Reading claims and comparisons

Carry the boundary with the claim

A passage can report a precise figure while supporting only a limited conclusion. “Twenty of the eighty returned forms mentioned noise” describes returned forms, not all residents and not all forms distributed. Keep the reference group attached to the number when you paraphrase. Replacing “respondents” with “residents” expands the claim, even if the new sentence sounds natural. The same discipline applies to a time period, location or eligibility condition. Those boundaries tell you what the evidence measures.

Start by identifying four parts of a comparison: what is measured, among whom, during which period, and against which baseline. A claim that morning deliveries became faster may concern a different set of parcels from a claim about all deliveries. If the passage does not connect those groups, do not silently merge them. This is a reading task before it is an arithmetic task. A perfectly calculated percentage can still answer the wrong question if its denominator comes from the wrong population.

Separate a count from a share

An absolute number and a proportion can move in different directions. Suppose one centre receives 12 complaints among 120 visits, while another receives 15 among 300 visits. The second has more complaints but a smaller complaint share: 5% rather than 10%. A passage comparing counts alone cannot establish which has the higher rate without the visit totals. Conversely, a passage giving rates alone need not establish which has more cases without the group sizes.

Notice also what the rate counts. “Complaints per visit” is different from “visitors who complained” when one visitor can make several visits or complaints. Do not replace one measure with another unless the text equates them. A small phrase such as “at least once” can change the counting unit. If a question asks what must be true, an answer depending on every visitor making exactly one visit requires an extra assumption that may not be available.

Read changes with their starting point

“A higher number than last month” is not the same as “a faster increase than last month”. The first compares levels; the second compares changes. Imagine membership moves from 100 to 130 and then to 150. The level rises in both steps, but the absolute increase falls from 30 to 20. A claim that membership growth accelerated would misdescribe those numbers. Before selecting an option, decide whether its words refer to the level, the difference or the percentage change.

The baseline matters for percentage language too. An increase from 40 to 50 is 10 additional units, which is 25% of the original 40. A decrease from 50 to 40 is 20% of 50. The same numerical difference can have a different percentage when the reference value changes. If the passage says only that a share rose from 20% to 25%, the rise is five percentage points. Calling it five per cent confuses the difference in shares with a relative change.

Distinguish observation from explanation

A passage may report that satisfaction rose after a new reception desk opened. That is an observation about timing. It does not by itself establish that the desk caused the increase. If staffing levels or the visitor mix also changed, they may matter. For a reading inference, preserve any uncertainty the writer expresses. “May explain” cannot become “fully explains”; “associated with” cannot become “caused by”. You need not reject the proposed explanation, only avoid making it more certain than the text permits.

Reported views are observations of what people said. If a survey finds that most respondents expect shorter queues, the supported conclusion concerns expectations, not necessarily future queues. Predictions can be sincere and still wrong. Similarly, a proposed plan is not a completed action. Carry the verb's status along with its subject: considered, authorised, begun and completed describe different stages. An answer may match every noun in a passage while changing its central claim through one verb.

Try to defeat an inference

To test whether an answer must follow, imagine a situation where the passage is true but the answer is false. If only returned forms are described, perhaps non-respondents have very different views. If an average rises, perhaps some individual values fall. If one branch improves, perhaps another worsens. These possibilities do not prove that the answer is false in reality; they show that the passage does not require it. That is enough to reject it as a necessary inference.

Finally, distinguish absence of evidence from evidence of absence. If no passage sentence mentions evening visitors, you cannot conclude that there were none. You can say only that their number is not established. Good reading answers are often restrained: they retain the measured group, comparison and uncertainty. Before accepting a dramatic statement, underline every word that expands its scope, then find the passage language that authorises that expansion.