Reference

SAT Glossary

Plain-English definitions of adaptive testing, Item Response Theory, and the psychometric concepts behind how the Digital SAT works and how your score is calculated.

Understanding these terms gives you a strategic edge: you will know exactly which questions carry the most scoring weight and why hard practice is the highest-leverage preparation approach.

Jump to:Adaptive Testing Item Response Theory (IRT)Discrimination Parameter (a)Difficulty Parameter (b)Ability Estimate (θ)Multi-Stage Testing (MST)Score Equating

Adaptive Testing

A testing method in which the difficulty of subsequent questions is adjusted based on the test-taker's performance on earlier questions.

Full Definition

Adaptive testing (also called Computer Adaptive Testing or CAT) is a form of standardized assessment in which the algorithm selects each question — or in the SAT's case, each second module — based on the ability estimate built from prior responses. The goal is to present questions that are neither too easy nor too difficult for the individual student, which improves measurement precision and reduces test length compared to fixed-form exams. The Digital SAT uses a multi-stage adaptive design: Module 1 is the same for all students, but Module 2 is routed to either a higher-difficulty or lower-difficulty form based on Module 1 performance.

Why it matters for your prep

Students who have only practiced at average difficulty will be unprepared for Module 2 Hard — the version that leads to scores above 1400. Adaptive testing means your preparation must include hard-difficulty questions, not just test-level questions.

Further reading

How the Digital SAT Algorithm Works Adaptive SAT Practice Questions

Item Response Theory (IRT)

A psychometric framework that models the probability of a correct answer as a function of both student ability and item characteristics.

Full Definition

Item Response Theory (IRT) is the mathematical foundation behind modern standardized test scoring, including the SAT, ACT, GRE, and many state assessments. Unlike classical test theory — which simply counts correct answers — IRT assigns a probability function to each question based on its difficulty, discrimination, and guessing parameters. A student's score is not a raw count; it is a maximum-likelihood estimate of their underlying ability (θ), derived from the pattern of rights and wrongs across all questions they answered. The College Board uses a three-parameter logistic (3PL) IRT model for the SAT, meaning each item is described by three values: difficulty (b), discrimination (a), and pseudo-guessing (c).

Why it matters for your prep

Understanding IRT explains why a student who answers 20 hard questions correctly can outscore a student who answers 25 easy questions correctly. It also explains why consistent hard practice, not just higher volume of practice, is the most efficient way to move your score.

Further reading

How the Digital SAT Algorithm Works

Discrimination Parameter (a)

A value describing how sharply a question separates students at different ability levels.

Full Definition

The discrimination parameter (a) in an IRT model quantifies how well a question distinguishes between students just below and just above a given ability level. A question with a high a value (typically above 1.5) produces a steep probability curve — a small difference in ability leads to a large difference in the probability of a correct response. This means high-discrimination items are very good at sorting students at the margin: for example, distinguishing a student at the 1380 level from one at the 1420 level. The Digital SAT intentionally includes high-discrimination items in both modules, and especially in Module 2 Hard, because these are the questions that produce the most information about a student's true ability at the high end of the scale.

Why it matters for your prep

High-discrimination questions are the hardest to fake your way through. Students who have not built genuine skill in a domain will get them wrong regardless of strategy or timing. This is why deep practice — not just practice volume — is required to improve at the high end of the score range.

Further reading

How SAT Prep Mastery Builds Real Skill

Difficulty Parameter (b)

A value representing the ability level at which a student has a 50% chance of answering the question correctly.

Full Definition

The difficulty parameter (b) in IRT defines a point on the ability scale. At exactly that ability level, a student has a 50% probability of answering correctly (ignoring the guessing floor). Questions with a high b value are considered hard: a student must be at a higher ability level to have a 50% chance of success. The College Board assigns b values to every question through a process called item calibration, conducted using field-test data from real students. Questions in Module 2 Hard skew toward higher b values, while questions in Module 2 Easy skew toward lower b values. Within Module 1, b values are spread across a range to sample student ability broadly before routing.

Why it matters for your prep

When you practice exclusively on medium-difficulty questions (average b), you never encounter the items with high b values that populate Module 2 Hard. SAT Prep Mastery's hard-mode training exposes you to the full upper range of b values in the question bank.

Further reading

Module structure and difficulty distribution

Ability Estimate (θ)

Your estimated latent ability on a continuous scale, updated in real time as you answer questions during a test.

Full Definition

In IRT, θ (theta) represents the underlying ability of a test-taker on a latent trait — in this case, SAT math ability or reading/writing ability. θ is expressed on a continuous scale (typically −3 to +3 in standard psychometric notation, though the SAT translates this to a 200–800 scale). During a test, the algorithm calculates a maximum-likelihood or Bayesian estimate of your θ after each response, using the known parameters of every question you have answered. A correct answer to a hard question (high b) increases your θ estimate more than a correct answer to an easy question (low b). Your final scaled score is a transformation of your θ estimate at the end of the test, adjusted for the specific equating constants used on that test form.

Why it matters for your prep

Your ability estimate is the mechanism by which the SAT routes you into Module 2 Hard or Easy. Building a higher true θ through structured practice — particularly hard-difficulty practice — is the most direct way to ensure you access the high-scoring module.

Further reading

How the SAT estimates your ability in real time Score improvements from targeted hard practice

Multi-Stage Testing (MST)

An adaptive testing design in which groups of questions (stages) are routed adaptively, rather than individual items.

Full Definition

Multi-Stage Testing (MST) is the specific variant of adaptive testing used by the Digital SAT. Unlike item-level CAT — where each individual question is chosen based on the current ability estimate — MST selects entire modules based on performance in the preceding module. The SAT uses a two-stage design: Stage 1 (Module 1, fixed for all students) followed by Stage 2 (Module 2, routed to hard or easy form). MST is preferred over item-level CAT in high-stakes testing because it allows test-takers to review and change answers within a module, which is not possible in item-level adaptive systems where each answer locks in the next item selection. The tradeoff is slightly lower measurement precision compared to item-level CAT, but greater test security and a more familiar test-taking experience.

Why it matters for your prep

The MST structure means you have one pivotal decision point: how well you perform in Module 1. There is no opportunity to "catch up" once you are routed into Module 2 Easy. Focused preparation on Module 1 difficulty — especially the latter half of the module — is the highest-leverage place to invest practice time.

Further reading

Full breakdown of the SAT module structure

Score Equating

A statistical process that adjusts scaled scores across different test forms so that equivalent ability levels receive the same score, regardless of which form a student took.

Full Definition

Score equating is the process the College Board uses to ensure that a 1400 on one test date means the same thing as a 1400 on a different test date, even though the questions differ. Because no two test forms are identical in difficulty, raw scores cannot be directly compared across forms. IRT-based equating solves this by anchoring all scaled scores to a common ability scale. Questions from each form are statistically linked to a reference form using their calibrated b, a, and c parameters, and the conversion from raw to scaled accounts for these differences. This is why you cannot simply count how many questions you got right to calculate your score — the difficulty of those questions matters.

Why it matters for your prep

Score equating means that getting harder questions right is always rewarded. A student who scores 80% correct on Module 2 Hard will receive a significantly higher scaled score than a student who scores 80% correct on Module 2 Easy — because the equating process values correct answers on harder items more.

Digital SAT Algorithm

Full technical breakdown of how the SAT adaptive system works.

How SAT Prep Mastery Works

How the platform trains you above the SAT's difficulty ceiling.

Adaptive Practice Questions

Questions calibrated to the hardest SAT difficulty bands.

Put the theory into practice

SAT Prep Mastery is built around the same IRT principles that power the real exam. Train on the hardest question bands and build the ability estimate you need to reach your target score.