Parallel Reasoning on the LSAT: Why the Smart Move Is to Read Less
Most LSAT logical reasoning questions reward slow, careful reading. Parallel reasoning questions quietly punish it. If you read all five answer choices in full, the way you would for an assumption or a flaw question, you will burn three or four minutes and still feel unsure. The entire point of this question type is that the content is a decoy. Once that clicks, these stop being time sinks and become some of the most mechanical points on the section.
A parallel reasoning question hands you an argument and asks which answer choice reasons in the most similar way. The stems are easy to spot: the reasoning above is most similar to which one of the following, or the pattern of reasoning is most closely paralleled by. The instant you recognize one, flip a switch in your head. Stop tracking what the argument is about and start tracking what it does.
This matters because the test mismatches topics on purpose. The stimulus might be about whether a bakery should buy a new oven, and the credited answer about migratory birds. That is not a coincidence or a trick; it is the design. Topic similarity is bait, and weaker scorers chase it.
So before you touch the answers, compress the stimulus into a structure. Drop the nouns, keep the logical moves, and ask three questions: what kind of premises do I have, what kind of conclusion, and how strong is that conclusion?
Two quick examples. Every certified electrician has passed the safety exam. Dana is a certified electrician. So Dana has passed the safety exam. That is clean, valid, and deductive: all A are B, x is an A, therefore x is a B. Now compare: most commuters own a bike, and most bike owners use the trails, so most commuters use the trails. That looks parallel on the surface, but it is flawed, because most and most do not chain together. A valid skeleton and a flawed skeleton are never parallel, no matter how alike the wording feels.
Get LSAT Guy’s stories in your inbox
Join Medium for free to get updates from this writer.
Conclusion strength is your fastest filter. A conclusion that says must be cannot match one that says probably or could be. Modality alone eliminates answers in seconds.
The move that saves the most time comes next: after you have your skeleton, read only the conclusion of each answer choice before reading a single premise. If your stimulus ends in a recommendation, such as the city should, throw out every answer that ends in a prediction or a flat statement of fact. It is common to knock out three of five choices on conclusion type alone.
Then check the finer features of that conclusion: positive or negative, conditional or absolute, about one specific case or a general rule. Each has to line up. Only once you are down to two survivors do you go back and compare premises in detail. Reading five full arguments is what eats your clock; reading five conclusions and then two full arguments does not.
Some of these questions add a wrinkle. The original argument is broken, and you need the answer that breaks in the same way. The stem tips you off, with phrasing like the flawed reasoning above is most similar to. Do not skip straight to matching. Name the flaw first. Is it confusing a necessary condition with a sufficient one? Reasoning from a part to a whole? Leaning on an irrelevant authority? Generalizing from a bad sample? Once you can put a label on it, you are hunting for the answer that commits that exact error. An answer that is logically valid, or flawed in some other way, is wrong even when it sounds like the stimulus.
Here is the whole method in order. Confirm from the stem that it is parallel reasoning. Abstract the stimulus into a skeleton, noting conclusion type and strength. Read only the conclusions and cut the mismatches. Compare premises for the two answers you have left. If it is a parallel flaw question, name the flaw before you match.
Run this on ten real questions and you will feel both speed and accuracy climb, because you stop getting pulled around by subject matter and start seeing the small set of structures the test reuses over and over.
I originally published the complete version of this framework on Verbloom.
