an image of the future reading room for a proposal. A clearly labelled concept image can help people imagine the design. It must not be presented as a photograph of a completed room or used to prove attendance. The distinction is between representation and record. Students should judge a media item in relation to the claim made about it. Generated does not automatically mean useless, and photorealistic does not automatically mean documentary.

75. Deepfake

A deepfake is synthetic or manipulated media that convincingly imitates a person’s appearance, voice or actions using machine-learning techniques. It can create the impression that someone said or did something they did not. Detection from visible oddities alone is not a dependable universal method; provenance, original sources and corroborating information remain important.

Use a suspected deepfake when authenticity has not been established. Do not accuse a real person or dismiss an inconvenient recording merely because manipulation is technically possible. In the fictional editorial case, an extraordinary voice clip arrives with no original source. The team withholds the allegation and seeks appropriate verification rather than forwarding the clip as a challenge for classmates. This is defensive media literacy, not an exercise in impersonation. The vocabulary should help learners recognise when an apparent record requires scrutiny and when uncertainty makes publication unfair or unsafe.

76. Content Credentials

Content Credentials refer here to provenance records associated with the C2PA approach to digital content. They can record information about creation and editing in a form designed to make tampering detectable. Their meaning is narrower than “this image is true.” They concern recorded history and its connection to the file, subject to the system’s trust and validation rules.

Use inspect the credential information rather than treating a badge as a complete judgement. A genuine camera record could depict a staged scene. A correct edit history would not, by itself, tell a reader that the scene was spontaneous. Equally, an ordinary photograph without credentials is not automatically false. Alicia therefore uses credentials as one possible piece of evidence within a wider evaluation. The student’s job is to ask what the record establishes and what remains outside it, not to replace careful interpretation with a new technical label.

77. Digital fingerprint

A digital fingerprint is a representation used to identify or compare digital material. The phrase can refer to different technical methods, so the method must be specified. A cryptographic hash can help check exact file identity; a perceptual fingerprint may be designed to recognise similarity after some changes. Those are different tasks.

Write an exact-file hash when exact identity is the point. Matching identifiers can support a conclusion about the file, not about the truth of every claim associated with it. Tricia explains this with an ordinary analogy: two copies of the same mistaken sentence remain the same sentence. The identity check has worked, while factual accuracy remains unresolved. Students need not implement forensic tools to understand the boundary. They should recognise that technical verification always verifies something specific, and should ask what that specific property is before expanding the conclusion.

78. Compression

Compression reduces the amount of data needed to store or transmit a file. Some methods preserve information exactly; others discard information to reduce size further. Repeated saving or platform processing can change visible detail without a person deliberately fabricating the scene. Unusual edges or blurred features can have several causes.

Use compression artefact for a distortion produced by compression, when that explanation is supported. Do not diagnose manipulation from a low-quality screenshot alone. Kai Kai sees a blurred hand in a repeatedly shared image and announces that it is generated. Tricia asks for the original file and context. The observation may justify checking, but it does not justify certainty. Advanced media analysis separates a warning sign from a conclusion. The same discipline applies to text: an apparent quotation error may come from transcription, editing or fabrication, and the evidence should distinguish those possibilities.

79. Cropping

Cropping removes parts of an image at its boundaries. It can improve composition or focus, but it can also remove context essential to interpretation. A tight photograph of three occupied seats may make a nearly empty room seem crowded. The visible pixels may remain genuine while the audience receives a misleading impression of the whole scene.

Write the crop excludes and identify what matters. Alicia keeps a close-up for a story about a particular reading activity but does not use it to illustrate a claim about total attendance. The purpose determines what context is needed. A crop is not automatically unethical; a mismatch between the crop and the claim can be. To practise, describe two possible captions for the same cropped image, one appropriately narrow and one misleadingly broad. The exercise teaches that a photograph’s evidential value depends partly on the sentence it is asked to support.

80. Recontextualisation

Recontextualisation places an item into a new setting of words, time, audience or purpose. An old photograph used in a historical comparison is openly recontextualised. The same photograph presented as today’s event can mislead. The file need not be digitally altered for its communicated meaning to change.

Use the image has been recontextualised and explain how. Tricia checks the event date separately from the date of a new post. Alicia labels an archive image clearly so it supports background rather than a claim about current conditions. The term connects visual evidence with temporal reasoning. A current caption can attach an outdated event to a present controversy, and a genuine quotation can be moved into an unrelated dispute. Always evaluate the relationship among item, caption and surrounding claim. Authenticity of the original does not guarantee accuracy of the new use.

Terms 81–90: turning evidence into writing

81. Paraphrase

A paraphrase restates a particular meaning in new wording while preserving its important content and limits. Replacing a few words with synonyms is not enough when the original sentence structure and distinctive expression remain. More importantly, a paraphrase must not strengthen the claim. “Some respondents preferred the room” cannot honestly become “students benefited from the redesign.”

Use paraphrase accurately and preserve the qualification. Alicia closes the source, explains the point in her own language, then reopens it to check population, certainty and causal wording. She still attributes the idea where appropriate. The method tests understanding rather than thesaurus use. A useful final check is to ask whether the source’s author could reasonably recognise the same claim in the new sentence. If the paraphrase removes the sample, timeframe or uncertainty, it may be shorter and smoother while becoming less accurate.

82. Summary

A summary selects and condenses the main information from a larger source. It differs from paraphrase because it usually reduces scope and detail rather than restating one passage at similar length. Selection is unavoidable, so the summary must preserve the source’s central meaning and important qualifications rather than only its most memorable sentence.

Write a concise summary of the findings. Kai Kai summarises a report about attendance, access and learner preferences. If he includes only the positive comments, he has produced a selective promotional account rather than a fair summary of the findings. Tricia asks him to name the report’s main result, strongest limitation and relevant context before drafting. The final summary can be brief while still retaining those essentials. In examinations, the exact task may require only selected points; read the question carefully rather than assuming every summary has the same purpose or length.

83. Quotation

A quotation reproduces someone’s exact words with clear attribution. Its value may lie in the speaker’s distinctive wording, an important admission or a phrase being analysed. A quotation is not automatically stronger evidence than a careful paraphrase. Its relevance depends on what the speaker could know and what the words actually establish.

Use quote accurately, quotation in context and the quotation supports. Alicia keeps the word feels when quoting a visitor who says the room feels calmer. Removing it could convert a personal impression into an objective description. She also checks the surrounding exchange so that a reply is not presented as answering a different question. Exact words can still be misleadingly selected. In your own work, choose short quotations that perform a clear analytical job, explain that job, and do not let quotation marks replace your reasoning.

84. Citation

A citation identifies the source supporting information or an idea so a reader can locate and evaluate it. Different tasks use different citation formats, but the underlying purpose is traceability. A long list of references does not prove that every claim is supported. Each important claim must connect to a source that actually addresses it.

Write provide a citation for and the cited source. Tricia checks whether the linked report contains the statistic used in the article. A reference to a general page about learning does not verify a specific attendance count. Nor should students invent a page number, author or publication title when details are missing. Record what has actually been inspected. A citation can identify a source whose conclusion you challenge; the surrounding sentence should make that relationship clear. Referencing is a tool for intellectual accountability, not a decorative sign that a paragraph is academic.

85. Synthesis

Synthesis combines information from several sources into a coherent account of a question. It does more than place one summary after another. A synthesis identifies agreement, disagreement, complementary evidence and remaining gaps. It may conclude that sources answer different questions rather than contradict one another.

Use synthesise the evidence and an evidence synthesis. Alicia combines attendance records, the new timetable and interviews. Her resulting account says visits rose, access increased and some users liked the room, while the effect on learning remains unmeasured. Each source contributes a different piece. She does not average their credibility or count how many favour the redesign. The synthesis is organised around claims and evidence roles. To practise, write one sentence beginning “Taken together…” and check that the conclusion genuinely follows from the combined material rather than merely sounding like a grander version of one source.

86. Counterargument

A counterargument is a reasoned challenge to a position. It may dispute a premise, offer a competing explanation, identify a cost or question the proposed action. A counterargument should be represented in a form its supporter could recognise. Weakening it into an obviously foolish version makes the response easier but the reasoning poorer.

Use consider a counterargument and a substantial counterargument. Kai Kai wants to keep the room open later. A counterargument is that staffing the added sessions would reduce support elsewhere. That is not equivalent to “some people dislike reading.” Tricia asks what evidence would show the staffing trade-off and whether alternatives exist. Engaging with a counterargument does not require abandoning the original goal. It may produce a more carefully limited proposal. Advanced writing becomes stronger when it addresses the best relevant challenge rather than merely announcing that opposing views exist.

87. Rebuttal

A rebuttal is a response that challenges an argument or part of it using reasons and evidence. It is not simply disagreement expressed firmly. If someone argues that extra opening hours are impossible because no staff are available, a relevant rebuttal might present a workable approved staffing arrangement. A statement about the beauty of the room would not address that objection.

Write rebut the claim, not “rebut the person.” Alicia learns to specify the part being answered. Her evidence may rebut the assertion that no additional session is possible without proving that every proposed session is affordable. A bounded rebuttal is more accurate than claiming total victory. In a timed essay, state the objection, provide the directly relevant response and explain its limit. This structure prevents a paragraph from drifting into unrelated praise for the writer’s preferred position while leaving the actual counterargument untouched.

88. Concession

A concession acknowledges a valid part of another position. It is not surrender of the whole argument. A writer may concede that longer opening hours require resources while still arguing that one additional session is justified. The concession improves the argument when it accurately recognises a real constraint and helps shape a feasible response.

Use concede that and make a limited concession. Tricia writes, “Although the interviews do not establish improved learning, they do provide useful evidence about how these visitors experienced the room.” She concedes the limitation without discarding the interviews entirely. This is a useful antidote to all-or-nothing evaluation. Evidence can be insufficient for one claim and valuable for another. A mature paragraph explains that distinction rather than branding a whole source “useless” after discovering one methodological boundary.

89. Hedging

Hedging uses language to express appropriate limits on certainty or generality. Words such as may, suggests and under these conditions can help align a statement with its evidence. Hedging becomes unhelpful when it is vague, excessive or used to protect a claim that still exceeds its support.

Write appropriate hedging when the uncertainty is justified. “The schedule may have contributed to the increase” is more defensible than “the colour definitely caused it.” However, “maybe perhaps the room possibly opened on Monday” weakens a fact established by a dated record. Alicia revises sentence by sentence, distinguishing documented events from interpretations. The goal is calibrated confidence, not timid prose. A useful hedge names a limit readers can understand. It should not become a cloud of hesitant words that conceals what the writer actually knows.

90. Scope

Scope is the range of people, circumstances, time or claims an investigation or statement covers. A survey of one class has a different scope from a census of the school. An analysis of attendance has a different scope from an assessment of learning. Stating scope helps readers understand both usefulness and limits.

Use within the scope of the study and outside the evidence’s scope. Kai Kai checks every broad noun in his draft: students, learning, improvement, success. Which group? Which outcome? Which period? Those questions often reveal a hidden expansion. A conclusion may become inaccurate through one unqualified noun even when its numbers are correct. Scope is also a planning tool. A manageable article can answer a narrow question well instead of pretending to settle an entire debate. The best boundary is explicit and useful rather than so broad that no evidence can satisfy it.

Terms 91–100: publishing with responsibility

91. Informed consent

Informed consent means that permission is given with an adequate understanding of the proposed action and relevant consequences. In a school media project, asking to record an interview is not the same as explaining that the clip will be published publicly. Age, school policy, safeguarding and local requirements affect what permissions are needed.

Use obtain appropriate informed consent without treating a simple checkbox as sufficient for every situation. Alicia explains the purpose, audience and intended use of the interview through the school’s approved process. She does not pressure a pupil who declines. The vocabulary lesson is the distinction between bare agreement and meaningful permission. This manual is not legal advice and cannot determine the requirements for a particular school or jurisdiction. For classroom practice, fictional transcripts and non-identifying examples can teach the same analytical skills without requiring students to expose personal experiences or private information.

92. Data minimisation

Data minimisation means collecting and retaining only the information reasonably needed for a defined purpose. A survey about preferred opening hours may not need names, home addresses or detailed personal histories. Extra data can create extra privacy responsibilities without improving the investigation.

Write minimise unnecessary collection and limit the dataset to the purpose. Tricia removes an identification field from a classroom demonstration because anonymous counts answer the question. She also checks whether supposedly anonymous answers could identify someone through a rare combination of details. The practical lesson is not that all data collection is wrong. It is that purpose should control collection rather than curiosity. When planning a project, explain why each field is needed, who can access it and when it will be removed under the relevant school process. Do not gather sensitive material merely to make the project appear sophisticated.

93. Accessibility

Accessibility concerns whether people with different abilities and access needs can use and understand a communication. A report may require meaningful headings, readable structure, captions for relevant audio and text alternatives for important images. Accessibility is part of communicating the evidence, not a decorative feature added after the content is finished.

Use accessible presentation and an accessibility barrier. Alicia’s chart uses two similar colours without labels. A reader who cannot distinguish them loses the comparison. Adding direct labels improves access and may also improve clarity for other readers. The W3C Web Accessibility Initiative’s image guidance explains how alternatives depend on an image’s purpose. Students should not assume that one technique solves every barrier. Ask what information a person needs and whether the chosen form makes that information available through an appropriate route.

94. Alternative text

Alternative text communicates the relevant information or function of an image when the image itself is not accessible to the reader. Its purpose is not to list every visible detail mechanically. A decorative flourish has a different role from a graph that supplies the article’s central evidence. Complex data may need a fuller text explanation or table as well as a short alternative.

Use write meaningful alternative text. For the invented attendance chart, a useful description states the comparison and points to the full figures in the adjacent text. “A blue chart” does not convey the evidence. Alicia also checks that the article does not depend on the reader seeing colour alone. The vocabulary distinction is between describing appearance and transmitting meaning. Decide what the image contributes to this particular page, then make that contribution available. Do not insert unrelated keywords or promotional language into an accessibility field.

95. Correction

A correction repairs an identified error in published material. A useful correction states what was wrong and supplies the accurate information. The appropriate visibility depends on the significance of the error and where audiences encountered it. Quietly changing a material claim may leave readers of the earlier version with a mistaken impression.

Write issue a correction and correct the reported figure. Kai Kai discovers that he called visits “unique visitors.” The correction names the distinction and updates both article and preview. It does not blame readers for misunderstanding wording that invited the error. Minor typographical repairs and substantive factual corrections need not receive identical treatment, but both should preserve accuracy. In a school project, keep an internal record of important changes and follow the publication’s approved process. Correcting responsibly demonstrates learning and accountability rather than proving that the work had no value.

96. Retraction

A retraction withdraws a published claim or work when a serious problem means it should no longer be relied upon in its previous form. It is more substantial than correcting one number while the main conclusion remains sound. The reasons and procedures depend on the publication setting.

Use retract the unsupported allegation and explain what is being withdrawn. Alicia’s fictional group discovers that an interview central to a story never occurred. Adding a small footnote would not repair an article built on fabricated evidence. The group must stop presenting it as valid and follow the school’s responsible correction process. Retraction should not be confused with removing a page merely because it attracted criticism. The relevant issue is the work’s evidential or ethical basis. A clear explanation helps readers understand what failed and what, if anything, remains supported.

97. Version history

A version history records successive forms of a document or file. It can show when wording, data or interpretation changed. Such a record helps collaborators compare revisions and recover earlier material when appropriate. It does not automatically explain why each change was made; explanatory notes may still be needed.

Write retain a version history and compare the revised version. Tricia saves the original headline, the corrected headline and a note explaining the scope change. Later, the group can distinguish a stylistic revision from a factual correction. Version history is especially useful when several people edit one project, but it should not expose private drafts or personal information indiscriminately. The educational purpose is traceable development. A polished final page can conceal both good reasoning and errors that were repaired; a well-kept process record makes the learning available for review.

98. Audit trail

An audit trail is a record that allows decisions or actions to be traced through their evidence and sequence. For a school article, it might connect a published claim to a source note, calculation, draft revision and final check. It is broader than version history because it records the reasoning and supporting material around the document as well as changes to the document itself.

Use a traceable audit trail. Alicia’s ledger includes the exact attendance claim, the record supporting it, the calculation and the person who checked the arithmetic. It does not claim an independent audit when only the authors reviewed their own work. That distinction matters: self-checking is useful but not identical to independent review. An audit trail should make the real process visible rather than manufacture an impression of authority. Record what was done, what was not done and which uncertainties remain open.

99. Editorial judgement

Editorial judgement is the reasoned selection of what to include, emphasise, explain, qualify or withhold for a particular audience and purpose. Evidence does not arrange itself into an article. Editors decide which questions matter, how much context is necessary and whether publication could mislead or unfairly harm someone.

Use exercise editorial judgement and explain the criteria. Tricia removes a sensational but irrelevant personal detail from an interview. The detail might attract attention, but it does not help answer the room-access question and could intrude on privacy. Alicia keeps a less dramatic paragraph explaining the survey’s limitations because readers need it. These choices show that good editing is not simply shortening. It is aligning the content with its reader job while maintaining evidence, fairness and clarity. Strong judgement can also mean deciding that a claim is not ready for publication.

100. Accountability

Accountability means being answerable for decisions and prepared to explain, correct or remedy their consequences. In publishing, it includes maintaining sources, acknowledging errors and responding appropriately when a claim is challenged. Accountability is not merely attaching a name to an article; it involves what the publication does after that name is attached.

Use accept accountability for the error and an accountable process. Kai Kai no longer treats the correction as an embarrassing interruption of the project. It is part of completing the communication responsibly. He explains which claim changed and why, then checks related versions so the same mistake does not remain in the poster. The final vocabulary lesson is practical: precision has consequences. A word such as caused, all or verified asks readers to trust something specific. Accountability means being prepared to show why that trust is deserved and to repair the statement when it is not.

Advanced investigation studio: use the words to change the answer

The following packets are invented teaching materials. Their organisations, quotations, numbers and outcomes are not evidence about real schools, products or platforms. They are deliberately complete enough to support bounded reasoning and incomplete enough to require qualification. Do the task before reading the worked response. The point is not to memorise a model paragraph. It is to learn which features of a new packet should trigger which vocabulary and which next question.

Investigation 1: the reading-room headline

Packet. During the four sessions before redesign, the room recorded forty entries. During six sessions afterward, it recorded seventy-two entries. A pupil who leaves and returns is counted again. The redesign included new wall colour, rearranged chairs and two additional opening sessions. Twelve visitors answered a voluntary survey placed inside the room; nine said they preferred its new appearance. The organiser said, “It feels more welcoming, and I have noticed more people coming in.” Three websites repeated versions of this comment. None collected separate learning data.

Task. Evaluate the headline “New Colour Improves Learning by 80%.” Identify the accurate arithmetic embedded in it, the changes of meaning that make it misleading, and a defensible replacement. Use denominator, operational definition, confounding and corroboration only where each performs a real explanatory job.

Worked reasoning. The entry total increased by thirty-two, from forty to seventy-two. Dividing thirty-two by forty gives eighty per cent. That calculation concerns entries, not learning. The original headline changes the measured outcome from attendance events to educational performance. It also attributes the change to colour even though opening hours and seating changed at the same time. The packet supplies no valid measure of learning and no comparison isolating the colour’s effect.

The denominator problem also appears when comparing activity. Before redesign, there were ten entries per session. Afterward, there were twelve entries per session. The average recorded entries per session therefore increased by twenty per cent, not eighty per cent. Even this adjusted comparison remains limited: the sessions may differ in length, timing or visitors, and repeated entries are not unique pupils. Adjustment improves the description but does not create causal proof. The three websites do not independently corroborate learning improvement because all derive their information from one comment.

Defensible headline. “Reading-Room Entries Rise as Opening Sessions Expand.” A suitable opening sentence is: “Recorded entries increased from forty across four sessions to seventy-two across six sessions after changes to the room and timetable; the records do not measure learning.” The survey can be reported separately as nine of twelve voluntary respondents preferring the new appearance. That result concerns the respondents and the stated question, not the whole school or academic performance.

Why the answer is advanced. It does not merely identify bias. It preserves the correct calculation, diagnoses the invalid interpretation, changes the unit of comparison and states what remains unresolved. Alicia keeps the story because the increased use is worth reporting. Tricia removes the unsupported causal claim. Kai Kai proposes a better follow-up question about access. Their result is not less interesting because it is accurate; it gives the reader a clearer account of what actually changed.

Investigation 2: the survey that doubled support

Packet. A notice says support for a new lunchtime programme doubled. In the first survey, twenty of two hundred respondents chose “strongly support.” In the second survey, twenty-eight of one hundred and forty respondents chose that option. The first invitation went to all classes. The second appeared only in the programme’s participant chat. Other response categories and non-response information are not supplied. The programme coordinator wants to claim that the whole school has become twice as supportive.

Task. Calculate the two reported proportions. Explain the difference between change in the number of strongly supportive respondents and change in their proportion. Then evaluate the population claim. Do not invent the preferences of people who did not respond.

Worked reasoning. Twenty divided by two hundred is ten per cent. Twenty-eight divided by one hundred and forty is twenty per cent. The proportion choosing “strongly support” doubled within these two respondent groups. Its absolute increase is ten percentage points. The count of strongly supportive responses rose by eight, which is a forty per cent increase relative to twenty. Both calculations can be correct because they describe different quantities. Saying “support doubled” without specifying the measure invites confusion.

The larger problem is comparison of samples. The second sampling frame is restricted to participants in the programme’s chat. Those pupils may differ systematically from the wider school. The packet does not show that the two respondent groups are comparable or representative of the same population. Therefore the figures cannot establish that school-wide support doubled. A careful report states the observed proportions and explains the change in recruitment. It should not assume that every participant is enthusiastic, but it should recognise the selection process as a material limitation.

Model evaluation. “The reported proportion of strongly supportive responses rose from 10% to 20%, an increase of ten percentage points. However, the surveys used different recruitment methods, with the second limited to a participant chat. This creates a selection concern and prevents a straightforward conclusion about change across the whole school. The result may reflect both changing views and changing respondents; the packet does not separate them.”

Transfer. Replace the programme with a product review, museum event or optional workshop. The same reasoning applies. First identify the quantity, then the denominator, then the population and selection method. More advanced vocabulary does not require a more complicated calculation. It requires keeping the meaning of the calculation intact as it moves into prose. Tricia’s final question is practical: what would a follow-up survey need to keep comparable? The answer concerns invitation, wording, timeframe and population, not merely a larger response target.

Investigation 3: two clubs, one misleading combined rate

Packet. A school offers a beginner workshop and an advanced workshop. In one term, Programme A has eight successful completions out of ten beginners and eighteen out of thirty advanced participants. Programme B has twenty-one successful completions out of thirty beginners and five out of ten advanced participants. The tasks differ in difficulty. A promotional article compares only total successful completions: A has twenty-six out of forty; B also has twenty-six out of forty. It concludes that the programmes perform identically in every relevant respect.

Task. Compare the within-workshop completion rates and the combined rates. Explain why an identical overall percentage does not establish identical performance for each subgroup. Then explain why the subgroup comparison still does not, by itself, prove that Programme A caused better outcomes.

Worked reasoning. Programme A’s beginner rate is 80%, compared with 70% for B. A’s advanced rate is 60%, compared with 50% for B. A has a higher observed completion rate within each workshop type. Both combined totals are 65%, because the groups contain different proportions of beginner and advanced participants. Programme A has more people in the harder workshop, while Programme B has more beginners. Combining the groups hides a relevant difference in their composition.

This does not mean the combined rate is mathematically wrong. It answers the question “What proportion of all recorded participants completed successfully?” It does not answer the separate question “How do completion rates compare within the same workshop type?” The appropriate summary depends on the reader’s question. A publication reporting both can avoid a false equivalence without claiming that one statistic should never be used.

Nor does the packet establish causation. Participants may differ in prior preparation, attendance, selection or support. The completion criteria also need to be comparable within each workshop type. A careful conclusion reports the pattern and its limitations: “The programmes share the same overall completion rate, but A has higher observed rates in both stated workshop categories. Different participant composition explains why the overall figures conceal that pattern. Further information is needed before attributing the differences to programme quality.”

Vocabulary gain. Use composition, subgroup, denominator, scope and confounding accurately. These extension words do not increase the numbered list; they help you connect its concepts. Kai Kai initially wants a winner. Alicia instead asks which comparison the article promised. Tricia checks that the promised comparison and supplied statistic match. The result is a report that teaches readers how to interpret the evidence rather than handing them a simple ranking unsupported by the design.

Investigation 4: the genuine photograph with the wrong date

Packet. A photograph shows a crowded room during a special author visit. The original archive entry dates it to the previous year. A new social post presents a cropped version with the caption “Every lunchtime this week.” The crop removes the event banner and most of the surrounding room. The image is genuinely associated with the school, and no evidence of face replacement or generated people is provided. A commenter says that because the photograph is genuine, the new caption must be reliable.

Task. Separate authenticity, integrity and context. Identify the claim the original photograph could support and the claim it cannot support without additional evidence. Write a correction that does not accuse the photographer of fabrication.

Worked reasoning. The photograph may be authentic as a record of the named school and earlier event. That does not verify its presentation as evidence of every lunchtime in the current week. The problem is recontextualisation: a genuine item has been attached to a different date and pattern of activity. Cropping removes context that might otherwise alert readers to the special event. Neither problem requires a deepfake. The image’s origin and the caption’s truth are separate questions.

The phrase “every lunchtime” is especially strong. One photograph of one moment cannot establish repeated conditions across several days. Even a photograph taken this week would need appropriate context before supporting that generalisation. A careful article might use the image as an explicitly labelled archive illustration, but not as measurement of ordinary attendance. The publication should avoid implying that the original photographer created the misleading new use unless evidence shows that responsibility.

Model correction. “The image in our earlier post showed a special author visit from the previous year, not ordinary lunchtime attendance this week. We have corrected the caption and removed the unsupported generalisation. Current attendance should be assessed from appropriately dated records rather than this archive photograph.” This correction identifies the error, supplies the right context and explains the appropriate evidence.

Transfer. The same structure applies to a real video with a false location, an accurate quotation placed in an unrelated debate, or a correct statistic from a different year. The key question is not only whether the item exists. It is whether it supports the particular claim now being made. Alicia adds an “item–caption fit” check to the publication process. That is a specific procedural improvement arising from the vocabulary, not a generic instruction to be careful online.

Investigation 5: the sponsored pencil-grip trial

Packet. A fictional stationery supplier provides thirty students with a new pencil grip. Students volunteer after seeing an enthusiastic demonstration. They copy a passage before the demonstration and another afterward. Different passages are used. The supplier chooses five attractive examples for its advertisement and reports, “Our grip makes every student’s writing clearer.” The complete set of pages, scoring criteria and comparison group are not supplied. The advertisement names the supplier openly.

Task. Identify at least four distinct limitations. Explain why open sponsorship is useful information but does not solve the methodological problems. Then propose a narrower statement the supplied evidence might support and a better school-level follow-up design.

Worked reasoning. The volunteer group may not represent every student. Different passages may vary in difficulty. Practice, attention or the demonstration could influence the second attempt. Selecting only five favourable examples creates a selection problem in the public presentation. Without a clear operational definition of “clearer,” the judgement may depend on subjective preference. The universal phrase “every student” is unsupported by the selected examples even if those examples are genuine.

Disclosure of sponsorship improves transparency because readers can see an interest in positive results. It does not make the trial valid, independent or representative. Equally, sponsorship alone does not prove the product has no benefit. A fair evaluation addresses the actual design and wording rather than treating commercial involvement as a substitute for analysis. The narrow supported statement is that the supplier displayed five examples it judged favourable after a demonstration. That is a much weaker claim than a universal causal effect.

A follow-up could use a teacher-approved, low-risk comparison with clear scoring criteria, comparable tasks, appropriate order control and complete reporting of results. The design should avoid collecting unnecessary identifying information. The exact research procedure would need to fit the school setting. Students can also analyse a supplied fictional dataset instead of testing classmates. The vocabulary goal is to recognise how selection, measurement and comparison shape a conclusion, not to claim that a small classroom exercise can certify a commercial product.

Model paragraph. “The advertisement overgeneralises from a selected set of examples. Its volunteer sample and changing tasks limit the comparison, while the absence of stated scoring criteria makes the outcome difficult to evaluate. Sponsorship is disclosed, but disclosure does not provide independent corroboration or isolate the grip’s effect. The evidence may justify further investigation, not the claim that every student benefits.” Notice the precise verbs: limits, overgeneralises and does not isolate. They diagnose the problem without claiming more than the packet establishes.

Investigation 6: the campaign with excellent engagement

Packet. A school publication promotes a reading event. Its reporting tool records one thousand impressions, four hundred reached accounts, one hundred link clicks and twenty completed registration forms. Sixty comments appear under the post. Forty ask whether the date is Friday or Saturday because the poster and caption disagree. The organiser announces that the campaign achieved “excellent understanding and overwhelming support.” No comprehension check or classification of comments was completed.

Task. State what each metric can reasonably describe. Calculate the registration rate among recorded link clicks. Explain why the comments are not direct evidence of support and why the denominator must be named when reporting a rate. Recommend one correction and one better measure of the communication goal.

Worked reasoning. Impressions describe recorded displays; reach describes distinct accounts under the tool’s rules; clicks describe recorded link interactions; completed forms describe registrations under the form’s rules. Twenty forms divided by one hundred clicks gives twenty per cent. That is a forms-to-clicks rate, not necessarily a people-to-people conversion probability, because the packet does not establish whether clicks are unique or whether every form came through that link.

The comment total indicates interaction but not approval. In fact, the supplied comment content identifies confusion about a crucial detail. Calling the interactions evidence of excellent understanding reverses their meaning. The campaign should first correct the inconsistent date in every controlled version of the message. A useful follow-up might ask a small, appropriately selected group to state the event date and registration action after viewing the corrected material. Such a check has its own sampling limits but is more directly connected to understanding than a raw comment count.

Defensible report. “The campaign generated substantial recorded interaction and twenty completed registrations. However, many comments sought clarification because the poster and caption gave different dates. Engagement should therefore not be interpreted as evidence of clear understanding. The team corrected both versions and will assess whether readers can identify the event details accurately.” This is a synthesis of activity data and qualitative evidence, not a rejection of analytics.

Transfer. Apply the same distinction to a popular video, a frequently downloaded worksheet or a widely shared petition. Distribution, interaction, comprehension and agreement are different outcomes. A publication should choose metrics that fit its purpose rather than redefining success after seeing which number looks strongest. Kai Kai’s vocabulary changes the project’s decision: instead of celebrating confusion as engagement, he repairs the message that produced it.

Investigation 7: the source that should not be discarded

Packet. A student writes a personal account of difficulty accessing the reading room because her club meets during its opening sessions. Her experience is accurately recorded with appropriate permission. A second student says the account is “only anecdotal” and should be removed because it cannot represent everyone. An official schedule confirms that the club and room opening times overlap, but the packet contains no school-wide survey of access problems.

Task. Explain the account’s legitimate evidential role and its boundary. Write a paragraph that uses the experience without generalising it to the entire school. Identify what additional evidence would answer a population-level question about access.

Worked reasoning. The personal account can provide evidence about that student’s reported experience, and the schedule corroborates the relevant timing conflict. It cannot establish how common the difficulty is across the school. Removing it solely because it is not a population estimate would confuse two different research questions. Evidence can be useful for identifying a problem or understanding an experience without measuring its prevalence.

Model paragraph. “One interviewed pupil reported that the room’s opening hours conflict with her club schedule. The published timetable supports that specific access difficulty. Her account does not establish how many other pupils face the same problem, but it identifies a relevant issue for further investigation. A broader assessment would need to examine schedules and experiences across an appropriately defined population.” The paragraph makes a limited concession to the critic while preserving the account’s value.

The follow-up should not begin with the assumption that every pupil is affected. It could compare activity timetables, investigate alternative sessions and gather appropriately approved feedback from relevant groups. The aim is to determine scale and possible solutions. Tricia recognises that methodological vocabulary can be misused as a weapon to silence inconvenient experiences. “Anecdotal” describes a type of evidence; it does not mean imaginary or worthless. Advanced evaluation gives each item the role it can support rather than demanding that every source answer every question.

Transfer. A customer complaint, a historical diary and an interview with a community member raise similar questions. What can the source reveal directly? What broader claim would require additional evidence? This exercise develops fairness alongside precision. The publication should neither turn one experience into a universal conclusion nor erase the experience because it is not universal. Both errors come from failing to match evidence to scope.

Investigation 8: the apparently contradictory sources

Packet. The room manager reports that total visits rose in the new term. A survey reports that the proportion of respondents who visit every week fell. An interviewee says that the room seems quieter. Kai Kai writes, “The sources contradict each other, so none can be trusted.” The records later show that the term contains more open sessions, the survey recruited a wider group than before, and the interview took place during an early session rather than lunchtime.

Task. Determine whether the three claims necessarily contradict one another. Identify differences in outcome, denominator, sample and timeframe. Then write a synthesis that preserves the useful information without forcing agreement or declaring universal unreliability.

Worked reasoning. Total visits can rise while the proportion of a wider respondent group reporting weekly use falls. The measures have different denominators and may concern different populations. An individual early session can seem quiet even if total term attendance rises. The observations do not logically conflict simply because one sounds positive and another sounds negative. Their exact definitions and contexts matter.

A synthesis might say, “The records indicate more total visits across a term with more opening sessions. The survey concerns the proportion of a differently recruited respondent group reporting weekly use, while the interview describes one early session. These sources address different aspects of use and should not be treated as direct contradictions. Comparable definitions and sampling would be needed to assess a single overall trend.” This conclusion is neither automatic trust nor automatic dismissal.

Why this matters for reading. Texts often use broad words such as improvement, popularity and success while measuring different things underneath. An advanced reader reconstructs the operational definitions before comparing conclusions. Alicia adds a column labelled “What exactly is measured?” to her source notes. It prevents her from arranging sources into opposing camps solely because their emotional tone differs. The task is to understand the relationship among claims, not to stage a debate where the evidence does not support one.

Transfer. Two reports about a business may concern revenue and profit; two reports about a food product may concern preference and texture. Both can be correct while answering different questions. This is where interdisciplinary vocabulary becomes useful. The habit of defining the outcome transfers across subjects even though the specialist nouns change.

Sentence workshop: precise words in complete reasoning

A vocabulary entry becomes productive when the learner can place it naturally in a sentence that explains something. Begin with a plain explanation, then introduce the technical word only when it improves precision. Do not write “The corroboration is a provenance of credibility.” The words are related, but that arrangement does not express a coherent relationship. A better sentence is: “The second report does not provide independent corroboration because its information comes from the first report.”

Repair 1: infer and imply

Weak: “The reader implies that the organiser is worried.” Repair: “The reader may infer that the organiser is worried from her repeated requests for confirmation.” The reader performs the inference. The wording or behaviour supplies the implication. The repaired sentence also gives evidence and uses an appropriate hedge. It does not diagnose a real person’s mental state from a fragment; it analyses the fictional text.

Repair 2: evidence and prove

Weak: “The three websites prove the result.” Repair: “The three websites repeat the same result, but their common source means they do not provide independent confirmation.” The repair changes both noun relationship and verb strength. It preserves what is known—the repetition—while removing the unsupported leap to proof. A strong answer can be critical without using the strongest possible negative label.

Repair 3: percentage language

Weak: “Support increased by ten per cent, from forty per cent to fifty per cent.” Repair: “Support increased by ten percentage points, from 40% to 50%; relative to the original rate, that is a 25% increase.” The arithmetic should be checked before the prose is polished. Once the correct quantity is identified, choose the wording that preserves its denominator and reference value.

Repair 4: reliability and validity

Weak: “The attendance count is invalid because two observers agree.” Repair: “Agreement between observers supports the consistency of the attendance count, but attendance is not by itself a valid measure of comprehension.” The original sentence confuses the evaluation dimensions. The repaired version shows how a measure can be useful in one respect while insufficient in another. That structure is more advanced than rejecting or approving the entire dataset.

Repair 5: sponsorship and falsity

Weak: “The sponsor paid, so the findings are false.” Repair: “The sponsorship creates a potential conflict of interest, making transparent methods and independent scrutiny important; the findings still require evaluation on their evidence.” This does not excuse weak research. It directs criticism towards what can be demonstrated. A conflict of interest is a reason to investigate the process, not a logical proof of a particular outcome’s falsity.

Repair 6: authenticity and context

Weak: “The photograph is authentic, therefore the caption is accurate.” Repair: “The photograph’s origin may be authentic, but the caption’s date and description need separate verification.” The connective therefore is the danger point in the original. A technical term cannot make an invalid inference valid. Always inspect the relationship the connective asserts between the two clauses.

Repair 7: a useful concession

Weak: “The interview is only one person, so it is useless.” Repair: “Although the interview cannot establish how common the experience is, it provides evidence about that participant’s account and identifies a question for further investigation.” The concession acknowledges the limitation before explaining the remaining value. This structure is useful in source-based essays because it avoids treating evidence quality as a single all-or-nothing score.

Repair 8: scope in the subject of a sentence

Weak: “Students prefer the new opening hours.” Repair: “Nine of the twelve voluntary respondents preferred the new opening hours.” A scope error can occur before the main verb. The unqualified subject students suggests a population-wide claim. Replacing it with the observed group fixes the meaning without adding a cloud of hesitant language. Precision often comes from a better noun phrase rather than another adverb.

Word families and collocations: building sentences that sound natural

Corroborate → corroboration → corroborative. Write “The register corroborates the reported count,” “We need independent corroboration,” and “The photograph provides corroborative evidence for the room’s layout.” Notice that the object of the verb is a claim, account or detail. The adjective should not be added to evidence that merely repeats the same source. Grammar and reasoning work together: the sentence structure asks what, exactly, is being supported.

Infer → inference → inferential. “We infer a possible concern,” “The inference is limited,” and “The report makes an inferential step beyond the direct observations” express related jobs. Avoid treating inferential as a synonym for uncertain in every case. Some inferences can be strongly supported. What matters is the connection from evidence to conclusion and whether alternatives remain plausible.

Qualify → qualification → qualified. “Qualify the claim,” “State an important qualification,” and “Offer a qualified conclusion” describe limiting a statement. In another context, a qualified teacher has a recognised qualification. The shared word family has different senses. Advanced vocabulary learning should include sense selection, not only morphological change. Read the whole sentence before choosing a replacement word.

Represent → representation → representative. A chart represents data; an article creates a representation of an event; a sample may be representative of a defined population. These meanings are connected but not interchangeable. “The sample represents fifty students” may mean that it contains fifty records, not that it is representative of all students. Spell out the intended relationship when ambiguity matters.

Verify → verification → verifiable. A claim is verifiable when there is a way to check it; verified means the relevant checking has actually been completed to a stated standard. Do not promote possibility into completion. Alicia writes “The date can be checked against the archive” before she checks it, and “The archive confirms the earlier date” afterward. Tense and aspect can reveal whether a claim about process is honest.

Synthesise → synthesis. Synthesis usually takes several sources or strands of evidence as input. It does not mean simply extending one source with invented detail. A natural sentence is “The paragraph synthesises the attendance record and interview evidence while retaining their different scopes.” The verb tells the reader that the writer has organised relationships, not merely copied a sequence of quotations.

Accountable → accountability. A publication is accountable for its reporting; an accountability process makes responsibility and repair possible. Do not confuse accountability with certainty. An accountable team can acknowledge that an answer remains unresolved. The difference is that it records its reasons, does not conceal known limitations and responds appropriately when better evidence changes the conclusion.

Worked writing: from a weak paragraph to an evidence-led argument

Original student paragraph. “The redesigned room is clearly a huge success because three websites say learning improved by eighty per cent. Most students love it, and the photograph shows a crowded room. Although some people complain about the hours, they are probably just resistant to change. Therefore the school should spend more on the same colour in every classroom.”

Before rewriting, diagnose the jobs the paragraph tries to perform. It reports a result, cites apparent corroboration, generalises preferences, uses an image as evidence, dismisses an objection and proposes a policy. Each job requires support. The paragraph’s fluency hides the number of unsupported transitions. A list of synonyms for success will not repair them. The first revision should address reasoning and scope, not vocabulary density.

The opening phrase “clearly a huge success” needs criteria. Success at what: increasing visits, improving appearance, widening access or improving learning? The cited eighty per cent concerns entries, not learning. The three websites are dependent accounts, not independent corroboration. “Most students” changes the survey’s denominator from twelve respondents to the school population. The photograph comes from an earlier special event. The sentence about complaints invents motives and ignores a potentially relevant access problem. The final policy proposal extends the claim across rooms and users without evidence about cost or effectiveness.

Revised model. “The available evidence supports a narrower conclusion than the claim that repainting improved learning. Recorded entries increased from forty across four sessions to seventy-two across six sessions, while the room and timetable changed together. The average entries per session rose from ten to twelve, but the records count visits rather than unique pupils or comprehension. Three websites repeat the same organiser’s comment, so they do not provide independent corroboration. Nine of twelve voluntary respondents preferred the new appearance; this is useful feedback from that group, not evidence of school-wide preference. Before extending the design to other classrooms, the school would need clearer criteria, appropriate learning evidence, information about access and a realistic assessment of costs.”

The revised paragraph makes a conclusion, supplies evidence, explains its limits and identifies the next decision need. It does not pretend the redesign failed. It distinguishes what improved in the records from what remains unmeasured. Notice how ordinary and technical language cooperate. Average entries per session is precise but accessible. Independent corroboration adds a useful distinction. The final sentence turns uncertainty into a research and decision agenda rather than ending with “more research is needed” without saying what kind.

Extension task. Write the strongest fair argument for one additional opening session using only the packet. You may use the access concern and attendance pattern, but you must acknowledge that staffing and cost information are missing. Then write a reasonable counterargument. Finally, produce a limited recommendation that states which additional evidence would determine the decision. This sequence trains argument, counterargument and qualification without forcing students into a predetermined opinion.

A possible recommendation is to investigate a small, approved trial of an additional session where the timetable conflict is greatest, while recording demand and resource requirements. That is a proposal, not a finding. Its justification concerns learning what users need, not claiming that the colour has already proved effective. Keeping proposals separate from evidence prevents future plans from being reported as completed results.

Independent assessment: a new evidence packet

Fictional packet: the revision-planner launch. A student team introduces a printable revision planner. Forty volunteers receive it after attending a study-skills talk. Thirty return a questionnaire. Twenty-four respondents say the planner helped them organise tasks. The team compares the volunteers’ average score on a ten-question practice exercise before the talk with their average score on a different ten-question exercise one week later. The average rises from six to seven. The exercises were written by different people and were not checked for comparable difficulty. No group without the planner is included. A sponsor paid for printing but did not collect the responses.

The team publishes a poster: “Planner Raises Results by 17%: Proven by Student Research.” The article underneath says that eighty per cent of respondents found the planner useful for organisation. A social account repeats the poster, and another blog cites that social account as an independent confirmation. A photograph shows pupils holding planners after the talk. The post records eight hundred impressions, three hundred reached accounts and ninety comments. Fifty comments ask how to obtain the planner; twenty question the result; twenty are unrelated reactions.

Questions. First, identify the population the poster appears to address and the sample actually observed. Second, calculate the response rate among the forty volunteers and the positive-response proportion among respondents. Third, explain the arithmetic behind the rounded seventeen per cent figure. Fourth, identify two reasons it does not establish the planner’s causal effect on results. Fifth, evaluate the word proven. Sixth, explain whether the blog supplies independent corroboration. Seventh, state what the photograph can and cannot establish. Eighth, distinguish the three types of comments. Ninth, write a corrected headline. Tenth, write a 180–250-word evaluation that preserves the useful findings.

Answer discussion: population, response and arithmetic

The poster appears to make a broad claim about students’ results, although the actual observed group consists of forty volunteers who also attended a talk. Thirty of forty returned the questionnaire, giving a response rate of seventy-five per cent within that volunteer group. Twenty-four of thirty respondents reported help with organisation, giving eighty per cent. Twenty-four of the original forty volunteers is sixty per cent. The last figure should not be interpreted as a complete positive-response rate for all volunteers because the preferences of the ten non-respondents are unknown.

The average practice score rose by one point, from six to seven out of ten. One divided by six is approximately 16.7%, which rounds to seventeen per cent. That is the relative increase in the stated averages. It is not seventeen percentage points. Expressed as percentages of the maximum score, the averages are sixty and seventy per cent, a difference of ten percentage points. The arithmetic can be correct while the causal headline remains unsupported.

Answer discussion: causation and evidence roles

The talk, practice, passage difficulty and other changes could contribute to the score difference. The exercises were not shown to be comparable, and no suitable comparison isolates the planner’s effect. The volunteer group may also differ from the wider population. These limitations mean the packet does not establish that the planner caused the score change. The word proven gives excessive certainty, while the broad term results expands a particular practice comparison into a general educational outcome.

The blog does not provide independent corroboration because it cites a social account repeating the team’s poster. The photograph supports that pupils held planners at the photographed moment, subject to its provenance and context. It does not measure organisation or learning. Sponsorship should be disclosed, but the packet does not justify claiming that the sponsor manipulated the responses. The relevant concern is transparency about the relationship and scrutiny of the actual methods.

Answer discussion: audience metrics and headline repair

The comments contain requests, criticism and unrelated reactions. Their total measures interaction, not uniform approval. Fifty requests may indicate interest in obtaining the planner, but they do not establish completed use or benefit. The impressions and reach figures describe exposure under the reporting system’s rules, not proven full reading. A suitable headline is “Volunteer Respondents Report Better Organisation with Revision Planner.” A more compact alternative is “Planner Feedback Is Positive; Learning Effect Remains Unclear.” Both need the article to explain the sample and outcome.

Model evaluation. “The packet provides useful feedback about organisation but does not prove a causal improvement in learning. Thirty of forty volunteers returned the questionnaire, and twenty-four of those thirty reported that the planner helped them organise tasks. This supports an eighty per cent positive-response figure among respondents, not a claim about all students. The average practice score rose from six to seven, but the two exercises were not shown to be comparable. Participants also attended a study-skills talk, so the design cannot isolate the planner’s effect. The blog repeats the team’s claim through a social account and therefore adds no independent corroboration. The photograph documents distribution at one moment, not educational benefit. A responsible report would distinguish the positive organisational feedback from the unresolved learning claim, disclose the printing sponsorship and explain the need for a more suitable comparison.”

Marking principle. Credit a correct distinction only when the explanation fits the packet. Writing the words bias, validity and corroboration without identifying the actual selection, measurement or source relationship should not receive the same credit as a precise analysis. Equally, an answer in plain English that correctly explains the problem demonstrates understanding. The vocabulary should make that understanding easier to express, not become an obstacle between the learner and the reasoning.

Twenty retrieval challenges with reasons

1. Three sites copy one press release. Which word identifies the missing property? Independence. The sites add no separate evidence for the copied claim. Corroboration requires attention to what each source contributes, not simply how many URLs appear in a reference list.

2. A report moves from twelve respondents to “all pupils.” What changed? Scope and denominator. The observed group has been replaced by a larger population without adequate support. Repair the subject of the sentence before adding another hedge.

3. A percentage rises from 30% to 36%. What is the absolute difference between the rates? Six percentage points. The relative increase is twenty per cent because six is one fifth of thirty. The answer must name which comparison is being made.

4. A genuine old photograph is labelled as today’s event. What is the central problem? Recontextualisation. Authenticity of the original file does not verify the new date or caption. A deepfake is not required for an image to mislead.

5. A questionnaire asks why the successful programme should expand. What should you inspect? Presupposition. The wording assumes success before asking about expansion. A neutraler question would separate evaluation of the programme from views about future action.

6. An attendance measure is recorded consistently but used to claim comprehension. Which distinction matters? Reliability versus validity. Consistent counting can support attendance reporting without measuring the learning outcome. Do not discard the count; limit its job.

7. The room is repainted and opening hours change together. What complicates the causal story? Confounding. The comparison cannot isolate colour when another relevant factor changes at the same time. The alternative explanation should be named explicitly.

8. Pupils who already enjoy reading visit more often. Which alternative direction should be considered? Reverse causality. Existing reading habits may influence visits, rather than visits alone causing stronger reading habits. The relationship could also work in both directions.

9. A source has an attractive seal but no clear ownership. What reading move helps? Lateral reading. Investigate the organisation outside its own page. Do not treat an unsuccessful search as proof of dishonesty; record the uncertainty.

10. A source is funded by a supplier. What should be disclosed? A potential conflict of interest. The relationship matters to evaluating the process, but it does not logically prove that every finding is false.

11. A writer restates a source but removes “some” and “may.” What has gone wrong? The paraphrase strengthens the claim. Preserving meaning includes preserving qualifications, not merely changing vocabulary or sentence order.

12. A document lists sources without showing which claim each supports. What is weak? Citation traceability. A bibliography can be useful, but the reader still needs a clear connection between important assertions and their relevant evidence.

13. A personal interview cannot measure prevalence. Must it be removed? No. It may provide evidence of experience or identify a problem. Its scope should be limited rather than its value erased.

14. A critic invents a foolish version of the opposing view. What should the writer do instead? Represent the strongest relevant counterargument fairly, then respond to its actual premises or consequences. An easy rebuttal to a distorted view is not a strong argument.

15. A writer accepts one objection while retaining a narrower proposal. What is this? A concession. It can improve feasibility and fairness without abandoning the entire position.

16. One account produces several displays of a post. Which metric can increase without another unique account? Impressions. Reach and impressions should not be treated as interchangeable counts of readers.

17. A post attracts angry comments. Do they prove support? No. Engagement measures defined interactions. Their meaning requires examining what the interactions contain and how they relate to the communication goal.

18. A file matches an earlier exact-file hash. What remains unresolved? Whether the content’s claims are true. An integrity or identity check does not establish factual accuracy, context or representativeness.

19. A survey asks for addresses although only aggregate preferences are needed. Which principle applies? Data minimisation. Collect only what the defined purpose requires, using the school’s approved privacy and safeguarding process.

20. A publication discovers a material error. What completes the responsibility? An appropriate correction, an accurate explanation and checks of related versions. Accountability continues after the original page is published.

A six-week advanced vocabulary sequence

Week 1: claim structure. Work with Terms 1–10 and one short paragraph each session. Identify the proposition, premises, conclusion and warrant before rewriting. The student should produce a plain-language explanation first. Success means the reasoning is visible, not that every technical word appears. Finish the week by comparing two different phrasings of the same proposition and explaining whether the meaning has changed.

Week 2: evidence history. Use Terms 11–30 with a teacher-prepared source chain. Trace original material, secondary accounts and repeated reporting. Keep a three-column record: what the source is, what it supports and what it cannot establish. Finish with a short explanation of why three pages may contain only one evidential origin. The teacher should reward accurate boundaries, not blanket distrust.

Week 3: wording and numbers. Combine Terms 31–50. Students analyse a headline’s framing, then inspect the figures beneath it. Practise absolute changes, relative changes and percentage points using fictional data. Require each numerical answer to be followed by one complete sentence stating the population, measure and timeframe. This prevents arithmetic from becoming detached from meaning.

Week 4: causation and platforms. Use Terms 51–70 to examine before-and-after stories and campaign metrics. Ask students to generate plausible alternatives without declaring those alternatives proven. Then separate exposure, interaction, agreement and understanding. The output is a short evaluation with one supported conclusion, one limitation and one useful next question.

Week 5: digital evidence and writing. Combine Terms 71–90. Analyse teacher-provided images or verbal image descriptions with clear provenance. Practise accurate paraphrase, bounded quotation and synthesis. Do not require students to investigate private people or create impersonations. Finish with a model paragraph that explains why an authentic item can be used in a misleading context.

Week 6: publication and transfer. Use Terms 91–100 in a small approved project. Plan audience, sources, privacy, accessibility, review and correction. The final product may remain within the classroom. Public posting is not necessary to demonstrate learning. Evaluate both the finished article and the process record: did the student revise a claim when evidence required it, and can another reader trace the important statements?

Between sessions, retrieve a small number of terms without looking. Ask for meaning, a near-neighbour contrast and a fresh example. Revisit errors after a delay rather than immediately repeating a definition until it feels familiar. This is a proposed teaching routine, not a validated promise that every learner will master the collection in six weeks. Adjust the pace according to the learner’s explanations and transfer performance.

Assessment rubric: recognise, distinguish, apply, transfer

Recognition. The learner identifies a term in a familiar example. This is a starting point. Matching corroboration to a definition does not show whether the student can detect dependent sources in a new packet.

Distinction. The learner explains a relevant boundary: authenticity versus truthful context, reliability versus validity, or percentage change versus percentage points. Give credit for an accurate explanation in ordinary language. Technical vocabulary should sharpen the explanation rather than replace it.

Application. The learner uses the term to diagnose a specific feature and revises the claim accordingly. “There may be selection bias because the invitation reached only current visitors” is application. “The source has bias” without a mechanism is only a label.

Transfer. The learner applies the distinction to a new topic without copying the original model. A pupil who can analyse the room survey should also recognise a denominator change in a product review or a food-preference test. Transfer is the strongest sign that the vocabulary has become a reasoning tool.

Responsible judgement. The learner avoids unsupported accusations, protects private information, represents counterarguments fairly and states uncertainty proportionately. An analysis can be technically sharp yet irresponsible if it exposes classmates or invents motives. The advanced standard includes both precision and care.

Teacher and parent guidance

Keep the work focused on claims rather than personalities. A teacher can ask, “What does this evidence establish?” without asking students to decide which classmate is trustworthy. Use fictional packets or public materials appropriate to the learners’ age. Avoid assignments that reward collecting private screenshots, identifying anonymous people or publishing allegations. The same reasoning skills can be taught through safe, bounded cases.

When a student uses a technical term incorrectly, first recover the intended meaning. Ask them to explain the problem without the term. If the explanation is sound, teach the correct label and collocation. If the explanation is weak, return to the evidence relationship. Correcting pronunciation or spelling alone will not repair a confused distinction. Conversely, a student who understands the concept should not be held back merely because the specialist noun is temporarily unavailable.

Parents can support the work through one question at a time: “What exactly is the claim?” “Who observed that?” “Could these reports share one source?” “Percentage of what?” “What would change your conclusion?” Avoid turning every ordinary conversation into an interrogation. The aim is a repeatable habit of thoughtful checking when the consequence matters, not anxiety about every message a child encounters.

Do not praise a learner only for catching an error. Also praise a fair concession, a correctly supported conclusion and a decision not to overclaim. A student who concludes that a source is useful within limits has done more sophisticated work than one who calls everything unreliable. The vocabulary should support informed trust as well as scepticism.

Frequently asked questions

Is advanced vocabulary mainly about longer words? No. Words such as scope, claim and sample are short but become advanced when used precisely. The challenge is selecting the right meaning, explaining a boundary and applying it to unfamiliar evidence.

Should students use all one hundred terms in an essay? No. That would usually make the essay worse. Select the few terms that clarify the reasoning. A paragraph using three concepts accurately is stronger than a paragraph crowded with specialist nouns that do not connect to evidence.

Does source evaluation mean distrusting everything? No. It means matching trust to evidence and purpose. An official timetable may be the right source for opening hours, while a personal interview may be the right source for an individual’s experience. Neither source has to answer every possible question.

Can a biased source contain true information? Yes. Selection and framing can coexist with accurate details. Check the specific claim, its supporting evidence and the information needed for context. Do not treat agreement with your own view as evidence of neutrality.

What should a student do when sources disagree? First check whether they measure the same outcome, population and period. Then compare methods and evidence. Some apparent disagreements disappear when definitions are clarified; others remain and should be reported honestly.

Can generated text be used in a school research task? Follow the task’s rules. Generated wording does not verify its own factual claims or citations. Any permitted assistance still requires the student to understand the argument, check sources and represent the contribution honestly. Do not submit assistance as independent work when the rules prohibit it.

How does this differ from the foundation glossary? The foundation page introduces broad media terminology. This extension focuses on evidential boundaries, causal reasoning, quantitative interpretation, synthesis and accountable publication. It asks students to repair claims and defend conclusions, not simply recognise labels.

Continue through the collection

Use the Secondary 2 Advanced Vocabulary Collection to find the companion extension manuals. Food-science investigation applies measurement, validity and evidence to ingredient behaviour. Entrepreneurship applies assumptions, denominators and uncertainty to customer research and unit economics. The subject changes; the discipline of making a claim fit its evidence remains.

For broader routes, the Vocabulary Article Directory organises related reading, while the English Vocabulary Lists page provides grade and word-list entry points. Keep those different jobs in mind: a list supplies language, a learning method helps it become usable, and an investigation tests whether the language transfers into judgement.

Reference shelf and limits of the examples

The external references linked in this manual provide different kinds of support. The UNESCO media and information literacy framework supplies broad educational context. The Digital Inquiry Group provides explicit lateral-reading instruction. C2PA explains the limits and role of technical provenance information. W3C provides accessibility guidance. None of those organisations has reviewed or endorsed the fictional cases or this vocabulary collection.

The calculations, source chains, survey designs, model paragraphs and assessment answers here are original teaching constructions. They should not be quoted as findings about real learners or products. Where a task moves into real publication, current school rules, safeguarding requirements and relevant local guidance take priority over a general classroom example.

The final test: can the sentence survive checking?

Alicia returns to the headline that began the project. She can still make it clear and engaging, but she no longer asks a photograph to prove learning or three copied pages to become three independent witnesses. Tricia’s source notes now identify which claim each item supports. Kai Kai’s next question is not how to make the percentage larger. It is what the percentage actually describes.

That is the practical meaning of advanced vocabulary. The learner can name the claim, trace the evidence, calculate the comparison, recognise the limitation and write a conclusion that another person can inspect. The sentence does not become weaker when it becomes checkable. It becomes something a reader can responsibly use.