The format: two short passages, two authors
A cross-text item gives you two brief passages, Text 1 and Text 2, by different authors on the same topic. The question asks how the second author would relate to the first: whether Text 2 supports, challenges, qualifies, or complicates a claim in Text 1. It sits in the Craft and Structure domain and is one of the rarer question types, so it is not where your study hours should pile up. But when it appears it is winnable, because the answer is never a matter of taste, only of reading two positions and one relationship correctly.
The reason students miss these is not comprehension. It is bookkeeping: two authors, two claims, and a clock competing for the same working memory until the attributions blur. The method below locks each position down before it can smear.
One line per author: the position summary
Before you read the answer choices, write or think one sentence for each author. Not a summary of the whole passage. The author's position: the single claim they are actually arguing.
Force it into this frame:
- Text 1 says: ______
- Text 2 says: ______
Keep each line to the claim, stripped of the supporting detail. If Text 1 spends four sentences describing a study and one sentence drawing a conclusion, the conclusion is the line. The study is scenery.
This is the highest-leverage habit on the question type, and a defense against anchoring. Skip straight to the choices and they start proposing positions to you, three of them engineered to sound like something an author could have said. Committing to your own one-liner first is the anti-anchoring move taught in Predict-Then-Peek: decide what you think before the test shows you what it wants you to think.
Then name the relationship: agree, disagree, or scope
Once you have two clean lines, put them side by side and name the link. This is the Same-or-Opposite Test you already use on transition questions, scaled up from two sentences to two passages. Ask the one question: does Text 2 point the same direction as Text 1, or against it?
But most cross-text pairs are not a flat agree or a flat disagree. The interesting relationship usually lives in one of these:
- Agree on the finding, disagree on the cause. Both authors accept the same result, then split on why it happened.
- Agree in general, disagree on scope. Text 2 accepts the claim but narrows it: true here, not everywhere.
- Concede, then qualify. Text 2 grants Text 1's point, then adds a limit that changes what it means.
Naming which of these you are looking at is the whole game. A choice that says the authors flatly disagree, when Text 2 accepts the finding and only questions the cause, states the wrong logical relationship: a wrong-relationship answer, a real link dressed as a different one. So is the choice that overstates the friction, claiming Text 2 "rejects" Text 1 when the tone is a measured qualification. Match the temperature of the answer to the temperature of the text, or you walk into a too-strong answer.
The trap that catches everyone: swapped attributions
Here is the most expensive mistake on this question type, because it costs points even when your reading was perfect.
The test builds a distractor by taking Text 2's actual view and stapling it to Text 1, or by inverting who supports and who challenges. Everything in the choice is textually true; someone really did say it. Only the direction is wrong, and direction is exactly what scrambles when two positions sit in a tired head at once. In our taxonomy this is a contradicted or flipped answer: the polarity or the attribution is reversed while every other word stays faithful to the text, and that faithfulness is what makes it land.
Your one-line summaries are the defense. When a choice describes a position, run the Receipt Rule on it: point to the exact author and the exact line that proves the attribution, not a feeling that someone said it. If the choice hands Text 1 an argument you can only find in Text 2, kill it, no matter how accurate the words are. This is a wasted point: nothing to do with what you know, everything to do with what your hand does under a clock.
A worked example
Two invented passages, so nothing here is a real test item:
Text 1. A city added bike-share stations, clustered downtown. The author argues that because the stations sit where car-owning commuters already work, the program mostly serves people who least need it and widens the city's mobility gap.
Text 2. After a later expansion pushed stations into outer neighborhoods, ridership among residents without cars rose sharply. The author concludes that bike-share can close the mobility gap when stations are placed where carless residents actually live.
One line each:
- Text 1 says: bike-share, as it was built, widens the mobility gap.
- Text 2 says: bike-share can close that gap if the stations are placed differently.
Name the relationship. They land on opposite final verdicts, widens versus can close, but they do not contradict each other on the diagnosis: both treat station placement as decisive. So Text 2 is not a flat rejection. It accepts the concern and reframes it as a fixable design flaw rather than a flaw in the concept: concede-then-qualify.
Now watch the traps:
- A choice saying Text 1's author believes better placement would solve the problem. That is Text 2's view attributed to Text 1. Swapped attribution. Dead.
- A choice saying Text 2 flatly rejects Text 1's argument. Too strong; Text 2 shares the placement diagnosis. Dead.
- A choice saying both authors agree bike-share is a failed idea. Neither says that; it is unsupported by either passage, an out-of-text answer. Dead.
- The credited answer: Text 2's author would respond that the disparity Text 1 identifies stems from station placement, not from the program itself, and can be corrected. It survives, because you can point to the line in Text 2 that says it.
The method, in two lines
Strip the whole thing to a routine you can run in under a minute:
- Write one line per author before reading any choice: Text 1 says ___, Text 2 says ___.
- Name the relationship between those two lines: same, opposite, or same-with-a-limit (cause, scope, or concession).
Then eliminate: any choice that misstates one of your two lines is out, and any choice that names a relationship your lines do not support is out. That is Kill-Three by attribution and direction, not by vibe. Reading the choices first, uncommitted, is the first-instinct override failure in a new costume.
Run this self-test today: grab any cross-text pair from a practice section. Cover the question and the choices, and write your two author-lines and one relationship word in the margin. Then uncover the stem. If your lines already contain the answer, your reading is fine and any misses are pure attribution slips you can drill away. If you could not write clean lines, the leak is upstream in how you read the passages, and that is what to slow down.
Practice it now: the free simplesat command-of-evidence drill trains the exact habit cross-text questions reward, forcing you to point at the line that proves each claim before you commit. No signup, works as a guest.