Efficient Reading with Evidence

Start with the question to identify the job: summarize the passage, retrieve a stated detail, or infer something the passage supports. Then read the full short passage with that job in mind. Task-first reading is a useful default, not permission to read only the final sentence. A final claim may qualify an earlier claim; a detail may refer to a group introduced several sentences earlier. Your first interpretation is provisional until the whole passage makes sense.

Compress meaning, not just length. After reading, form a brief account such as 'Old explanation predicts X; new finding shows Y; author limits the conclusion to Z.' For a central-idea question, the answer must represent the passage's main point at the right level of breadth. An accurate description of an experiment's equipment can still be a poor summary of a passage about what that experiment reveals. Conversely, 'Scientists study plants' is too broad to preserve a specific finding about shade tolerance.

For a detail question, identify the requested person, object, time, or comparison and locate the statement about it. Return to that sentence and enough surrounding context to resolve references. For an inference, separate the stated premise from the conclusion you can safely add. If a passage says a technique works only on dry samples, it supports a limitation on that technique; it does not establish that every wet sample is impossible to analyze by any technique. Watch the answer's subject and scope as carefully as its main verb.

When two choices remain, name their decisive difference: all versus some, observation versus cause, one method versus every method, or the whole argument versus one example. Check that difference against the text. An extreme word is not automatically wrong; it is wrong when the passage does not justify it. Move on when one choice is supported and each rival has a demonstrable defect. Further rereading without a disputed claim to investigate usually adds little.

Evidence questions ask for a relationship between a claim and a finding. Before evaluating choices, state the claim in plain language, including its comparison and qualification. 'The lining lowers noise at high speeds' is different from 'The lining lowers noise at every speed.' Also note the direction of the question: support, weaken, illustrate, or complete. A finding can concern the right topic while doing the opposite of what the question requests.

For textual evidence, ask what you would expect to observe if the claim were true. Suppose researchers propose that brief darkness, rather than a temperature change, triggers an animal's movement. A useful comparison changes light while holding temperature stable. A finding about warmer animals in darker habitats may leave the two explanations entangled. You need not design an entire perfect experiment; identify the evidence that bears most directly on the stated claim among the choices. Do not require a single result to prove every possible implication.

For a table, read the title, row labels, column labels, and units before choosing numbers. Then translate the claim into an exact comparison: which two cells, which difference, or which sequence? If the claim concerns improvement, compare before and after for each group; the highest final value alone may be misleading. If groups differ in size, raw counts may not answer a question about rates. Keep the denominator attached to the number. Eight successes out of ten attempts is not the same result as eight successes out of twenty.

Match the strength of the answer to the design described. A table may show an association without establishing its cause; a finding from one tested condition may support a conclusion about that condition without supporting every setting. Resolve the final two options by doing the smallest necessary check, such as calculating two changes or locating the clause that isolates a variable. State the result with its units. Extra arithmetic and outside scientific knowledge usually distract from the evidence comparison the question actually asks you to make.

Words-in-context questions reward local meaning, not the most familiar dictionary meaning. Read the entire sentence and enough surrounding text to identify the relationship around the tested word. Predict a plain substitute, then test the actual choices in context. A scholar who 'qualifies a claim' limits it; the scholar does not necessarily earn a credential. Your substitute is a guide, not a commitment: a choice may capture the meaning more accurately than your first prediction.

For a blank, identify what the sentence requires before choosing among polished words. Is the author describing a result as expected or surprising, a pattern as stable or variable, an explanation as convincing or incomplete? Pay attention to contrast markers and negation. In 'Although the samples differed in color, their composition was ___,' the contrast makes similarity in composition plausible. But the rest of the passage must still support the particular degree of similarity the answer states.

For purpose questions, separate what a sentence says from what it does. 'A visitor counts five windows' describes content. 'It introduces an observation that a later discovery corrects' describes a role. Summarize the sentence immediately before and after the target; those neighboring roles often reveal its function. Distinguish the author's voice from a view the author reports. An objection quoted in a passage may be included so the author can answer it, not because the author accepts it.

For overall structure, map the sequence of ideas using ordinary verbs: introduces a problem, describes an approach, reports a result, limits an interpretation. A valid structure answer must fit the sequence and the relationships, not merely mention the same subjects. Watch for choices that turn a hypothesis into an established result or present a single study as a debate between researchers. If a passage contains technical vocabulary, first identify these familiar logical roles. Look up nothing and import no outside facts; use the passage's own explanations to test the choices.

Read the question first, then keep the two texts separate long enough to identify each author's claim. Reduce each to a short statement with its qualification: 'Text 1: rankings probably change because of reader preference. Text 2: rankings may change because the catalog changed.' Both concern rankings, but the second offers a competing explanation. A shared topic does not establish agreement, and different emphases do not automatically establish disagreement.

For a likely-response question, focus on the exact statement one author is asked to evaluate. Ask whether the other author would accept it, narrow it, question its evidence, or offer an alternative cause. Use only the second author's stated commitments. An author who questions the strength of a conclusion need not believe its opposite. If Text 2 says the available evidence cannot distinguish two explanations, do not choose an answer claiming that Text 2 has disproved one of them.

For difficult language, extract the sentence's basic claim before interpreting every descriptive phrase. Find who or what is being discussed, what is asserted, and which clause restricts that assertion. In 'Although the revised catalog, compiled from records of uneven completeness, appears exhaustive, its silence about an item does not establish the item's absence,' the core is 'the catalog's silence does not establish absence.' The interruption explains why. Preserve not, only, unless, some, and other small words that can reverse or limit the claim.

After simplifying, return to the original text to verify your paraphrase. Compression is useful only when it preserves the author's degree of certainty and the relationship between ideas. If you are stuck, choose a narrow question to investigate: 'Which explanation does Text 2 actually challenge?' or 'What does this not rule out?' A targeted reread can resolve that question. Repeatedly restarting both texts without a target often wastes time. If a productive check is still unclear and other questions remain, use the guide's deferral routine rather than forcing certainty.

Keep these checks in mind

• Identify the task, read the full short passage, and compress its claims without losing their scope or certainty.

• For close answers, name a disputed word or relationship and check that precise difference against the evidence.

• For data, translate the claim into the needed comparison and keep units and denominators attached to the values.

• For paired texts, distinguish a shared topic from agreement and insufficient evidence from proof of the opposite.

• Use targeted rereading when you can name what it will resolve; use a provisional answer and a useful flag when further work is stalled.

Begin with four short drills, one per lesson, then complete twelve multiple-choice questions, three per lesson. Labels distinguish SAT-format questions from strategy decisions. Give the evidence or stated condition that decides each item. Model drill responses are examples; another well-supported response may also work.