Cross-Text Connections

What this skill tests

College Board calls this skill Cross-Text Connections. It sits inside the Craft and Structure part of the test, alongside Words in Context and Text Structure and Purpose. Cross-Text questions present two short, labeled texts on a related subject, Text 1 and Text 2. They may agree, disagree, qualify one another, address different aspects, or support a common conclusion from different evidence. Your first job is to find out which of these is actually happening before you touch the answer choices.

Read each text separately first

The paired texts share a topic, and locating the exact point where they meet or diverge is the real work. A reader who treats the two texts as one continuous passage arrives at the answer choices holding a single blurred position, and nothing then shows which author owns which half of it.

Read the stem first. It tells you whose view or which relationship is at issue. Then read Text 1 on its own and summarize its claim. Then read Text 2 on its own and ask what it asserts, without yet asking what it answers. Label the two texts briefly, mentally on a straightforward pair and on paper when the texts run dense. The relationship between those two labels is the entire answer, and you want to hold it in your own words before any choice has a chance to suggest one to you.

Record each text's claim and basis

• The claim. What does this author actually assert? Two texts can share a finding and divide entirely over what to make of it.
• The basis. What does the claim rest on: data, a method, a reading of a source, or a definition? Many pairs diverge here rather than at the claim itself.
• The scope and strength. How far does the claim reach, and is it argued outright or merely offered as a possibility? A cautious, narrow claim is hard to reject and easy to limit.
• The cause, where one exists. Does the author explain why something happened, and how firmly? Add this only when the pair actually concerns an explanation.

Locate the exact point of agreement or disagreement

Do not arrive expecting full agreement or full disagreement. A pair can share a finding and split over its cause, share a cause and split over its reach, or work on the same subject without ever meeting the same question at all. When Text 2 opens by confirming what Text 1 reported, any choice that has the second author denying the finding fails right there, on that opening line, regardless of how the rest of the choice reads.

Stop exactly where Text 2 stops

A cross-text answer can take one reasonable-sounding step beyond Text 2 and still be wrong. If Text 2 offers two possible explanations, says the cause remains open, or refuses to choose between alternatives, the correct response has to preserve that same limit. Do not finish the second author's argument for them. Watch for may, could, might, remains unknown, cannot determine, and either/or constructions. A choice that turns an open question into a settled conclusion has changed the author's position without warning, even when the conclusion sounds plausible.

Same finding, different explanation

Text 1. Songbirds in cities sing at a higher pitch than birds of the same species in nearby woodland. Low-frequency traffic noise masks low notes, and Hans Slabbekoorn argues that raising the pitch keeps an urban male audible to females.
Text 2. Recordings from twelve cities confirm the pitch difference. It appears as well at dawn, before traffic builds, and in quiet parks where sound levels match woodland. Urban songs are also shorter and louder, and singing loudly raises pitch on its own.

How would Text 2's author most likely respond to Slabbekoorn's explanation? Text 1 offers a finding and a reason. Text 2 opens by confirming the finding, reports the same pitch difference in two settings where the proposed reason cannot operate, and closes with a rival mechanism. Accepting the result while doubting the explanation is the answer. "Rejecting the claim that urban birds sing at a higher pitch" collides with Text 2's opening word, confirm. A choice saying traffic noise affects only birds singing at dawn reverses what dawn shows, since dawn is a setting without traffic.

Use timing to test a proposed cause

Text 1. Historians long credited the fall in urban tuberculosis deaths to the sanatoria built in the 1930s.
Text 2. Death rates in three large cities fell at a steady pace from 1900 onward, decades before any sanatorium opened its doors. Real wages climbed across the same period, and with them diet, ventilation, and living space.

The chronology weakens the claim that sanatoria explain the beginning of the decline. The correct response is precise and limited: it does not claim sanatoria had no later effect, only that they cannot be the starting cause. A cause cannot start a trend that began thirty years before it existed.

Same evidence, different interpretation

Two authors can accept identical facts and land in different places. If one reads a painter's late still lifes as a retreat from politics and the other notes they date from years when the painter was barred from exhibiting, the disagreement is about meaning, not about what the paintings are. Neither author disputes the evidence; each reads it differently. The same holds for a pottery scatter that one author reads as sudden destruction and another as gradual abandonment. When a text names the other author directly, the disagreement is easier to spot, but the choices still have to be checked clause by clause.

Narrowed, extended, and unrelated claims

• A narrowed claim. If Text 1 says a diet lowers a health risk in all age groups and Text 2 confirms it for adults over fifty while saying the data on people under thirty is too sparse to conclude anything, Text 2 keeps the claim and limits it. That is not a rejection, and it is not evidence of harm.
• An extended claim. A larger survey that finds the same stress pattern and adds a detail about which views matter most supports Text 1. New evidence extending an idea is not automatically a disagreement with it.
• Different questions, same subject. A debate about why wolves declined and a report about radio-collaring efforts both concern the same wolves without answering the same question. Topic overlap is not a connection by itself.

Shared conclusions and partial agreement

Two authors can land in the same place by different routes. If one credits remote workers' higher satisfaction to schedule autonomy and the other credits the missing commute, an agreement question wants the shared bottom line, that remote workers report higher satisfaction, not either author's mechanism. A choice that hands one author's explanation to both overstates what they share.

The reverse also happens: a shared point can hide a larger disagreement. Two texts may agree that food truck registrations fell, while one traces the fall to a licensing fee and the other uses it to argue the whole permitting system must be rebuilt. And a narrow disagreement can hide a larger agreement: two engineers may dispute whether a coating works through thickness or chemistry and still agree entirely that the coating reduces corrosion.

Attack the evidence, not the conclusion

Questioning the data, the sample, or the method weakens the support for a claim without showing the claim is false. A text that points out that a poll reached only landline users has not said the sixty percent figure is wrong. It has questioned whether the sample can support the number. Match the strength of the response: questions and doubts are not disproves.

How far does each author go?

Two authors can accept the same data and differ in how much weight it bears. A field trial with forty plots may be strong enough for one author to recommend wider use and, to another, only a reason for further study. When a method differs, such as a phone survey versus in-person canvassing, the difference matters only if the texts tie it to the disagreement, for instance by saying phone surveys in the area reach an older population.

Harder relationships

• Answers built from both texts. Sometimes neither text mentions the other's finding, and the question asks what both teams would accept. If one text shows warmer-bay fish grow smaller and another shows smaller fish produce fewer eggs, the narrowest statement both force together is that warmer-bay fish probably produce fewer eggs. Reduce each text to one finding, find the shared variable, and reject answers that need an extra assumption about cause, prevalence, or magnitude.
• Thresholds that change. If Text 2 revises a classification rule, apply the new number to the case from the other text. A river that floods five years out of ten is flood-prone under a three-in-ten rule but not under a six-in-ten rule. The same move works with dates, percentages, and ranges.
• Different time horizons. A drop in bridge traffic six months after a toll increase and a full recovery three years later can both be true. Check whether two results measure the same window before calling them a conflict.
• Related but different variables. What people say they intend to do and what they actually do are different variables. Prevalence and incidence, self-report and observed behavior, and total amount and rate of change can all move independently.
• Incomplete, not incorrect. Evidence that an account cannot fully explain one case does not disprove it. Language that fits: incomplete, insufficient, requires revision, does not account for, leaves unexplained. Keep these separate from false, disproven, and invalid.

Classify Text 2's response to Text 1

Use this method

1. Read the question stem first. It names the claim you have to respond to and the author who has to respond to it.
2. Read Text 1 alone and write its claim in about eight words, noting whether it is argued outright or only offered as a possibility.
3. Read Text 2 alone and write what it finds, without yet asking how it relates to Text 1.
4. Name the relationship between your two sentences: agreement, extension, a shared conclusion with different reasons, a narrowed claim, a disputed cause, two texts that never meet on the same question, a boundary carried from one text to the other, results separated by time, a difference in what was measured, or evidence that leaves an account incomplete rather than wrong.
5. Predict the response in your own words before opening the choices. Then match, checking that each view sits with its own author and that your answer preserves any limit or uncertainty Text 2 actually states.

Final checks before you answer

When two choices still look possible, the miss is usually one small substitution: the wrong author, the wrong part of the claim, a conclusion stronger than the text supports, or an agreement that exists in only one of the two texts. On an agreement question, a detail discussed by only one text cannot be part of the answer, however true it is. When a question names one specific claim inside a longer text, the correct response has to address that exact claim, not a different disagreement nearby. And a choice can overreach by claiming rejection where the text only complicates, proof where it only suggests, or an absolute word such as all or entirely that neither text supports. Write out in one line exactly what the two texts share and exactly where they split, then check every clause of the choice you prefer: a single invented clause makes the whole choice wrong.