The May 2026 SAT: easy math, and a Reading section that punished people

It depends entirely on which section you mean, and May is the clearest case of that we have seen. Difficulty reports and reassurances came in at almost the same rate, which usually signals a middling exam. Reading what students actually said, the split was not disagreement at all. The people saying it was easy meant math. The people saying it was brutal meant Reading and Writing.

The second module of Reading and Writing drew the strongest reaction, described as a marked step up from the March exam, with lengthy scientific passages and a heavy load of graphical analysis. Students reported spending time on individual questions that they would normally have banked for the end of the module.

Math, by contrast, ran through standard ground: proportional relationships, a perpendicular slope, a probability read from a table, area of a rectangle, centre and spread from a small data set. Where it was hard it was hard in the familiar way, asking which function displays a property rather than asking for a value.

The single most useful observation about May is that its hardest reading questions were not about reading in the sense most students prepare for. They were about evaluating whether a piece of evidence supports a claim, frequently with a figure or a table attached, and frequently in dense scientific prose about an unfamiliar system.

This is a real shift in emphasis, and it has a practical consequence. Students prepare for Reading and Writing as though it rewards comprehension, when the hardest items reward argument analysis. Understanding every sentence of a passage about contaminant levels in river species does not get you the point. Identifying that the researchers’ proposal depends on one species predicting the other, and that the weakening finding must break exactly that link, does.

The reason it costs so much time is that comprehension feels like progress. A student who does not know what the question is looking for will read the passage twice, understand it well, and still be choosing between four plausible findings at the end of it.

Plenty of students left May pleased about the math, and it is worth saying what that does and does not mean. It does not mean a higher score. Equating adjusts for the difficulty of the form you sat, so a comfortable math section converts less generously than a punishing one. Feeling good about a section is not evidence about the score it will produce.

There is a second reason not to relax. On an adaptive test, a math section that felt easy throughout may mean you were routed onto the lower-ceiling second module after the first one, in which case the comfort was a symptom rather than an achievement. Students consistently read an easy second module as good news, and it is the one interpretation the design rules out.

The useful reading of May is therefore the opposite of the comfortable one: if math felt easy and Reading and Writing hurt, the preparation to change is not the section that hurt more. It is the balance between them, because most students are already spending the majority of their time on math.

May’s hard reading questions were overwhelmingly evidence questions: a researcher proposes something, and you are asked which finding would most weaken or most support it. The passages were long and technical, which is what made them feel hard, but the technical content is almost never where the difficulty lives.

Every one of these questions rests on a single link. A proposal that crayfish can stand in for trout depends on the two species’ contaminant levels rising and falling together. A claim that languages spread after farming arrived depends on the date farming arrived. Once you can state that link in one sentence, three of the four choices become visibly irrelevant, and you never need to have understood the biology.

The math was broad and shallow: inequalities, perpendicular lines, probability from a two-way table, similarity criteria, area, centre and spread. Nothing exotic, and students noticed.

The exception was a small cluster asking which of two defined functions displays some property in its equation, such as showing its minimum value as a constant. These are the questions where a calculator is least useful and where confident students go wrong, because a decaying exponential approaches a value it never reaches while a growing one attains its minimum at the left end of the domain. The distinction is conceptual and takes ten seconds if you know it.

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

May is the sitting that should change your study balance. Most students allocate the majority of their time to math because progress there is legible: you either solve it or you do not. Reading and Writing feels less trainable, so it gets the remainder.

The hardest questions on this form say otherwise. Evidence questions are as mechanical as any algebra item once you know what to look for, and they are worth more of your attention than their reputation suggests, because the students losing points on them are usually the ones who read the passage most carefully.