SAT Intel Report · August 2026
August 2026 SAT: the questions that cost people points
Eight things this test did, each with a real example and what to practice. Most of the hard math was in Module 2, so start there if you’re retaking.
Was the August 2026 SAT hard?
Math was the hard half, and nearly all the complaints were about Module 2. Reading was about average apart from vocab, which people argued over more than anything else on the test. Most of the hard math was a standard topic with one extra step bolted on. If you found the algebra fine and the geometry brutal, you weren’t alone.
Vocab: sentences that call something an exaggeration
The most argued-about question of the test was a vocab one. When a sentence says it would be an exaggeration to call something something, you aren’t looking for the word that describes the thing. You want the word that would be too strong.
What one looked like
The sentence ran roughly “it would be an exaggeration to call the effects ____”. The passage then said the effects were moderate. The last two options meant harmful and in name only.
Where people went wrong
Most people picked harmful, because the passage is about downsides. But moderate effects can still be harmful. So calling them harmful isn’t an exaggeration, and the sentence needs one.
What it actually is
In name only is the answer. Moderate effects are clearly more than nothing, so saying they exist in name only is the overstatement the sentence rules out. Another question did the same thing with water use called negligible in a dry region, where the options meant make worse and ease.
How to prep it
When a sentence admits something is an exaggeration, an overstatement or not quite true, flip it. You want the word that would be wrong, not the word that is right. Then check it against the one fact the passage gives you.
Rewritten from the August 2026 SAT: same idea, new words and numbers
Commentary written in the months after the 1992 moratorium predicted that the closure of the northern cod fishery would hollow out the regional economy within a decade. Reconstructing county-level employment records, the economist Halvard Nyquist finds that it would be an exaggeration to call the eventual effects ______, since displaced workers moved into processing and offshore services at rates nobody had forecast, but the losses in the smallest outports were real and persist in the record to this day. Which choice completes the text with the most logical and precise word or phrase?
Data questions: sometimes nothing is proved
Several questions gave you a figure and asked how well it supports a claim. On this test the answer was often that it barely supports it, or that you can’t tell from what’s shown. People aren’t used to picking that.
What one looked like
A figure gave a proportion of about 0.50, and the question asked how well the data support the researcher’s claim.
Where people went wrong
Picking the option saying the data strongly support it. There is a real result on the page, so it feels like it has to back something up.
What it actually is
0.50 is a coin flip. So the support is partial at best, and the option saying the data only moderately support the claim is the right one. A brands question went further. The figure never gave the percentages you would need to compare them, so the answer was that you can’t make the comparison.
How to prep it
Before you read the options, ask what number would make the claim true, then check whether you were given it. If the value sits near the middle, or the figure is missing the thing you would compare, pick the hedged option.
Explain-the-result questions: use the reason the text backs up
A few questions reported a result and asked what explains it. Several options are believable. Only one of them is the reason the passage bothers to support with a second fact.
What one looked like
A passage said one group of apes was far more right-handed than another. The question is whether that preference was learned or inherited.
Where people went wrong
Picking inherited, or anything else biological, like the two groups being different species. It sounds like the right kind of reason and the passage never offers it.
What it actually is
Learned is the answer. The group with more human contact picked it up from people. You can tell because the passage also says how strongly right-handed humans are, and that second fact is only there to make this explanation work.
How to prep it
Look for the fact in the passage that seems like a spare detail. It is usually there to support one specific option. Match your answer to that detail rather than to what sounds most scientific.
Rewritten from the August 2026 SAT: same idea, new words and numbers
Roughly 90 percent of people in every population yet studied favor the right hand, a skew considerably stronger than anything documented in other primates. The primatologist Emeka Adeyemi scored hand preference in 212 chimpanzees performing a standardized reaching task at two separate locations in the same region. Animals housed at a sanctuary, where keepers handle and feed them daily, reached with the right hand in 78 percent of trials, while animals in a forest enclosure with minimal human contact reached right in 54 percent of trials, a figure close to chance. Which choice best explains the difference Adeyemi recorded between the two groups?
Complete the logic: build it from the oddly specific detail
Complete-the-logic questions hid the answer in one specific detail rather than in the main point. If a passage bothers to tell you something oddly precise, that detail is the answer.
What one looked like
A passage on sumptuary laws, which banned showy displays of wealth. It said the fines were announced publicly, and that the laws didn’t work. You finish the argument about why enforcing them worked against their own aim.
Where people went wrong
Picking the option saying the fines were too small to put anyone off. It is a sensible reason for a law to fail, and the passage never says it.
What it actually is
The fines were announced publicly. So paying one told everybody you could afford it, which is itself a display of wealth. Enforcing the law produced the thing the law banned. A bird question worked the same way. Bill shape predicted which flowers got visited and nectar amount didn’t, so the birds end up on low-nectar flowers that happen to fit.
How to prep it
Underline the most specific detail in the passage, the one that seems too particular to be there by accident. Build your ending out of it. Then throw out any option that gives a general reason instead of one the passage supports.
Desmos: zoom out before you read anything off
One question set a function as the negative square root of a quadratic. It asked for the greatest x where the function hits zero. Graphing it was right. Stopping at the first root you can see was not.
What one looked like
The curve passes through (2, 0). At x equals 0 it sits at about -17.4, which is minus the square root of 302.
Where people went wrong
Reading 2 off the screen and gridding it. The default Desmos window runs to about 10 in each direction, and 2 is the only root inside it.
What it actually is
A quadratic has two roots, so the function has two zeros. One root is 2, and the quadratic is 302 at x equals 0, so the other root is 151, because 2 times 151 is 302. The question asked for the greatest, so 151 is the answer.
How to prep it
Zoom out once before you read a number off the screen. Do it every time a question asks for the greatest or the least value. Then count the roots and check you’ve got as many as the algebra says.
Parabola graphs: read the signs of a, b and c
A few questions showed a fitted parabola as y = ax2 + bx + c. You say which of a, b and c are positive. Two of the three are quick. The sign of b is the one people guess at.
What one looked like
The parabola opens upward and crosses the y-axis below the origin, at about -4. Its lowest point sits to the left of the y-axis, at about x equals -1.5.
Where people went wrong
Saying all three are positive, because the curve mostly sits above the x-axis and climbs steeply. But c is just the y-intercept, and here that is -4, so c is negative.
What it actually is
a is positive because it opens upward. c is -4, so that one is false. For b, the lowest point sits at x equals minus b over 2a. That’s -1.5, so b is 3 times a, which is positive. Two true, one false.
How to prep it
Practice reading all three signs straight off a sketch. Which way it opens, where it crosses the y-axis, then which side the lowest point is on. Say each sign out loud. You’ll get b wrong until you’ve done twenty of them.
Rectangle in a circle: the diagonal is the diameter
A rectangle was drawn inside a circle. All you were told was how the diagonal compared with the short side. That’s enough, because the diagonal of a rectangle inside a circle is the diameter.
What one looked like
The diagonal is twice the shortest side. Find the perimeter. The options looked like 136 plus 136 times the square root of 3, and 188 plus 188 times the square root of 3.
Where people went wrong
Reading “twice the shortest side” as the long side being twice the short one. With a short side of 68 that gives a perimeter of 408. It is a clean number and it is the wrong shape.
What it actually is
The diagonal is twice the short side, so the sides run 1, root 3, 2. That is a 30-60-90. With a diagonal of 136 the short side is 68 and the long side is 68 times the square root of 3. So the perimeter is 136 plus 136 times the square root of 3, about 372.
How to prep it
Whenever a rectangle sits inside a circle, draw the diagonal and label it the diameter. Then check whether the ratio you were given is side to side or side to diagonal. Side to diagonal of 1 to 2 is always a 30-60-90.
Spacing questions: count the gaps, then add one
One question put fasteners at equal spacing along a wall and asked for an expression for how many you need. The division is the easy bit. Whether you add one is the whole question.
What one looked like
The wall is 8x feet long and the spacing is 8/9 of a foot. Write an expression for the number of fasteners.
Where people went wrong
Dividing and stopping. 8x divided by 8/9 is 9x, so people wrote 9x. That is the number of gaps, not the number of fasteners.
What it actually is
9x gaps need 9x plus 1 fasteners, because there is one at each end. Sketch 3 fasteners on a short wall and you get 2 gaps, which is the check that settles it every time.
- things at intervals count the gaps, then add one for the far end. 9x gaps means 9x plus 1 posts.
- infinitely many solutions one equation is a multiple of the other. If one fifth of x plus three tenths of y equals c, and 2x plus 3y equals d, the second is 10 times the first, so d over c is 10.
- axis of symmetry two points with the same y sit either side of the lowest point, so the x you want is their midpoint.
- horizontal shifts y = f(x - h) moves the graph right by h. A shift right of 509 is f(x - 509), not f(x + 509).
- margin of error a bigger sample gives a smaller margin, and nothing else in the question changes that.
How to prep it
For any question about things placed at intervals, draw four of them and count. Posts, panels, trees, tiles. Decide whether the question wants gaps or objects before you divide anything.
Rewritten from the August 2026 SAT: same idea, new words and numbers
A carpenter drives fasteners along a straight rail that is 12x feet long, where x is a positive integer. She places one fastener at each end of the rail and spaces the rest evenly, with 4/7 of a foot between consecutive fasteners. Which expression represents the number of fasteners she uses?
What to work on before your next test
- Words in Context, do twenty questions where the sentence concedes or denies something, and write out what the sentence admits before you look at the options. Back to the example
- Nonlinear Graphs, sketch ten parabolas and read off all three signs each time, until b stops needing thought. Back to the example
- Area & Volume, practice spotting a 30-60-90 from a ratio, and draw the diameter on every inscribed figure you meet. Back to the example
- Inferences, do ten complete-the-logic questions and underline the supporting detail before reading any option. Back to the example
- Linear Equations & Inequalities, work through interval problems until adding the last one is automatic. Back to the example
Slow down for the last step
Nearly every mistake above happened at the last step, not the first. The right method, then 9x instead of 9x plus 1. The right graph, then the root you could see instead of the one you couldn’t. The right passage, then the sensible reason instead of the supported one.
So before you commit, read the question again and check it asks what you just worked out. Gaps or objects. Greatest or first. Supports or proves. That is a second a question and it is worth more than a new technique.
Practise the 47 questions modeled on the August 2026 SAT
Original questions written around what the August 2026 exam tested, tagged by skill and difficulty, each with a worked solution.
Students told us most about the second math module on this test, so that gets the most space here.The reading of it is ours. We hold no scoring data, we can’t identify unscored questions, and every practice question here is written from scratch.
Questions about what was on this test
What math was on the August 2026 SAT?
Mostly Module 2. Quadratics and their graphs, a rectangle inscribed in a circle, and two touching spheres. Also a system with a shared constant, an equation with infinitely many solutions, exponential growth run backwards, margin of error and a rolling wheel. A lot of it was a standard topic with one extra step added.
What vocab words were on the August 2026 SAT?
People reported a lot of these in pairs. The ones that came up meant in name only, harmful, to make worse, to ease, to permeate, to reconcile and deviating from what is expected. The pairs were the hard part. The last two options usually sat at opposite ends of a scale, so you had to work out which end the sentence needed.
How should I practice the August 2026 question types?
Pick one theme above and do ten questions on that skill alone. Look for the trap named in it rather than just checking the answer. Work slowly enough to say why each wrong option is there. That is what stops the same step going wrong on your own test.