An SHSAT practice test that actually adapts to you.
Most "SHSAT practice tests" online are PDFs of the same recycled questions. Ours is a working Computer Adaptive Test built on the same Item Response Theory framework the SAT and the official SHSAT-CAT use. Take it once and see where you really stand.
- How long is the test?
- Typically 30–60 minutes. Up to 12 questions per section. The engine stops early once it has a confident ability estimate.
- Is the test free?
- Yes. No signup. No payment. No email. Your responses stay in your browser and are not sent to any server.
- What format does the test simulate?
- The Computer Adaptive Test (CAT) format used by the official SHSAT since October 2025. Questions adapt to your ability after each response.
- What does my score mean?
- You get an estimated composite score (200–800) and a per-section scaled score (100–400 each). The test labels its confidence so you know how precise the estimate is.
How adaptive testing actually works.
If you've taken a digital SHSAT, GRE, GMAT, or NWEA MAP test in the last decade, you've sat through a Computer Adaptive Test without anyone explaining what was happening under the surface. Here's what's happening — and what makes our practice test different from the static PDFs most prep sites call "practice tests."
Every question has hidden numbers
Each item in our pool has three statistical properties that the engine uses but that you never see:
- Difficulty (b) — the ability level at which a typical test-taker has a 50% chance of getting this question right. Easier questions have lower difficulty.
- Discrimination (a) — how well this question separates strong test-takers from weak ones. A high-discrimination item is one where strong test-takers almost always get it right and weak ones almost always get it wrong.
- Guessing (c) — the floor probability of getting the question right by random guessing. For a 4-option multiple choice item, this is roughly 0.25.
These three numbers feed into the 3-Parameter Logistic model, which gives the probability that a test-taker at a specific ability level answers the question correctly. The math behind this has been a published, peer-reviewed framework in educational measurement since the 1960s and is what every major standardized test uses today.
The engine estimates your ability after every answer
You start the test at the population average. After each question, the engine performs Bayesian updating: given everything you've answered so far, what's the most likely ability that would have produced this response pattern? The answer is a number called theta (θ), which ranges from roughly −3 (low ability) to +3 (high ability) for typical test-takers.
The engine doesn't just calculate your current θ — it also calculates the standard error around that estimate. When the standard error is large, the engine isn't yet confident about where you are. When the standard error is small, the engine has converged.
The next question is the one that teaches the engine the most
This is the part that makes adaptive testing efficient. After estimating your ability, the engine looks at every unused question in the pool and calculates the Fisher Information that question would contribute at your current ability level. The question with the most information at your specific θ gets served next.
In practice, this means:
- If you're getting questions right, the engine starts serving you harder items.
- If you're getting questions wrong, the engine pivots toward easier items.
- If you're solidly in the middle, the engine keeps you near medium-difficulty items.
You can't game this by intentionally answering wrong on the first few. The engine looks at your full response pattern, and a pattern of "easy questions wrong, hard questions right" is statistically improbable and gets weighted differently than a consistent response pattern.
Your final scaled score comes from your full response pattern
When the test ends, the engine takes one final pass at the estimation and translates your θ into the SHSAT scaled score range (100–400 per section). This is not the same as "how many you got right divided by total." Two test-takers who got the same raw count of questions correct can have different scaled scores if one was answering harder questions than the other.
The composite score is the sum of your ELA and Math scaled scores. It ranges from 200 (worst case) to 800 (perfect on both sections).
A real exam-style interface.
The test interface is designed to feel like the digital SHSAT environment — focused, low-distraction, and similar enough that you won't be surprised on test day.
One question at a time, no skipping ahead
The real SHSAT-CAT serves questions one at a time and doesn't let you preview future items, because the next item depends on how you answered this one. Our test works the same way.
Flag for review
You can flag any question you want to come back to. Because the test is adaptive, you can't return to flagged items mid-test — but the flags are saved with your results, so when you review your performance afterward, you can see which questions you were unsure about.
Immediate explanations
After you submit each answer, the test reveals whether you were correct and shows a worked explanation. This is the opposite of how the real exam works (you get no feedback on the real test), but it makes our test genuinely useful for learning, not just for measuring. If you only want measurement and prefer no feedback during the test, that's coming as a toggle in a future update.
Confidence labeling on your results
When the test ends, your results screen shows your composite score along with a confidence label: high, medium, or low. This reflects how tight the engine's estimate is. A high-confidence result means the score is reliable; a low-confidence result means you got an unusual pattern of responses and the estimate has wider uncertainty. We label this explicitly because pretending to high precision when we don't have it is exactly the kind of thing AdSense flags and exactly the kind of thing parents shouldn't have to figure out for themselves.
A few things we want you to know
Our pool is smaller than the real exam’s. The official SHSAT-CAT pulls from hundreds of items calibrated against thousands of student responses. Our pool is 105 items (50 ELA + 55 Math), all reviewed by our Education Editor before going live. With fewer items, the engine still converges on a reasonable estimate, but the confidence interval is wider than the real test’s. We tell you the confidence on every result.
Every question is original. We didn't scrape the NYC DOE handbook or any prep publisher. Every item in the pool is written for this site and reviewed by our Education Editor, Elisa Ahmed, who teaches in the NYC public school system and works directly with SHSAT-eligible students.
Your data stays on your device. The test runs entirely in your browser. Your responses are not sent to any server. We don't know what you answered, what your score was, or whether you finished. We can't see your data because we don't collect it. This is by design, not by accident.
For the technical details — the IRT model parameters, how items are calibrated, how the EAP estimator works, the full conversion methodology — see our methodology page. We try to show our work.
Or take a full-length mock exam
If you prefer a fixed-sequence exam over the adaptive test, we offer three full-length mock exams. Each presents a different set of 35 questions in order, timed at 75 minutes, with scaled scoring at the end. The 3 mocks have no item overlap.
See the 3 mock examsFAQ
Is this practice test really adaptive?
Yes. The test uses a 3-Parameter Logistic Item Response Theory engine — the same statistical framework used by the SAT, GRE, and the official SHSAT-CAT. After each question, the engine estimates your ability and selects the next question to give the most information at your current level. The full implementation is open in our methodology page.
How long does the practice test take?
The adaptive test administers up to 12 questions per section (ELA and Math), so total length is typically 30 to 60 minutes depending on how quickly you read and decide. The full official SHSAT is 180 minutes; our test is shorter because the adaptive engine reaches a stable estimate faster than fixed-form testing.
Do I need to create an account?
No. The test runs entirely in your browser. Your responses, scores, and any data never leave your device. There is no signup, no email collection, and no account.
How accurate is the score estimate?
Your composite score estimate is based on a smaller question pool than the official SHSAT uses, so the confidence interval is wider. With our current 105-item pool, expect your estimate to be within roughly ±20–25 points of your true SHSAT ability. As the pool continues to grow, that interval tightens. The test labels the confidence of your result as high, medium, or low so you know how to interpret it.
Can I take the test more than once?
Yes, as many times as you want. Each session draws from the full pool, and the randomesque selection prevents you from seeing the exact same item order every time. That said, retaking the test back-to-back doesn't really tell you new information — you'd be better served by studying the topic areas where you struggled and coming back in a week or two.
What if I got a question that I think is wrong?
Email us at [email protected] with the question ID (shown at the top of each question card) and what you think the right answer should be. We review every correction request and update the pool when we find errors. Pages and items corrected based on reader feedback are logged in our methodology changelog.
How is this different from your score calculator?
The score calculator converts raw scores you already know into estimated scaled scores. The practice test actually administers questions and produces a measured ability estimate. Use the calculator if you've already taken a paper-and-pencil practice test elsewhere and just want to convert; use the practice test if you want the engine to estimate where you actually are.