The March 2026 SAT: answers that were one step short

Split, and along an unusual line. Students describe easy first modules in both sections, then a sharp turn. Several said they had finished Module 1 with time to spare and still ran out in Module 2.

Reading and Writing drew the steadier complaints, with vocabulary the loudest theme. Math was quieter until the last four or five items, which students describe as the hardest part of the day. That’s a normal shape, and it feels worse than it is because the earlier questions let you relax.

One more thing shows up in March and matters more than it looks. More students than usual mention using the calculator heavily. It will help you through the middle of the section and do nothing for you at the end.

The pattern runs through both sections. You do real work, reach a quantity, and the question asks for a different quantity built from it. Every wrong answer you can reach that way is in the choices.

Start with the percentage question. Visitors rise 184 percent, then fall 25 percent. Multiply 2.84 by 0.75 and you get 2.13. Stop there and you answer 213 percent. The question asked how much greater, so the answer is 113 percent. Both numbers are choices, and 213 is the one your working produces.

The quartic question does it with two factorings. The same expression factors once with integer constants and once with non-integer constants, and it wants the first constant from each added together. Most people find the integer pair, 5 and 11, and stall. The second pair comes from the same quadratic, and the sum is 5 plus 16.5, which is 21.5. Find only the integers and you have half the answer with no way to notice.

In both, your work is correct and incomplete, which is the worst combination because nothing feels wrong. A wrong method usually announces itself. A method that stops one step early hands you a clean number that appears on the screen.

The fix is almost embarrassingly small. Before you select, read the final clause of the question and name the quantity it asks for, out loud if you can.

Say it as a noun with its units. Percent greater, not percent. The sum of a and c, not a. The greatest value of x, not a value of x. The perimeter in the form a root 3, so the answer is a, not the perimeter.

That last one caught people. An equilateral triangle has height 51, and the perimeter is given as a times the square root of 3. The side is 34 root 3, the perimeter is 102 root 3, and the answer is 102. If you type 102 root 3 you did every step correctly and answered a question nobody asked.

Geometry and percentages carried the section, with statistics steady underneath. The difficult items clustered late and shared a habit of asking for a derived quantity.

One ran a sight line to the horizon. A planet has radius 6,400 kilometers and your eye sits 45 meters up. A line that grazes the surface is tangent to it, so it meets the radius at a right angle. That gives you a right triangle with hypotenuse 6,400.045 and one leg 6,400, and the sight line is about 24 kilometers. Leave the height in meters and you get 759, which looks plausible and is wrong by a unit.

Another gave an exponential through two points whose x-coordinates were c and 2c. That doubling is the whole gift. If you call a to the c by a single letter, the second point becomes its square, and subtracting the equations kills the other constant. You end up with u squared minus u equals 380, so u is 20.

A pyramid question handed you a perimeter where you wanted a side. Seventy-two is the perimeter, so the side is 18, the base area is 324, and with a height of 45 the volume is 4,860. Use 72 as the side and your base area is sixteen times too big. Statistics ran a cleaner idea. Shift every value in a set by the same amount and the mean and median move with it. The range doesn’t, because the constant cancels when you subtract.

Vocabulary drew the most argument, and three questions account for most of it. Each one sets up a concession, and the concession picks the word.

Take the first. A theory overturns an entire chronology, and the sentence adds that calling it that for its own sake would be a mistake. The phrase for its own sake tells you the blank names the overturning. Iconoclastic is the word. Provisional and derivative both describe real scholarly qualities, and the concession concedes neither of them.

Another opened with contrary to the belief that early metallurgy was primitive. It then reported furnaces hitting the same temperature as workshops five centuries later. Tantamount to is the answer because the evidence makes the two equivalent. At odds with reverses it. Preliminary to restates the belief the sentence rejects.

The third turned on however. New technologies signal progress, and then lost skills undercut that. Evidence refutes a claim. A reader misconstrues a text. That difference is why the second choice felt close and wasn’t.

Beyond vocabulary, the section leaned on questions where the data can’t settle the claim. One gave a single average across all categories and a claim about one kind of category. Without a per-category figure, the honest answer is that you can’t check it, which students reported feeling like a trick until they went looking for the missing number.

Original questions written around what this exam tested, one for each argument above. Nothing here is a real test question.

March didn’t ask for unfamiliar methods. It asked you to carry each one all the way to the quantity named in the question. Percent greater rather than the multiplier. The letter rather than the expression. Both constants rather than the pair you found first.

Practice that separately from solving. Take twenty questions you already know how to do, cover the choices, and write down the quantity each one asks for before you touch the math. You’ll find the habit takes about a week to build and saves you points on every form after this one.